CN115127545B

CN115127545B - Motion state identification method and device, electronic equipment and storage medium

Info

Publication number: CN115127545B
Application number: CN202210333905.5A
Authority: CN
Inventors: 黄连; 梁杰
Original assignee: Guangdong Genius Technology Co Ltd
Current assignee: Guangdong Genius Technology Co Ltd
Priority date: 2022-03-31
Filing date: 2022-03-31
Publication date: 2024-09-24
Anticipated expiration: 2042-03-31
Also published as: CN115127545A

Abstract

The embodiment of the application discloses a motion state identification method, a motion state identification device, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring motion state characteristics of the terminal equipment at each moment in a first time period; inputting the motion state characteristics of each moment into a first model, and determining the motion state corresponding to each moment according to the motion state characteristics of each moment through the first model; inputting the motion states corresponding to all the moments into a second model, constructing a motion time sequence according to time sequence for the motion states corresponding to all the moments through the second model, and denoising the motion time sequence; and determining the target motion state of the terminal equipment in the first time period according to the motion time sequence after denoising. By implementing the embodiment of the application, the accuracy of the final output target motion state can be improved, and the number of identifiable motion states is increased.

Description

Motion state identification method and device, electronic equipment and storage medium

Technical Field

The present application relates to the field of motion recognition technologies, and in particular, to a motion state recognition method, a motion state recognition device, an electronic device, and a storage medium.

Background

In recent years, as the functions of the terminal device are increased, the terminal device can identify the motion state of the user, so as to provide better motion advice for the user. In the related art, a motion state is generally identified through sensors such as an accelerometer and a gyroscope, but the types of motion states which can be identified are limited. With more and more types of movement available to people, false recognition is easy to occur when the prior art is adopted to recognize the movement state.

Disclosure of Invention

The embodiment of the application discloses a method, a device, electronic equipment and a storage medium for identifying a motion state, which can improve the accuracy of motion state identification.

An embodiment of the present application provides a method for identifying a motion state, where the method includes:

Acquiring motion state characteristics of the terminal equipment at each moment in a first time period, wherein the motion state characteristics are extracted according to motion state data acquired by the terminal equipment, and the motion state data comprise acceleration data and barometric pressure data;

Inputting the motion state characteristics of each moment into a first model, and determining the motion state corresponding to each moment according to the motion state characteristics of each moment through the first model;

Inputting the motion states corresponding to the moments into a second model, constructing a motion time sequence according to time sequence through the second model for the motion states corresponding to the moments, and denoising the motion time sequence;

And determining the target motion state of the terminal equipment in the first time period according to the motion time sequence after denoising.

As an optional implementation manner, in a first aspect of the present embodiment, the determining, by the first model, a motion state corresponding to the respective time according to a motion state feature of the respective time includes:

Determining the probabilities of the respective moments corresponding to the respective preset motion states according to the motion state characteristics of the respective moments through the first model, wherein the sum of the probabilities of the respective preset motion states is 1;

Comparing the probabilities respectively corresponding to the preset motion states at all the moments with probability thresholds corresponding to the preset motion states, wherein the probability thresholds corresponding to the preset motion states are all larger than 0.5;

If the probability of the first moment corresponding to the first motion state is larger than the probability threshold value corresponding to the first motion state, determining the motion state corresponding to the first moment as the first motion state, wherein the first moment is any one of the moments, and the first motion state is any one of the preset motion states.

As an optional implementation manner, in the first aspect of the present embodiment, before comparing the probabilities that each moment corresponds to each preset motion state with the probability threshold value that each preset motion state corresponds to, the method further includes:

acquiring historical state data corresponding to the terminal equipment, wherein the historical state data comprises motion states respectively corresponding to a plurality of historical moments;

Classifying each preset motion state according to the number of the historical moments corresponding to each preset motion state to obtain a first type motion state and a second type motion state, wherein the number of the historical moments corresponding to the first type motion state is larger than the number of the historical moments corresponding to the second type motion state;

And adjusting the probability threshold value corresponding to each preset motion state contained in the first type of motion state and the probability threshold value corresponding to each preset motion state contained in the second type of motion state so that the probability threshold value corresponding to each preset motion state contained in the second type of motion state is larger than the probability threshold value corresponding to each preset motion state contained in the first type of motion state.

As an optional implementation manner, in a first aspect of the present embodiment, the denoising processing for the motion time sequence includes:

Determining an abnormal time according to the motion states corresponding to all the moments in the motion time sequence, wherein the motion state corresponding to the abnormal time is different from the motion state corresponding to the previous moment of the abnormal time and the motion state corresponding to the next moment of the abnormal time, and the motion state corresponding to the previous moment is the same as the motion state corresponding to the next moment;

and correcting the motion state corresponding to the abnormal moment according to the motion state corresponding to the previous moment or the motion state corresponding to the next moment.

As an optional implementation manner, in the first aspect of the present embodiment, before the acquiring the motion state feature of the terminal device at each moment, the method further includes:

Data cleaning is carried out on the motion state data collected by the terminal equipment according to a state value interval, so that cleaned motion state data are obtained, wherein the cleaned motion state data are motion state data in the state value interval;

Separating acceleration data in the cleaned motion state data to obtain multi-axis acceleration data;

And extracting the motion state characteristics of the terminal equipment at each moment in a first time period from the multi-axis acceleration data and the air pressure data in the cleaned motion state data.

As an optional implementation manner, in a first aspect of the present embodiment, the extracting, from multi-axis acceleration data and barometric pressure data included in the motion data set, a motion state feature of the terminal device at each moment in a first period of time includes:

data segmentation is carried out on each axis of acceleration data in the multi-axis acceleration data according to a preset repetition rate and a sample time unit, and segmented acceleration data are obtained;

Performing data segmentation on the air pressure data in the cleaned motion state data according to the sample time unit to obtain segmented air pressure data;

and extracting the motion state characteristics of the terminal equipment at each moment in the first time period from the segmented acceleration data and the segmented air pressure data.

As an optional implementation manner, in a first aspect of the present embodiment, the determining, according to a motion time sequence after denoising processing, a target motion state of the terminal device in the first period includes:

Determining the time quantity corresponding to each motion state in the motion time sequence after denoising, and determining the motion state with the largest proportion in the motion time sequence after denoising according to the time quantity corresponding to each motion state;

and if the corresponding duty ratio of the motion state with the largest duty ratio in the motion time sequence after the denoising processing is larger than a duty ratio threshold value, determining the motion state with the largest duty ratio as a target motion state of the terminal equipment in the first time period.

A second aspect of an embodiment of the present application provides a motion state identifying apparatus, the apparatus including:

The device comprises a characteristic acquisition module, a characteristic extraction module and a characteristic analysis module, wherein the characteristic acquisition module is used for acquiring motion state characteristics of a terminal device at each moment in a first time period, the motion state characteristics are extracted according to motion state data acquired by the terminal device, and the motion state data comprise acceleration data and barometric pressure data;

The state acquisition module is used for inputting the motion state characteristics of each moment into a first model, and determining the motion state corresponding to each moment according to the motion state characteristics of each moment through the first model;

the denoising processing module is used for inputting the motion states corresponding to the moments into a second model, constructing a motion time sequence according to time sequence through the second model for the motion states corresponding to the moments, and denoising the motion time sequence;

and the state determining module is used for determining the target motion state of the terminal equipment in the first time period according to the motion time sequence after the denoising processing.

A third aspect of the embodiment of the present application provides an electronic device, including a memory and a processor, where the memory stores a computer program, where the computer program when executed by the processor causes the processor to implement any one of the motion state identification methods disclosed in the embodiments of the present application.

A fourth aspect of the embodiment of the present application provides a computer readable storage medium storing a computer program, where the computer program when executed by a processor implements any one of the motion state recognition methods disclosed in the embodiment of the present application.

Compared with the related art, the embodiment of the application has the following beneficial effects:

The method comprises the steps of acquiring the motion state characteristics of the terminal equipment at each moment in a first time period, determining the motion state corresponding to each moment according to the motion state characteristics of each moment through a first model, accurately outputting the motion state corresponding to each moment, constructing a motion time sequence according to time sequence through a second model on the basis of the motion state corresponding to each moment, denoising the motion time sequence, determining the target motion state of the terminal equipment in the first time period according to the motion time sequence after denoising, analyzing and denoising the motion state corresponding to each moment identified from the angle of the time sequence, improving the accuracy of the final output target motion state while improving the accuracy of the motion state corresponding to each moment identified, obtaining the acquired motion state characteristics according to the acceleration data and the air pressure data acquired by the terminal equipment, increasing the identifiable motion state quantity, and improving the identification accuracy of the motion state of a vertical space.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic diagram of an application scenario of a motion state recognition method disclosed in an embodiment of the present application;

FIG. 2 is a schematic flow chart of a method for identifying a motion state according to an embodiment of the present application;

FIG. 3 is a flow chart of another method for identifying motion states according to an embodiment of the present application;

FIG. 4 is a graph of the results before and after denoising a motion time series according to one embodiment;

FIG. 5 is a flow chart of adjusting probability thresholds in accordance with one embodiment disclosure;

FIG. 6 is a flow chart of motion state data processing as disclosed in one embodiment;

Fig. 7 is a schematic structural diagram of a motion state recognition device according to an embodiment of the present application;

FIG. 8 is a schematic diagram of another motion state recognition device according to an embodiment of the present application;

Fig. 9 is a schematic structural diagram of an electronic device according to an embodiment of the disclosure.

Detailed Description

The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.

It should be noted that the terms "comprising" and "having" and any variations thereof in the embodiments of the present application and the accompanying drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus.

The embodiment of the application discloses a method, a device, electronic equipment and a storage medium for identifying a motion state, which can improve the accuracy of motion state identification. The following will describe in detail.

Referring to fig. 1, fig. 1 is a schematic diagram of an application scenario of a motion state recognition method according to an embodiment of the present application. As shown in fig. 1, the terminal device 10 and the cloud 20 may be included, where the terminal device 10 is communicatively connected to the cloud 20. Cloud 20 may include, but is not limited to, a server or a cluster of servers. The terminal device 10 may collect motion state data through the sensing device, and extract motion state features at various moments according to the collected motion state data, where the motion state data includes acceleration data and barometric pressure data. The terminal device 10 may upload the extracted motion state features at each time to the cloud 20, so that the cloud 20 obtains the motion state features at each time in the first time period. The cloud 20 inputs the motion state characteristics of each moment into a first model, determines the motion state corresponding to each moment according to the motion state characteristics of each moment through the first model, inputs the motion state corresponding to each moment into a second model, constructs a motion time sequence according to time sequence for the motion state corresponding to each moment through the second model, performs denoising processing for the motion time sequence, and determines the target motion state of the terminal equipment 10 in the first time period according to the motion time sequence after denoising processing. Or the terminal device 10 acquires the motion state characteristics of the terminal device 10 itself at various moments in the first period. The terminal device 10 inputs the motion state characteristics of each moment into a first model, determines the motion state corresponding to each moment according to the motion state characteristics of each moment through the first model, inputs the motion state corresponding to each moment into a second model, constructs a motion time sequence according to time sequence for the motion state corresponding to each moment through the second model, performs denoising processing for the motion time sequence, and determines the target motion state of the terminal device 10 in the first time period according to the motion time sequence after denoising processing.

Referring to fig. 2, fig. 2 is a flow chart of a method for identifying a movement state according to an embodiment of the application, and the method can be applied to the cloud end 20. As shown in fig. 2, the method may include the steps of:

210. And acquiring the motion state characteristics of the terminal equipment at each moment in the first time period, wherein the motion state characteristics are extracted according to motion state data acquired by the terminal equipment, and the motion state data comprise acceleration data and barometric pressure data.

In the embodiment of the application, the terminal equipment acquires the motion state data of the terminal equipment in real time through the sensing equipment, wherein the sensing equipment at least comprises an accelerometer and a barometer, the motion state data at least comprises acceleration data and barometric pressure data, and the acceleration data can be triaxial acceleration data.

The terminal equipment performs feature extraction on the motion state data acquired in real time, and motion state features of the terminal equipment at all times in a first time period are obtained through extraction. The cloud acquires the motion state characteristics of the terminal equipment at each moment in a first time period, wherein the motion state characteristics are extracted by the terminal equipment in a terminal equipment uploading mode. Wherein the motion state features include acceleration data features and barometric pressure data features. The first time period is a set time period for analyzing the motion state.

In some embodiments, the acceleration data features may include time domain features of the acceleration data and frequency domain features of the acceleration data. The time domain features of the acceleration may include mean, median, polar difference, absolute deviation sum, maximum, minimum, absolute sum, sample entropy, quartile range, and variance, and the frequency domain features of the acceleration data may include mean, variance, frequency domain entropy, kurtosis, skewness, band energy, and the like. The barometric pressure characteristics may include slope, range, mean and variance, etc. The quartile moment represents the difference between the data arranged at the 75% position and the data arranged at the 25% position after the data are ordered in the order from the top to the bottom; the median represents the data ranked at the 50% position after ranking; sample entropy represents a similarity measure of data. The frequency domain features are the change rule of the data in terms of frequency, and the time domain features are the change rule of the data along with time change.

In the embodiment of the application, the air pressure data are data reflecting the height difference of the terminal equipment in the vertical horizontal plane. Under the condition that the trend of movement on the same horizontal plane is gentle, the change trend of the air pressure data is slow or randomly fluctuated, so that the air pressure data has a good effect of identifying the spatial movement in the vertical direction. The terminal device reflects the air pressure change trend at least by extracting the slope and the range characteristics of the air pressure data.

In the embodiment of the application, the average value characteristics are in different motion states such as sitting, standing, lying, walking and running, and the motion intensity in unit time is different, so that the terminal equipment evaluates the motion intensity in unit time by extracting the average value characteristics.

Under different motion states such as sitting, running, standing, sitting and lying, the change range of the acceleration data is wider, so that the terminal equipment distinguishes the motion with larger motion intensity difference through the extremely poor characteristic of the acceleration data.

Under the motion states of upstairs, downstairs and the like, the kinematics of the terminal equipment user in the same period are single-foot stress, so that the gravity center of the acceleration data is offset, the offset directions of the upstairs and the downstairs are just opposite, the offset condition can be evaluated through the mean value characteristic and the median characteristic of the acceleration data, and the symmetrical motion and the asymmetrical motion can be distinguished.

Under the motion state of riding a bicycle, riding a scooter and the like, due to the fact that the road surface is uneven, the motion randomness is large, the obtained acceleration data does not normally show a regular state, and therefore regular and irregular motions can be distinguished through sample entropy characteristics.

In the embodiment of the application, the accuracy of identifying the target motion state of the terminal equipment can be improved and the number of the motion states which can be identified can be increased by extracting the specific motion state characteristics.

220. And inputting the motion state characteristics of each moment into a first model, and determining the motion state corresponding to each moment according to the motion state characteristics of each moment through the first model.

In the embodiment of the application, the cloud end inputs the motion state characteristics of each moment in the first time period into the first model, and the first model determines the motion state corresponding to each moment according to the type and the number of the state characteristic intervals of the feature values of the motion state characteristics of each moment after receiving the motion state characteristics of each moment. For example, the state characteristics at each time include an acceleration range, an acceleration average and a barometric average, the acceleration range corresponding to time a, the acceleration average and the barometric average are (1,2,0.1), the acceleration range corresponding to time B, and the acceleration average and the barometric average are (10,1,0.1). The running exercise state can be set, the acceleration range is very poor, the intervals of the acceleration average value and the air pressure average value are respectively [8, 15], [0,3] and [0,1], and the intervals of the acceleration average value and the air pressure average value are respectively [0,5], [0,3] and [0,1] in the running exercise state. At this time, it can be seen that the three state feature intervals corresponding to the three motion state feature values corresponding to the time a are all in the walking state and are in the two state feature intervals corresponding to the running state, and then the first model can output that the motion state corresponding to the time a is the walking state. Similarly, the three state feature intervals corresponding to the running states where the three motion state feature values corresponding to the time B are all located are the two state feature intervals corresponding to the walking state, and then the first model can output that the motion state corresponding to the time B is the running state.

In the embodiment of the application, the acquired motion state data can be segmented according to a certain repetition rate to obtain a plurality of groups of sample data, and each group of sample data is input into the first model as a training set to train the first model. The repetition rate is the same ratio between the sample data and the collected motion state data, so that the training data can be attached to the collected motion state data, and the first model can be trained by enough sample data.

In some embodiments, the cloud end inputs the motion state features of each moment into the first model, and after the first model receives the motion state features of each moment, the first model may further output probability values corresponding to different motion states of each moment, and select the motion state corresponding to the maximum probability value as the motion state corresponding to the moment. For example, the first model outputs probability values of the motion state a and the motion state B corresponding to the time a and the time B according to the motion state characteristics of the respective time, wherein the probability value of the motion state a corresponding to the time a is 0.6, the probability value of the motion state B corresponding to the time a is 0.4, the probability value of the motion state a corresponding to the time B is 0.3, and the probability value of the motion state B corresponding to the time B is 0.7. The first model in the terminal device may then determine that the motion state corresponding to time a is a and the motion state corresponding to time B is B.

230. And inputting the motion states corresponding to the moments into a second model, constructing a motion time sequence according to the time sequence for the motion states corresponding to the moments through the second model, and denoising the motion time sequence.

In the embodiment of the application, after the cloud outputs the motion state corresponding to each moment in the first model, the motion state corresponding to each moment can be input into the second model, the second model sorts the moments according to the sequence of time and constructs a motion time sequence, and the constructed motion time sequence comprises the moments and the motion states corresponding to the moments. After the second model builds the motion time sequence, the motion state which is possibly abnormal in the motion time sequence is corrected, so that the motion time sequence is denoised. The denoising process may be data smoothing, that is, a motion state with a relatively high frequency of occurrence in the entire motion time sequence is allocated to a motion state with a relatively low frequency of occurrence according to a certain policy. For the denoising processing of the motion time series, the method may include determining an abnormal motion state in the motion time series, and correcting the abnormal motion state to a motion state corresponding to a time preceding or following the corresponding time. Wherein less than a number of threshold states of motion may be determined as abnormal states of motion. For example, the motion time sequence includes 20 times and motion states corresponding to the 20 times, and then the total number of motion states less than 2 in the motion time sequence can be determined as abnormal motion states, and the motion states are corrected to motion states corresponding to a time before or a time after the corresponding time, so as to finally implement denoising processing of the motion time sequence. In addition, according to the motion states corresponding to all moments in the motion time sequence after denoising, one motion state can be determined to correspond to the motion time sequence after denoising.

240. And determining the target motion state of the terminal equipment in the first time period according to the motion time sequence after denoising.

In the embodiment of the application, after the cloud acquires the denoised motion time sequence output by the second model, the cloud can determine the motion state with the largest corresponding time according to the number of times corresponding to different motion states in the motion time sequence, and determine the motion state as the target motion state of the terminal equipment in the first time period. Or the cloud determines the motion state corresponding to the motion time sequence after denoising as the target motion state of the terminal equipment in the first time period.

In the embodiment of the application, the cloud transmits the target motion state to the terminal equipment after determining the target motion state of the terminal equipment in the first time period, so that the terminal equipment outputs the target motion state. Wherein the first model and the second model are stored in the cloud.

By adopting the embodiment, the motion states at all times can be denoised from the time sequence, the accuracy of the finally output target motion state is improved, the acquired motion state characteristics are obtained according to the acceleration data and the air pressure data acquired by the terminal equipment, the number of identifiable motion states is increased, and the accuracy of identifying the motion states in the vertical space is improved. And the cloud is used for executing complex operations such as determining the motion state corresponding to each moment, constructing a motion time sequence, denoising the motion time sequence to obtain a target motion state and the like, so that the operation pressure of the terminal equipment can be effectively reduced.

In some embodiments, the motion state identification method flow of steps 210 to 240 may be applied to the terminal device 10 in fig. 1.

In the embodiment of the application, the terminal equipment acquires the motion state data in real time through the sensing equipment, performs feature extraction according to the motion state data, and can acquire the motion state features of the terminal equipment at all times in the first time period after acquiring the motion state features of the terminal equipment at all times in the first time period.

After the terminal equipment acquires the motion state characteristics of the terminal equipment at each moment in the first time period, the motion state characteristics of the terminal equipment at each moment are input into a first model, and the motion state corresponding to each moment is determined according to the motion state characteristics of the terminal equipment at each moment through the first model. The terminal equipment inputs the motion states corresponding to the moments output by the first model into the second model, constructs a motion time sequence according to time sequence for the motion states corresponding to the moments through the second model, and performs denoising treatment for the motion time sequence. And the terminal equipment determines the target motion state of the terminal equipment in the first time period according to the motion time sequence after the denoising processing. The terminal device outputs the target motion state. Wherein the first model and the second model are stored in the terminal device.

In the embodiment of the application, the sensing equipment in the terminal equipment is used for collecting the motion state data, and after the data processing such as feature extraction is carried out on the motion state data, the terminal equipment automatically executes complex operations such as determining the motion state corresponding to each moment, constructing the motion time sequence and denoising the motion time sequence to obtain the target motion state, so that the data transmission operation in the motion state identification process can be reduced, the motion state identification error caused by data loss in the data transmission process is avoided, and the accuracy of the motion state identification is improved.

In an embodiment, referring to fig. 3, fig. 3 is a flow chart of another exercise status recognition method according to an embodiment of the present application, and the method can be applied to the cloud end 20. As shown in fig. 3, the method may include the steps of:

310. And acquiring the motion state characteristics of the terminal equipment at each moment in the first time period, wherein the motion state characteristics are extracted according to motion state data acquired by the terminal equipment, and the motion state data comprise acceleration data and barometric pressure data.

320. And inputting the motion state characteristics of each moment into a first model, and determining the probability that each moment corresponds to each preset motion state respectively according to the motion state characteristics of each moment through the first model, wherein the sum of the probabilities that each preset motion state corresponds to each preset motion state is 1.

In the embodiment of the application, the cloud end inputs the motion state characteristics of each moment into the first model, and the first model can output probability values corresponding to different preset motion states at each moment according to an algorithm between the characteristic values and the probability values of the different preset motion states after receiving the motion state characteristics of each moment. For example, according to the motion state characteristics of each moment, the probability values of the output moment a corresponding to the preset motion state a, the preset motion state B and the preset motion state c are respectively 0.6, 0.3 and 0.1, and the probability values of the output moment B corresponding to the preset motion state a, the preset motion state B and the preset motion state c are respectively 0.1, 0.1 and 0.8. The preset motion state is a set identifiable motion state. For each moment, the sum of probability values of the corresponding preset motion states is 1, that is, the motion state corresponding to each moment is one of the preset motion states. The algorithm for calculating the probability values corresponding to different preset motion states for the motion state features at each time is not particularly limited herein.

In the embodiment of the application, the first model may be a model formed by a plurality of decision trees, each decision tree outputs a probability value corresponding to a preset motion state, the cloud traverses the motion state features of each moment through all the decision trees, and each moment obtains a corresponding array formed by a plurality of probability values.

330. And comparing the probabilities respectively corresponding to the preset motion states at all times with probability thresholds corresponding to the preset motion states, wherein the probability thresholds corresponding to the preset motion states are all larger than 0.5.

In the embodiment of the application, the cloud obtains probabilities that each moment corresponds to each preset motion state respectively, that is, for each moment, each probability value is respectively available for each preset motion state corresponding to different preset motion states. And the cloud compares each probability value in each moment with a probability threshold corresponding to a preset motion state corresponding to the probability value. For example, the probability values of the time a corresponding to the preset motion state a, the preset motion state b, and the preset motion state c are respectively 0.6, 0.3, and 0.1, the probability threshold value of the preset motion state a corresponding to 0.5, the probability threshold value of the preset motion state b corresponding to 0.6, and the probability threshold value of the preset motion state c corresponding to 0.5. The first model may compare the probability 0.6 of the preset motion state a corresponding to the time a with the probability threshold value corresponding to the preset motion state a being 0.5, compare the probability 0.3 of the preset motion state b corresponding to the time a with the probability threshold value corresponding to the preset motion state b being 0.6, and compare the probability 0.1 of the preset motion state c corresponding to the time a with the probability threshold value corresponding to the preset motion state c being 0.5. The probability threshold value of each preset motion state corresponding to each moment is larger than 0.5, so that the probability threshold value of each preset motion state corresponding to each subsequent moment can be guaranteed to be at most one preset motion state.

340. If the probability of the first moment corresponding to the first motion state is larger than the probability threshold value corresponding to the first motion state, the motion state corresponding to the first moment is determined to be the first motion state, the first moment is any one of the moments, and the first motion state is any one of the preset motion states.

In the embodiment of the application, for any one of the moments, namely the first moment, the cloud terminal compares each probability value in the first moment with a probability threshold corresponding to a preset motion state corresponding to the probability value, and if the probability value corresponding to the first motion state at the first moment is larger than the probability threshold corresponding to the first motion state, the cloud terminal can determine the first motion state as the motion state corresponding to the first moment. Wherein the first motion state is any one of the preset motion states. Repeating the steps until the terminal equipment determines the motion state corresponding to each moment. For example, the probability values of the time a corresponding to the preset motion state a, the preset motion state b, and the preset motion state c are respectively 0.6, 0.3, and 0.1, the probability threshold value of the preset motion state a corresponding to 0.5, the probability threshold value of the preset motion state b corresponding to 0.6, and the probability threshold value of the preset motion state c corresponding to 0.5. The probability value 0.6 of the preset motion state a corresponding to the time A is larger than the probability threshold value 0.5 corresponding to the preset motion state a, so that the terminal equipment determines the preset motion state a as the motion state corresponding to the time A.

In the embodiment of the application, the motion state corresponding to each moment can be more accurately determined by determining the probability that each moment corresponds to each preset motion state and determining the motion state corresponding to each moment according to whether the probability meets the probability threshold, and the sum of the probability values of the corresponding preset motion states is 1 for each moment, so that the motion state corresponding to each moment is ensured to be one of the preset motion states, and meanwhile, the probability threshold of each preset motion state corresponding to each moment is greater than 0.5, so that the motion state corresponding to each subsequent moment can be ensured to be at most one preset motion state, and the determined motion state corresponding to each moment is more reasonable.

350. And inputting the motion states corresponding to each moment into a second model, constructing a motion time sequence according to time sequence for the motion states corresponding to each moment through the second model, determining an abnormal moment according to the motion states corresponding to each moment in the motion time sequence, wherein the motion states corresponding to the abnormal moment are different from the motion states corresponding to the previous moment of the abnormal moment and the motion states corresponding to the next moment of the abnormal moment, and the motion states corresponding to the previous moment are the same as the motion states corresponding to the next moment.

In the embodiment of the application, the cloud inputs the motion states corresponding to the moments output by the first model into the second model. The second model sorts the moments according to the sequence of time and constructs a motion time sequence, wherein the constructed motion time sequence comprises the moments and the motion states corresponding to the moments. The second model judges whether each moment is an abnormal moment according to the motion state corresponding to each moment, the motion state corresponding to the last moment in the motion time sequence and the motion state corresponding to the next moment. For any one time, if the motion state corresponding to the time is different from the motion state corresponding to the last time and the motion state corresponding to the next time in the motion time sequence, and the motion state corresponding to the last time and the motion state corresponding to the next time in the motion time sequence are the same, the time can be determined to be an abnormal time. For example, referring to fig. 4, fig. 4 is a schematic diagram of the results before and after denoising the motion time series according to an embodiment. In fig. 4, the motion time sequence at least includes 8 moments, and the motion states corresponding to the first 8 moments sequenced in sequence are lying, sitting, lying, standing, lying respectively. At this time, the motion state corresponding to the 3 rd time is different from the motion state corresponding to the 2 nd time, the motion state corresponding to the 3 rd time is also different from the motion state corresponding to the 4 th time, and the motion state corresponding to the 2 nd time is the same as the motion state corresponding to the 4 th time. Similarly, the motion state corresponding to the 6 th time is different from the motion state corresponding to the 5 th time, the motion state corresponding to the 6 th time is also different from the motion state corresponding to the 7 th time, and the motion state corresponding to the 5 th time is the same as the motion state corresponding to the 7 th time. The second model may determine the 3 rd and 5 th moments as abnormal moments.

In the embodiment of the present application, for any time, if the motion state corresponding to the time is different from the motion state corresponding to the previous time and the motion state corresponding to the next time in the motion time sequence, but the motion state corresponding to the previous time and the motion state corresponding to the next time in the motion time sequence are different, the time is not determined as the abnormal time. As shown in fig. 4, at the 5 th time, the motion state corresponding to the 5 th time is different from the motion state corresponding to the 4 th time, and the motion state corresponding to the 5 th time is also different from the motion state corresponding to the 6 th time, but the motion state corresponding to the 4 th time is different from the motion state corresponding to the 6 th time. Then the second model may not determine the 5 th time instant as the abnormal time instant.

360. And correcting the motion state corresponding to the abnormal moment according to the motion state corresponding to the previous moment or the motion state corresponding to the next moment to obtain a motion time sequence after denoising.

In the embodiment of the present application, after determining the abnormal time in the motion time sequence in the second model, for each abnormal time, the motion state corresponding to the abnormal time is corrected to the same motion state as the motion state corresponding to the previous time or the motion state corresponding to the next time of the abnormal time, so as to achieve the correction of the motion state corresponding to the abnormal time. And correcting the motion state corresponding to each abnormal moment in the motion time sequence, so as to realize denoising processing of the motion time sequence and obtain a denoised motion time sequence. For example, in fig. 4, the motion states corresponding to the 2 nd and 4 th moments are all lying, so the motion state corresponding to the 3 rd moment can be corrected from sitting to lying. Similarly, the movement states corresponding to the 5 th time and the 7 th time are all lying, so the movement state corresponding to the 6 th time can be corrected to lie.

In the embodiment of the present application, for any time, if the motion state corresponding to the time is different from the motion state corresponding to the previous time and the motion state corresponding to the next time in the motion time sequence, but the motion state corresponding to the previous time and the motion state corresponding to the next time in the motion time sequence are different, the second model may not output the motion state corresponding to the time or output other motion states as the motion states corresponding to the time when the motion time sequence is processed. As shown in fig. 4, the second model may not output the lying motion state corresponding to the 5 th moment, or may output other motion states such as the standing as the motion state corresponding to the 5 th moment.

In the embodiment of the application, since the time ordered at the first time in the motion time sequence does not have the last time and the time ordered at the last time in the motion time sequence does not have the last time, when the second model performs denoising processing on the motion time sequence, the motion states corresponding to the first time and the last time can not be output.

In the embodiment of the application, in the second model, abnormal time in the motion time sequence can be sequentially determined according to the sequence of each time in the motion time sequence, and the motion state corresponding to the abnormal time can be sequentially corrected after the abnormal time is determined. The abnormal time in the motion time sequence can be determined without ordering the time in the motion time sequence, and the motion state corresponding to the abnormal time can be corrected one by one or simultaneously.

In the embodiment of the application, whether the motion state corresponding to each moment is reasonable or not is analyzed according to the motion state corresponding to each moment and the motion state corresponding to the last moment and the next moment, and the motion state corresponding to the unreasonable moment is corrected to realize the denoising processing of the motion time sequence, so that the motion state corresponding to each moment in the motion time sequence is more reasonable, and the accuracy of motion state identification can be integrally improved from a period of time.

370. Determining the corresponding time quantity of each motion state in the motion time sequence after denoising, and determining the motion state with the largest proportion in the motion time sequence after denoising according to the corresponding time quantity of each motion state;

In the embodiment of the application, after the motion time sequence after the denoising processing is output by the second model, the cloud side counts the time quantity corresponding to each motion state according to each time and the motion state corresponding to each time contained in the motion time sequence after the denoising processing, calculates the proportion of the time quantity corresponding to each motion state to the time quantity of all times in the motion time sequence after the statistics, and determines the motion state with the largest proportion in the motion time sequence after the denoising processing.

380. And if the corresponding duty ratio of the motion state with the largest duty ratio in the motion time sequence after the denoising processing is larger than the duty ratio threshold value, determining the motion state with the largest duty ratio as the target motion state of the terminal equipment in the first time period.

In the embodiment of the application, after determining the motion state with the largest duty ratio in the motion time sequence after denoising, the cloud compares the duty ratio with the duty ratio threshold, and if the duty ratio is larger than the duty ratio threshold, the cloud can determine the motion state with the largest duty ratio as the target motion state of the terminal equipment in the first time period; if the duty ratio is smaller than or equal to the duty ratio threshold value, the cloud end does not determine the motion state with the largest duty ratio as the target motion state of the terminal equipment in the first time period, namely the target motion state of the terminal equipment in the first time period is not output temporarily.

In the embodiment of the application, the target motion state of the terminal equipment in a period of time is determined according to the motion state with the largest proportion in the motion time sequence, so that the motion state of the terminal equipment can be observed from a longer time scale, and the accuracy of the identified motion state of the terminal equipment can be improved.

In some embodiments, another motion state identification method flow of steps 310 to 380 may be applied to the terminal device 10 in fig. 1.

In one embodiment, referring to fig. 5, fig. 5 is a flow chart of adjusting probability thresholds as disclosed in one embodiment. Before the process of comparing the probabilities corresponding to the preset motion states at the respective moments and the probability thresholds corresponding to the preset motion states in step 330, the method further includes:

510. and acquiring historical state data corresponding to the terminal equipment, wherein the historical state data comprises motion states respectively corresponding to a plurality of historical moments.

In the embodiment of the application, the cloud acquires the historical state data corresponding to the terminal equipment, wherein the historical state data can be stored in a database of the terminal equipment or in the cloud. Before the current movement state identification process, the history state data is the movement state corresponding to each determined moment in the movement state identification process of the terminal equipment or the cloud end in the past.

520. Classifying each preset motion state according to the number of the historical moments corresponding to each preset motion state to obtain a first type motion state and a second type motion state, wherein the number of the historical moments corresponding to the first type motion state is larger than the number of the historical moments corresponding to the second type motion state.

In the embodiment of the application, the cloud counts the acquired historical data, determines the number of the historical moments corresponding to each preset motion state, classifies each preset motion state according to the number of the historical moments corresponding to each preset motion state, and can be divided into two types, namely a first type motion state and a second type motion state, wherein the number of the historical moments corresponding to the first type motion state is larger than the number of the historical moments corresponding to the second type motion state. Specifically, a number threshold is set, and the number of historical moments corresponding to each preset motion state is compared with the set number threshold. If the number of the historical moments corresponding to one preset motion state is greater than the set number threshold, the cloud end can classify the preset motion state into a first type of motion state. Otherwise, if the number of the historical moments corresponding to one preset motion state is smaller than or equal to the set number threshold, the cloud end can classify the preset motion state into a second type of motion state.

530. And adjusting the probability threshold value corresponding to each preset motion state contained in the first type of motion state and the probability threshold value corresponding to each preset motion state contained in the second type of motion state so that the probability threshold value corresponding to each preset motion state contained in the second type of motion state is larger than the probability threshold value corresponding to each preset motion state contained in the first type of motion state.

In the embodiment of the application, the cloud adjusts probability thresholds corresponding to the preset motion states included in the first type of motion state, and adjusts probability thresholds corresponding to the preset motion states included in the second type of motion state. Specifically, the probability threshold value corresponding to each preset motion state contained in the second type of motion state is increased, and meanwhile, the probability threshold value corresponding to each preset motion state contained in the first type of motion state is reduced; or only increasing the probability threshold value corresponding to each preset motion state contained in the second type of motion state; or only reducing the probability threshold value corresponding to each preset motion state contained in the first type of motion state. So as to achieve the purpose that the probability threshold value corresponding to each preset motion state contained in the second type of motion state is larger than the probability threshold value corresponding to each preset motion state contained in the first type of motion state. Wherein, the probability threshold value corresponding to each preset motion state contained in the first type of motion state and the probability threshold value corresponding to each preset motion state contained in the second type of motion state still need to be larger than 0.5 after adjustment.

In the embodiment of the application, the probability threshold corresponding to the less recognized motion states is increased or the probability threshold corresponding to the more recognized motion states is reduced, so that the more frequently-occurring motion states are easier to recognize and the less-occurring motion states are harder to recognize, and the intelligent level of motion state recognition can be improved.

In one embodiment, referring to FIG. 6, FIG. 6 is a flow chart of motion state data processing as disclosed in one embodiment. Before the process of acquiring the motion state characteristics of the terminal device at each moment in step 210 or step 310, the method further includes:

610. and carrying out data cleaning on the motion state data acquired by the terminal equipment according to the state value interval to obtain cleaned motion state data, wherein the cleaned motion state data is the motion state data in the state value interval.

In the embodiment of the application, after the terminal equipment collects the motion state data, the motion state data collected by the terminal equipment can be compared with the set state value interval to judge whether the motion state data is in the corresponding state value interval. And removing the motion state data which are not in the corresponding state value interval so as to clean the motion state data. The state value interval may be an empirical value, that is, an interval in which data may be located in a normal state, for example, the state value interval corresponding to the acceleration data is generally (0, 20). In addition, each motion state data is compared with a corresponding state value section, such as an acceleration data is compared with an acceleration section, and a barometric pressure data is compared with a barometric pressure section.

620. And separating acceleration data in the cleaned motion state data to obtain multi-axis acceleration data.

In the embodiment of the application, the terminal equipment separates acceleration data in the cleaned motion state data to obtain multi-axis acceleration data, such as triaxial acceleration data, and obtains four-axis acceleration data according to the acceleration data before separation, namely the total acceleration data, so as to reflect the overall acceleration condition and the acceleration condition in the triaxial direction.

630. And extracting the motion state characteristics of the terminal equipment at each moment in the first time period from the multi-axis acceleration data and the air pressure data in the cleaned motion state data.

In the embodiment of the application, the terminal equipment performs feature extraction on the multi-axis acceleration, and performs feature extraction on the air pressure data in the cleaned motion state data at the same time, so as to obtain the motion state features of the terminal equipment at all times in a first time period. The extracted motion state features are the same as those described in the above embodiments, and will not be described in detail here.

In the embodiment of the application, the motion state data used for extracting the state characteristics is more reasonable by cleaning the data and separating the acceleration data, and the extracted motion state characteristics can better reflect the motion condition of the terminal equipment, so that the accuracy of the follow-up target motion state identification is improved.

In one embodiment, the process of extracting the motion state characteristics of the terminal device at each moment in the first time period from the multi-axis acceleration data and the air pressure data included in the motion data set in step 630 may include:

In the embodiment of the application, the terminal equipment performs data segmentation on each axis of acceleration data in the multi-axis acceleration data according to the preset repetition rate and the sample time unit, and segments the multi-axis acceleration data into a plurality of segments of acceleration data to obtain segmented acceleration data. The terminal equipment also performs data segmentation on the air pressure data in the cleaned motion state data according to the same sample time unit, and segments the air pressure data into multiple segments of air pressure data to obtain segmented air pressure data. For example, each axis of acceleration data is sliced at a 50% repetition rate and 5s sample time units, and the cleaned barometric pressure data is sliced at a 5s sample time answer. The terminal equipment performs feature extraction on the segmented acceleration data and the segmented air pressure data to obtain motion state features of the terminal equipment at all times in a first time period.

In the embodiment of the application, the motion state data is subjected to data segmentation processing, so that a specific quantity of motion state data can be obtained from a plurality of motion state data acquired in real time, and the motion state characteristics of each moment in a first time period are obtained. The quantity of the motion state data to be processed can be reduced, and the efficiency of motion state identification is improved.

In some embodiments, if the sampling frequency of the collected acceleration data is different from the sampling frequency of the collected air pressure data, the terminal device may further use different sample units to perform data slicing on each axis of acceleration data and the cleaned air pressure data. For example, the frequency of collecting the acceleration data is 10 times per second, and the frequency of collecting the air pressure data is 2 times per second, so that the data segmentation can be performed on each axis of acceleration data by taking 100 sampling points as one sample unit, and the data segmentation can be performed on the cleaned air pressure data by taking 20 sampling points as another sample unit.

Referring to fig. 7, fig. 7 is a schematic structural diagram of a motion state recognition device according to an embodiment of the present application, where the motion state recognition device can be applied to a terminal device. As shown in fig. 7, the motion state recognition apparatus 700 may include: a feature acquisition module 710, a state acquisition module 720, a denoising processing module 730, and a state determination module 740.

The feature obtaining module 710 is configured to obtain a motion state feature of the terminal device at each moment in a first period, where the motion state feature is extracted according to motion state data collected by the terminal device, and the motion state data includes acceleration data and barometric pressure data;

The state acquisition module 720 is configured to input the motion state features at each time into the first model, and determine, according to the motion state features at each time, a motion state corresponding to each time through the first model;

The denoising processing module 730 is configured to input the motion states corresponding to the respective moments into the second model, construct a motion time sequence according to the time sequence for the motion states corresponding to the respective moments through the second model, and denoise the motion time sequence;

the state determining module 740 is configured to determine a target motion state of the terminal device in the first period according to the motion time sequence after the denoising process.

In one embodiment, the status acquisition module 720 is further configured to:

determining the probabilities of the respective preset motion states at the respective moments according to the motion state characteristics at the respective moments through a first model, wherein the sum of the probabilities of the respective preset motion states is 1;

The probability corresponding to each preset motion state at each moment is compared with the probability threshold corresponding to each preset motion state, and the probability threshold corresponding to each preset motion state is larger than 0.5;

If the probability of the first moment corresponding to the first motion state is larger than the probability threshold value corresponding to the first motion state, the motion state corresponding to the first moment is determined to be the first motion state, the first moment is any one of the moments, and the first motion state is any one of the preset motion states.

In one embodiment, the status acquisition module 720 is further configured to:

In one embodiment, the denoising processing module 730 is further configured to:

In one embodiment, the status determination module 740 is further configured to:

determining the corresponding time quantity of each motion state in the motion time sequence after denoising, and determining the motion state with the largest proportion in the motion time sequence after denoising according to the corresponding time quantity of each motion state;

And if the corresponding duty ratio of the motion state with the largest duty ratio in the motion time sequence after the denoising processing is larger than the duty ratio threshold value, determining the motion state with the largest duty ratio as the target motion state of the terminal equipment in the first time period.

Referring to fig. 8, fig. 8 is a schematic structural diagram of another motion state recognition device according to an embodiment of the present application, and the motion state recognition device can be applied to a terminal device. As shown in fig. 8, the movement state recognition device 700 further includes: a data processing module 750.

The data processing module 750 is configured to perform data cleaning on the motion state data collected by the terminal device according to the state value interval, so as to obtain cleaned motion state data, where the cleaned motion state data is motion state data in the state value interval;

and extracting the motion state characteristics of the terminal equipment at each moment in the first time period from the multi-axis acceleration data and the air pressure data in the cleaned motion state data.

In one embodiment, data processing module 750 is further configured to:

In some embodiments, the movement state recognition device in the above embodiments may also be used for cloud.

Referring to fig. 9, fig. 9 is a schematic structural diagram of an electronic device according to an embodiment, which is applicable to driving a vehicle, and is not particularly limited herein. As shown in fig. 9, the electronic device 900 may include:

a memory 910 storing executable program code;

A processor 920 coupled with the memory 910;

The processor 920 invokes executable program codes stored in the memory 910 to perform any one of the motion state recognition methods disclosed in the embodiments of the present application.

The embodiment of the application discloses a computer readable storage medium which stores a computer program, wherein the computer program enables a computer to execute any motion state identification method disclosed by the embodiment of the application.

Embodiments of the present application disclose a computer program product comprising a non-transitory computer readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform any of the motion state identification methods disclosed in the embodiments of the present application.

It should be appreciated that reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present application. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. Those skilled in the art will also appreciate that the embodiments described in the specification are alternative embodiments and that the acts and modules referred to are not necessarily required for the present application.

In various embodiments of the present application, it should be understood that the sequence numbers of the foregoing processes do not imply that the execution sequences of the processes should be determined by the functions and internal logic of the processes, and should not be construed as limiting the implementation of the embodiments of the present application.

The units described above as separate components may or may not be physically separate, and components shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiment.

In addition, each functional unit in the embodiments of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.

The integrated units described above, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present application, or a part contributing to the prior art or all or part of the technical solution, may be embodied in the form of a software product stored in a memory, comprising several requests for a computer device (which may be a personal computer, a server or a network device, etc., in particular may be a processor in a computer device) to execute some or all of the steps of the above-mentioned method of the various embodiments of the present application.

Those of ordinary skill in the art will appreciate that all or part of the steps of the various methods of the above embodiments may be implemented by a program that instructs associated hardware, the program may be stored in a computer readable storage medium including Read-Only Memory (ROM), random access Memory (Random Access Memory, RAM), programmable Read-Only Memory (Programmable Read-Only Memory, PROM), erasable programmable Read-Only Memory (Erasable Programmable Read Only Memory, EPROM), one-time programmable Read-Only Memory (OTPROM), electrically erasable programmable Read-Only Memory (EEPROM), compact disc Read-Only Memory (Compact Disc Read-Only Memory, CD-ROM) or other optical disk Memory, magnetic disk Memory, tape Memory, or any other medium that can be used for carrying or storing data.

The above describes in detail a method, an apparatus, an electronic device and a storage medium for identifying a motion state disclosed in the embodiments of the present application, and specific examples are applied to illustrate the principles and implementation manners of the present application, where the foregoing description of the embodiments is only used to help understand the method and core idea of the present application. Meanwhile, as those skilled in the art will have variations in the specific embodiments and application scope in accordance with the ideas of the present application, the present description should not be construed as limiting the present application in view of the above.

Claims

1. A method of motion state identification, the method comprising:

determining a target motion state of the terminal equipment in the first time period according to the denoised motion time sequence;

The determining, by the first model, the motion state corresponding to each moment according to the motion state features of each moment includes:

determining the probabilities of the respective moments corresponding to the respective preset motion states according to the motion state characteristics of the respective moments through the first model, wherein the sum of the probabilities of the respective preset motion states is 1; comparing the probabilities respectively corresponding to the preset motion states at all the moments with probability thresholds corresponding to the preset motion states, wherein the probability thresholds corresponding to the preset motion states are all larger than 0.5; if the probability of the first moment corresponding to the first motion state is larger than the probability threshold value corresponding to the first motion state, determining the motion state corresponding to the first moment as the first motion state, wherein the first moment is any one of the moments, and the first motion state is any one of the preset motion states;

Before the probabilities respectively corresponding to the preset motion states at the moments are compared with the probability threshold values corresponding to the preset motion states, the method further comprises:

Acquiring historical state data corresponding to the terminal equipment, wherein the historical state data comprises motion states respectively corresponding to a plurality of historical moments; classifying each preset motion state according to the number of corresponding historical moments of each preset motion state in the historical state data to obtain a first type motion state and a second type motion state, wherein the number of the corresponding historical moments of the first type motion state is larger than that of the corresponding historical moments of the second type motion state; and adjusting the probability threshold value corresponding to each preset motion state contained in the first type of motion state and the probability threshold value corresponding to each preset motion state contained in the second type of motion state so that the probability threshold value corresponding to each preset motion state contained in the second type of motion state is larger than the probability threshold value corresponding to each preset motion state contained in the first type of motion state.

2. The method of claim 1, wherein said denoising said motion time series comprises:

3. The method according to claim 1, characterized in that before the acquisition of the motion state characteristics of the terminal device at the respective moments in time within the first period of time, the method further comprises:

4. A method according to claim 3, characterized in that extracting the motion state characteristics of the terminal device at each moment in time within a first period of time from the multi-axis acceleration data and the barometric pressure data contained in the motion state data comprises:

5. The method according to any one of claims 1 to 4, wherein the determining, according to the denoised motion time sequence, the target motion state of the terminal device in the first period of time includes:

6. A motion state recognition apparatus, the apparatus comprising:

The state determining module is used for determining a target motion state of the terminal equipment in the first time period according to the motion time sequence after denoising;

the state acquisition module is further used for determining the probability that each moment corresponds to each preset motion state respectively according to the motion state characteristics of each moment through the first model, and the sum of the probabilities that each preset motion state corresponds to is 1; comparing the probabilities respectively corresponding to the preset motion states at all the moments with probability thresholds corresponding to the preset motion states, wherein the probability thresholds corresponding to the preset motion states are all larger than 0.5; if the probability of the first moment corresponding to the first motion state is larger than the probability threshold value corresponding to the first motion state, determining the motion state corresponding to the first moment as the first motion state, wherein the first moment is any one of the moments, and the first motion state is any one of the preset motion states;

The state acquisition module is further used for acquiring historical state data corresponding to the terminal equipment, wherein the historical state data comprises a plurality of motion states corresponding to historical moments respectively; classifying each preset motion state according to the number of corresponding historical moments of each preset motion state in the historical state data to obtain a first type motion state and a second type motion state, wherein the number of the corresponding historical moments of the first type motion state is larger than that of the corresponding historical moments of the second type motion state; and adjusting the probability threshold value corresponding to each preset motion state contained in the first type of motion state and the probability threshold value corresponding to each preset motion state contained in the second type of motion state so that the probability threshold value corresponding to each preset motion state contained in the second type of motion state is larger than the probability threshold value corresponding to each preset motion state contained in the first type of motion state.

7. An electronic device comprising a memory and a processor, the memory having stored therein a computer program which, when executed by the processor, causes the processor to implement the method of any of claims 1 to 5.

8. A computer readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, implements the method according to any one of claims 1 to 5.