CN108523868A - Self-calibration system and method for blood pressure measurement - Google Patents

Abstract

The invention relates to the technical field of blood pressure detection, in particular to a self-calibration system and method for blood pressure measurement. The system includes a blood pressure collection unit, a human body posture collection unit and a processing unit. The blood pressure collection unit is used to collect human blood pressure detection signals. The human body posture collection unit is used to collect human body movement signals. The processing unit is equipped with a blood pressure detection model and a human body posture recognition model. and records the eigenvalue table of the coefficients of the blood pressure detection model under different human behavior modes; the processing unit is used to identify the current human behavior mode through the human body posture recognition model according to the human motion signal, and bring the corresponding coefficients into the blood pressure according to the eigenvalue table The detection model is used to obtain the human blood pressure detection value through the blood pressure detection model according to the human blood pressure detection signal. The method is realized based on the system. The invention can preferably realize the automatic calibration of the blood pressure detection model in the process of continuous blood pressure monitoring.

Description

Self-calibrating system and method for blood pressure measurement

technical field

The invention relates to the technical field of blood pressure detection, in particular to a self-calibration system and method for blood pressure measurement.

Background technique

Blood pressure (Blood Pressure, BP) is an important physiological health indicator of the human body, which can provide important parameter basis for the diagnosis and treatment of cardiovascular diseases.

Existing blood pressure measurement can be divided into two categories: invasive blood pressure measurement and non-invasive blood pressure measurement. In invasive blood pressure measurement, the catheter of the pressure sensor needs to be inserted into the aorta or heart of the measurement object to detect the blood pressure signal. The measurement result is accurate, but preparation Long time, high requirements on the subjects and easy to cause complications; non-invasive blood pressure measurement generally obtains blood pressure by contacting the surface of the human body to obtain characteristic signals for analysis and processing, and will not cause trauma to the subjects, so non-invasive blood pressure measurement is more suitable for routine blood pressure measurement needs.

The automatic and continuous measurement of blood pressure has great practical significance in medicine. For example, in clinical medicine, continuous monitoring of blood pressure is required for critically ill patients and critically ill patients during surgery, so that once the patient has an accident, the medical staff can take effective rescue in time. measure. Existing non-invasive blood pressure measurement methods such as Korotkoff audiometry, oscillometric method, arterial tension method, volume compensation method, etc., cannot perform continuous monitoring of blood pressure due to limitations of factors such as blood vessel elasticity recovery.

At present, non-invasive and continuous monitoring of the blood pressure of the measurement object is generally carried out based on the volume pulse wave. This method is mainly realized by establishing a correlation model between the volume pulse wave transit time and blood pressure. However, this blood pressure measurement model has certain limitations and can only be applied under conditions consistent with the modeling environment. Because the coefficients in the blood pressure measurement model that represent properties related to blood vessel viscosity and elasticity will not change unless the modeling environment is similar. However, in practical applications, the behavior patterns and intensity of the subjects are constantly changing, and there are differences between the measurement environment and the modeling environment, and the vascular characteristic parameters such as vascular viscosity and elasticity vary, resulting in the blood pressure measurement model no longer being effective. It is necessary to continuously correct the model coefficients to carry out effective follow-up blood pressure measurement, which cannot meet the needs of health management and monitoring of continuous blood pressure measurement.

Contents of the invention

The present invention provides a self-calibration system for blood pressure measurement, which can overcome some or some defects of the prior art.

According to the self-calibration system for blood pressure measurement of the present invention, it includes a blood pressure acquisition unit, a human body posture acquisition unit and a processing unit, the blood pressure acquisition unit is used to collect human body blood pressure detection signals, the human body posture acquisition unit is used to collect human body movement signals, and processes The unit is equipped with a blood pressure detection model, a human body posture recognition model, and an eigenvalue table that records the coefficients of the blood pressure detection model under different human behavior modes; the processing unit is used to identify the current human body behavior mode through the human body posture recognition model according to the human motion signal, And according to the eigenvalue table, the corresponding coefficients are brought into the blood pressure detection model, and then the human blood pressure detection value is obtained through the blood pressure detection model according to the human blood pressure detection signal.

In the present invention, when the blood pressure is detected on the human body, the behavior pattern of the human body at this time can be detected synchronously, so that different coefficients can be brought into the blood pressure detection model according to different behavior patterns of the human body, thereby effectively reducing the measurement The difference between the environment and the modeling environment realizes the automatic calibration in the blood pressure measurement, and then better realizes the continuous measurement of the blood pressure.

Preferably, the blood pressure collection unit includes a photoelectric sensor for collecting volumetric pulse waves of the human body. With reference to the Chinese invention patent with the notification number CN107157461A, the continuous measurement of blood pressure in the present invention can obtain the systolic blood pressure SBP and diastolic blood pressure only by collecting single-point photoplethysmography signals, thereby greatly reducing the signal acquisition. Difficulty and better measurement comfort, and the accuracy of the prediction model results is high, so that continuous blood pressure monitoring is well realized.

Preferably, the human body posture acquisition unit includes a three-axis acceleration sensor, a three-axis geomagnetic sensor and a three-axis gyroscope sensor, the three-axis acceleration sensor is used to collect acceleration data generated during human body movement, and the three-axis geomagnetic sensor is used to collect human body movement process The magnetic field data generated in the three-axis gyroscope sensor is used to collect the angular velocity data generated during the movement of the human body. Therefore, the current behavior pattern of the human body can be better recognized.

Preferably, an output unit is connected to the control unit, and the output unit is used to output the currently detected blood pressure value. Data output can thereby be facilitated.

Preferably, the control unit is connected to a cloud server, and the human body gesture recognition model and feature value table are all stored in the cloud server. Therefore, the storage and processing of data can be better facilitated.

Based on any of the above self-calibration systems for blood pressure measurement, the present invention also provides a self-calibration method for blood pressure measurement, which includes the following steps:

Step 1, collect the human blood pressure detection signal through a blood pressure acquisition unit, and establish a blood pressure detection model between the human blood pressure detection signal and the blood pressure value according to the measured blood pressure value;

Step 2, collecting human motion signals through a human body posture acquisition unit, and establishing a human body posture recognition model between the human body motion signals and the human body behavior patterns according to the actual human behavior patterns;

Step 3, collect the corresponding coefficient values of the blood pressure detection model when the human body is in different behavior modes, so as to establish the characteristic value table;

Step 4: After the blood pressure detection model, the human body posture recognition model and the feature value table are established, when the human blood pressure is detected, the human blood pressure detection signal is collected through the blood pressure acquisition unit and the human body motion signal is collected through the human body posture acquisition unit, and then A control unit identifies the current behavior pattern of the human body according to the human body motion signal, and then updates the corresponding coefficient to the blood pressure detection model through the feature value table, and then obtains the human blood pressure measurement value according to the human blood pressure detection signal and the blood pressure detection model after the updated coefficient .

Through the method of the present invention, the behavior pattern of the human body during blood pressure detection can be detected synchronously, so that different coefficients can be brought into the blood pressure detection model according to the different behavior patterns of the human body, thereby effectively reducing the measurement environment and construction cost. The difference between the model environments realizes the automatic calibration in the blood pressure measurement, and then better realizes the continuous measurement of the blood pressure.

Preferably, the blood pressure acquisition unit collects the body volume pulse wave as the human body blood pressure detection signal through the photoelectric sensor. Therefore, continuous blood pressure monitoring can be well realized.

Preferably, the human body attitude acquisition unit collects the acceleration data generated during the human body movement, the magnetic field data generated during the human body movement, and the The angular velocity data is used as a human motion signal.

Preferably, the control unit outputs the currently detected blood pressure value through an output unit. Data output can thereby be facilitated.

Preferably, the control unit performs data interaction with a cloud server, and stores the human body gesture recognition model and feature value table in the cloud server. Therefore, the storage and processing of data can be better facilitated.

Description of drawings

FIG. 1 is a schematic diagram of a system block diagram of a self-calibration system for blood pressure measurement in Embodiment 1;

Fig. 2 is a schematic diagram of the system block diagram of the establishment of the human body gesture recognition model and the recognition system in embodiment 1;

Fig. 3 is a schematic diagram of a system block diagram of an attitude calculation unit in Embodiment 1;

4 is a schematic flow chart of a self-calibration method for blood pressure measurement in Embodiment 1;

FIG. 5 is a schematic flowchart of the establishment of the human gesture recognition model and the recognition method in Embodiment 1.

Detailed ways

In order to further understand the content of the present invention, the present invention will be described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the examples are only for explaining the present invention and not for limiting it.

Example 1

As shown in Figure 1, this embodiment provides a self-calibration system for blood pressure measurement, which includes a blood pressure acquisition unit, a human body posture acquisition unit and a processing unit, the blood pressure acquisition unit is used to collect human blood pressure detection signals, human body posture acquisition The unit is used to collect human body motion signals, and the processing unit is equipped with a blood pressure detection model, a human body posture recognition model, and an eigenvalue table that records the coefficients of the blood pressure detection model in different human behavior modes; The model identifies the current behavior pattern of the human body, and brings the corresponding coefficients into the blood pressure detection model according to the eigenvalue table, and then obtains the human blood pressure detection value through the blood pressure detection model according to the human blood pressure detection signal.

In this embodiment, when the blood pressure is detected on the human body, the behavior pattern of the human body at this time can be detected synchronously, so that different coefficients can be brought into the blood pressure detection model according to different behavior patterns of the human body, thereby effectively reducing the The difference between the measurement environment and the modeling environment realizes the automatic calibration in the blood pressure measurement, and then better realizes the continuous measurement of the blood pressure.

In this embodiment, the blood pressure collection unit includes a photoelectric sensor for collecting the volumetric pulse wave of the human body.

With reference to the Chinese invention patent with the notification number CN107157461A, the continuous measurement of blood pressure in this embodiment can obtain the systolic blood pressure SBP and diastolic blood pressure better only by collecting single-point photoplethysmography signals, thereby greatly reducing the signal It is difficult to obtain, better improves the comfort of measurement, and the prediction model results have higher precision, thus realizing continuous blood pressure monitoring well.

In this embodiment, the human body attitude acquisition unit includes a three-axis acceleration sensor, a three-axis geomagnetic sensor and a three-axis gyroscope sensor. The magnetic field data generated during exercise, and the three-axis gyroscope sensor is used to collect angular velocity data generated during human movement.

In this embodiment, the human body posture acquisition unit can be used to sense body movements of the human body, such as waving, bouncing, walking, jumping, etc., through the regular movements of different limbs of the human body, the current actions of the user can be better recognized.

Referring to FIG. 2 , the three-axis acceleration sensor, the three-axis geomagnetic sensor and the three-axis gyroscope sensor in this embodiment all use MEMS sensors, so as to better capture and record human body movements in real time. And since the three-axis acceleration sensor, the three-axis geomagnetic sensor and the three-axis gyroscope sensor can be integrated in a human body wearable device at the same time, by setting the human body wearable device at different positions of the user, the corresponding parts of the human body can be better realized. The detection of motion posture at the position can better realize the detection of human behavior patterns by wearing multiple human body wearable devices.

Wherein, the human body wearable device can also be provided with a data processing module for performing data processing on the human body posture acquisition unit, and of course the data processing module can also be provided at the control unit.

In this embodiment, the data processing module includes a data preprocessing unit, a state transition diagram establishment unit, an attitude calculation unit, a first weighted calculation unit, a fusion classification unit and a second weighted calculation unit; the data preprocessing unit is used for three-axis The signals collected by the acceleration sensor, the three-axis geomagnetic sensor and the three-axis gyroscope sensor are preprocessed, and a plurality of feature points are extracted; the state transition diagram establishment unit is used to establish or match the state transition diagram model according to the plurality of feature points The attitude calculation unit is used to calculate the attitude of the data processed by the data preprocessing unit to obtain the three-dimensional posture information of the human body; the first weighted calculation unit is used to compare one or more of the plurality of feature points with The weighted calculation is performed on the three-dimensional posture information of the human body, and the fusion classification unit is used to establish or match the human body posture pre-classification model according to the weighted calculation results; the second weighted calculation unit is used to perform weighted calculations on the matching results of the state transition diagram model and the human body posture pre-classification model , so as to establish or match the human gesture recognition model.

In this embodiment, the human body posture acquisition unit can be used to collect human body motion information, so that the waveform of relevant data can be obtained, and then the existing sliding window method can be used to extract multiple frequency domain, time domain, and time-frequency feature points from the signal waveform At the same time, the attitude calculation module can be used to obtain the three-dimensional attitude information of the human body, that is, the three-dimensional space attitude angle of human motion; after that, the eigenvalues of some feature points and the three-dimensional space attitude angle can be extracted for weighted calculations, and the fusion classification algorithm is used for training and compared with the actual movement of the human body. At the same time, according to the change relationship of the feature vectors of the extracted feature points, the state transition diagram can be established Model. After that, by setting the weight ratio of the human body pose pre-classification model and the state transition diagram model, the human body pose recognition model can be better obtained.

In this embodiment, the weight ratios adopted by the first weighted calculation unit and the second weighted calculation unit are obtained after certain comparative analysis and mathematical processing based on the model predicted value and the actual measured value.

After the above-mentioned model is established, the data collected by the human body posture acquisition unit can be matched with the established model, so that the motion posture of the human body can be obtained better. Therefore, the automatic and continuous recognition of human motion can be realized, and the human behavior pattern can be inverted according to the recognition result.

In this embodiment, the fusion classification algorithm used at the fusion classification unit is a combination of KNN (k-Nearest Neighbor) algorithm and SVM (Support Vector Machine) algorithm. The idea of the KNN algorithm is: if most of the k most similar (that is, the nearest neighbors in the feature space) samples of a sample in the feature space belong to a certain category, then the sample also belongs to this category, and SVM is through a The nonlinear mapping p maps the sample space to a high-dimensional or even infinite-dimensional feature space, so that the nonlinearly separable problem in the original sample space is transformed into a linearly separable problem in the feature space. By combining the two, it is possible to effectively identify the steady-state and transition-state actions of the human body, and at the same time maintain the diversity of feature vectors, so that each iteration will retain the feature quantities between samples and update the judgment criteria. After multiple iterations, the feature vector with the best matching degree will be generated, and the stability of the fusion algorithm is strong.

In this embodiment, the establishment of the steady-state model is to first extract the above-mentioned feature vectors (including the extracted multiple feature values and the three-dimensional posture information of the human body calculated by the solution) in each stable motion state, and use the fusion classification algorithm to train the feature vectors. Combined, the eigenvectors matched by each stable motion state are corrected, and finally the relationship model between the eigenvector of the measurement object and the stable motion state is obtained.

In this embodiment, the transition state model is to establish the nonlinear real-time relationship between the above-mentioned eigenvectors (including a plurality of extracted eigenvalues and the calculated three-dimensional posture information of the human body) and the transition state, and solve the delay in the relational model through large sample iterations. parameters and unknown coefficients.

In this embodiment, the stable motion state refers to repeated, continuous, and identical activities within a relatively period of time, such as standing still and running continuously; Sit up and wait.

In this embodiment, the transition relationship between the current state and the front and rear states during the movement of the human body can be better represented by establishing a state transition diagram.

In this embodiment, the current three-dimensional attitude information of the human body, that is, the three-dimensional space attitude angle, can be preferably processed by the attitude calculation unit, and the three-axis acceleration sensor, the three-axis geomagnetic sensor and the three-axis geomagnetic sensor can be preferably extracted by the data preprocessing unit. Multiple frequency-domain, time-domain, and time-frequency feature points in the signal collected by the axial gyroscope sensor, by weighting one or more of the multiple feature points and the three-dimensional attitude information, the recognition can be better improved the accuracy of the results.

In this embodiment, the three-axis acceleration sensor, the three-axis geomagnetic sensor and the three-axis gyroscope sensor can detect the motion state of the human body in real time, and can generate acceleration change waveforms, magnetic force change waveforms and gyroscope waveforms, and the data preprocessing unit It can preprocess the acceleration change waveform, magnetic force change waveform and gyroscope waveform and extract relevant feature points, which is convenient for subsequent data processing.

In conjunction with FIG. 3 , the attitude calculation unit in this embodiment includes a complementary filtering unit and a quaternion arithmetic unit, and the complementary filtering unit includes a median filtering unit for performing median filtering on the data collected by the triaxial acceleration sensor, The self-calibration unit used to calibrate the data collected by the three-axis geomagnetic sensor, the mean filter unit used to perform mean filter processing on the data collected by the three-axis gyroscope, and the mean value filter unit used to calibrate the median filter unit and the self-calibration unit A normalization processing unit for normalizing the processed data, and a data fusion unit for performing data fusion processing on the data processed by the normalization processing unit and the mean filtering unit to obtain a quaternion; quaternion algorithm The unit is used to process the quaternion obtained by the complementary filtering unit to obtain the three-dimensional posture information of the human body.

In this embodiment, after the three-axis acceleration sensor, the three-axis geomagnetic sensor, and the three-axis gyroscope sensor are calibrated, the three-axis acceleration sensor can use the median filter unit to perform median filtering to effectively filter out the three-axis acceleration sensor. The pulse error in the collected signal is then normalized with the signal collected by the self-calibrated three-axis geomagnetic sensor, and data fusion is performed with the three-axis gyroscope signal after the mean filtering process, which can effectively improve the The dynamic performance and static precision of the collected signal better guarantee the real-time and precision of the collected data.

In this embodiment, the feature point extraction of the signal waveform collected by the three-axis acceleration sensor can include such as acceleration mean value, variance, zero-crossing rate, mean square error, etc., and the feature point extraction of the signal waveform collected by the three-axis geomagnetic sensor can include such as Angle skewness, kurtosis, etc., the feature point extraction of the signal waveform collected by the three-axis gyroscope sensor can include, for example, the acceleration DC component after Fourier transform, power spectral density, angular velocity amplitude, frequency, DC component, etc.

In this embodiment, an output unit is connected to the control unit, and the output unit is used to output the currently detected blood pressure value. Data output can thereby be facilitated.

In this embodiment, the control unit is connected to a cloud server, and the human body posture recognition model and feature value table are stored in the cloud server. Therefore, the storage and processing of data can be better facilitated.

As shown in Figure 4, based on the self-calibration system for blood pressure measurement of this embodiment, this embodiment also provides a self-calibration method for blood pressure measurement, which includes the following steps:

Step 1, collect the human blood pressure detection signal through a blood pressure acquisition unit, and establish a blood pressure detection model between the human blood pressure detection signal and the blood pressure value according to the measured blood pressure value;

Step 2, collecting human motion signals through a human body posture acquisition unit, and establishing a human body posture recognition model between the human body motion signals and the human body behavior patterns according to the actual human behavior patterns;

Step 3, collect the corresponding coefficient values of the blood pressure detection model when the human body is in different behavior modes, so as to establish the characteristic value table;

Step 4: After the blood pressure detection model, the human body posture recognition model and the feature value table are established, when the human blood pressure is detected, the human blood pressure detection signal is collected through the blood pressure acquisition unit and the human body motion signal is collected through the human body posture acquisition unit, and then A control unit identifies the current behavior pattern of the human body according to the human body motion signal, and then updates the corresponding coefficient to the blood pressure detection model through the feature value table, and then obtains the human blood pressure measurement value according to the human blood pressure detection signal and the blood pressure detection model after the updated coefficient .

Through the method of this embodiment, the behavior pattern of the human body during blood pressure detection can be detected synchronously, so that different coefficients can be brought into the blood pressure detection model according to different behavior patterns of the human body, thereby effectively reducing the measurement environment and The difference between modeling environments realizes automatic calibration in blood pressure measurement, and thus better realizes continuous measurement of blood pressure.

In this embodiment, the blood pressure collection unit collects the volumetric pulse wave of the human body through the photoelectric sensor as the detection signal of the blood pressure of the human body. The entire content of the Chinese invention patent with the notification number CN107157461A is introduced here. In this embodiment, the same method as that of the patent application is adopted to realize the continuous detection of blood pressure. Therefore, the difficulty of signal acquisition is greatly reduced, the measurement comfort is better improved, and the prediction model results have high accuracy, which can well realize continuous blood pressure monitoring.

In this embodiment, the human body posture acquisition unit collects the acceleration data generated during human body movement, the magnetic field data generated during human body movement, and the The generated angular velocity data is used as a human motion signal.

In conjunction with FIG. 5 , it is a schematic flowchart of the establishment of the human body posture recognition model and the recognition method in this embodiment.

In this embodiment, the method for recognizing the human body posture includes the following steps:

Step 1, collecting human body movement information through the human body posture acquisition unit;

Step 2, processing the data collected by the human body posture acquisition unit through the data processing module, and matching the processed result with the human body posture recognition model, so as to obtain the prediction result of the human body movement posture;

In this step, a human body posture recognition model is first established, and after the human body posture recognition model is established, the information collected by the human body posture acquisition unit can be matched with the human body posture recognition model to obtain the current behavior pattern of the human body;

When establishing the human body posture recognition model and matching the human body motion information, a data preprocessing unit is used to preprocess the signal collected by the human body posture acquisition unit, and multiple feature points are extracted; a state transition diagram is used to establish the unit Establish or match the state transition graph model according to the plurality of feature points; perform attitude calculation on the data processed by the data preprocessing unit through a posture calculation unit, so as to obtain human body three-dimensional posture information; through a first weighted calculation unit One or more of the plurality of feature points and the three-dimensional posture information of the human body are weighted and calculated, and a fusion classification unit is used to establish or match the human body posture pre-classification model according to the weighted calculation results; the state transfer is performed by a second weighted calculation unit The matching results of the graph model and the human pose pre-classification model are weighted and calculated, and then the human pose recognition model is established or matched.

In this embodiment, in step 1, a three-axis acceleration sensor is used to collect acceleration data generated during human movement, a three-axis geomagnetic sensor is used to collect magnetic field data generated during human movement, and a three-axis gyro sensor is used to collect data during human movement. Generated angular velocity data and angle data.

In this embodiment, in step 2, a data preprocessing unit is used to preprocess the signals collected by the three-axis acceleration sensor, the three-axis geomagnetic sensor and the three-axis gyroscope sensor, and extract a plurality of feature points and send them to a processing unit. unit; use a posture calculation unit to perform posture calculation on the data processed by the data preprocessing unit, so as to obtain the three-dimensional posture information of the human body and send it to the processing unit; through the processing unit, one or more of the plurality of feature points A weighted process is performed with the 3D pose information and matched with the human pose recognition model to obtain the prediction result of the human behavior pattern.

In this embodiment, when the attitude calculation unit performs attitude calculation on the data processed by the data preprocessing unit, a complementary filter unit is used to process the data of the three-axis acceleration sensor, the three-axis geomagnetic sensor and the three-axis gyroscope sensor The quaternion is obtained, and a quaternion algorithm unit is used to process the quaternion obtained by the complementary filtering unit to obtain the three-dimensional posture information of the human body.

In this embodiment, when the complementary filter unit is used to process the relevant data, a median filter unit is used to perform median filter processing on the data collected by the three-axis acceleration sensor, and a self-calibration unit is used to perform median filter processing on the data collected by the three-axis geomagnetic sensor. The data is calibrated, using a mean value filter unit to perform mean value filter processing on the data collected by the three-axis gyroscope, and using a normalization processing unit to perform normalization processing on the data processed by the median value filter unit and the self-calibration unit, A data fusion unit is used to perform data fusion processing on the data processed by the normalization processing unit and the mean filtering unit to obtain the quaternion.

Through the human body motion posture recognition method in this embodiment, when the human body is moving, the three-axis acceleration sensor can collect the acceleration data of the human body motion, the three-axis geomagnetic sensor can collect the magnetic field data during the human body motion, and the three-axis gyroscope sensor can Collect the angular velocity and angle data of human body movement; after that, it can preprocess the waveforms of the signals collected by the three-axis acceleration sensor, three-axis geomagnetic sensor and three-axis gyroscope sensor and extract multiple frequency domain, time domain and time-frequency features point; after that, the data collected by the three-axis acceleration sensor can be processed by median filtering, the data collected by the three-axis geomagnetic sensor can be self-calibrated, and the data collected by the three-axis gyroscope sensor can be used for median filtering; after that, the The data collected by the processed three-axis acceleration sensor and the three-axis geomagnetic sensor are normalized and fused with the data collected by the processed three-axis gyroscope sensor, so that the quaternion can be obtained; after that, it can be used The quaternion algorithm performs quaternion calculation on the obtained quaternion, so that the three-dimensional posture information of the human body can be obtained, that is, the three-dimensional space posture angle of the human body movement; One or more of the eigenvalues are weighted and matched with the established human body posture recognition model, so that the current motion posture of the human body can be better recognized.

In this embodiment, the human body gesture recognition model is established based on the relevant parameters collected when the human body moves and the actual movement posture of the human body. By matching the collected and processed data with the human body posture recognition model, the current motion posture of the human body can be better recognized.

In this embodiment, the collection of human body motion data can be realized through a three-axis acceleration sensor, a three-axis geomagnetic sensor, and a three-axis gyroscope sensor, and then the collected data can be preprocessed and feature quantities can be extracted through the data preprocessing unit , and then the data processed by the data preprocessing unit can be used for attitude calculation by the attitude calculation unit. Among them, when building the model, one or more of the extracted feature quantities can be weighted with the result of the posture calculation and then the actual posture of the human body can be used to establish a human body posture pre-classification model. A state transition diagram model is established for actual posture changes, and a human body posture recognition model can be established according to the state transition diagram model and the set weight value of the human body posture pre-classification model. Among them, when recognizing gestures, one or more of the extracted feature quantities can be weighted according to the results of the gesture calculation and then matched with the human body posture pre-classification model, and the extracted feature quantities and the state transition diagram model can be used. Matching, and can carry out weight calculation according to the matching result of the state transition graph model and the matching result of the human body posture pre-classification model, and then obtain the current human body posture by matching the human body posture recognition model.

Through the method of this embodiment, the motion recognition of the steady state and transition state of the human body can be realized.

In this embodiment, the control unit outputs the currently detected blood pressure value through an output unit. Data output can thereby be facilitated.

In this embodiment, the control unit performs data interaction with a cloud server, and stores the human body posture recognition model and feature value table in the cloud server. Therefore, the storage and processing of data can be better facilitated.

The above schematically describes the present invention and its implementation, which is not restrictive, and what is shown in the drawings is only one of the implementations of the present invention, and the actual structure is not limited thereto. Therefore, if a person of ordinary skill in the art is inspired by it, without departing from the inventive concept of the present invention, without creatively designing a structural mode and embodiment similar to the technical solution, it shall all belong to the protection scope of the present invention .

Claims (10)

1. The self-calibration system for blood pressure measurement is characterized in that: it comprises a blood pressure collection unit, a human body posture collection unit and a processing unit, the blood pressure collection unit is used to collect human body blood pressure detection signals, and the human body posture collection unit is used to collect human body movement signals, The processing unit is equipped with a blood pressure detection model, a human body posture recognition model, and an eigenvalue table that records the coefficients of the blood pressure detection model in different human behavior modes; the processing unit is used to recognize the current human body behavior mode through the human body posture recognition model according to the human motion signal , and bring the corresponding coefficients into the blood pressure detection model according to the eigenvalue table, and then obtain the human blood pressure detection value through the blood pressure detection model according to the human blood pressure detection signal. 2. The self-calibration system for blood pressure measurement according to claim 1, characterized in that: the blood pressure acquisition unit includes a photoelectric sensor for acquiring volumetric pulse waves of the human body. 3. The self-calibration system for blood pressure measurement according to claim 1, characterized in that: the human body posture acquisition unit includes a three-axis acceleration sensor, a three-axis geomagnetic sensor and a three-axis gyroscope sensor, and the three-axis acceleration sensor is used to collect The acceleration data generated during the human body movement, the three-axis geomagnetic sensor is used to collect the magnetic field data generated during the human body movement, and the three-axis gyro sensor is used to collect the angular velocity data generated during the human body movement. 4. The self-calibration system for blood pressure measurement according to claim 1, wherein an output unit is connected to the control unit, and the output unit is used to output the currently detected blood pressure value. 5. The self-calibration system for blood pressure measurement according to claim 1, characterized in that: the control unit is connected to a cloud server, and the human body posture recognition model and feature value table are stored in the cloud server. 6. A self-calibration method for blood pressure measurement comprising the steps of: Step 1, collect the human blood pressure detection signal through a blood pressure acquisition unit, and establish a blood pressure detection model between the human blood pressure detection signal and the blood pressure value according to the measured blood pressure value; Step 2, collecting human motion signals through a human body posture acquisition unit, and establishing a human body posture recognition model between the human body motion signals and the human body behavior patterns according to the actual human behavior patterns; Step 3, collect the corresponding coefficient values of the blood pressure detection model when the human body is in different behavior modes, so as to establish the characteristic value table; Step 4: After the blood pressure detection model, the human body posture recognition model and the feature value table are established, when the human blood pressure is detected, the human blood pressure detection signal is collected through the blood pressure acquisition unit and the human body motion signal is collected through the human body posture acquisition unit, and then A control unit identifies the current behavior pattern of the human body according to the human body motion signal, and then updates the corresponding coefficient to the blood pressure detection model through the feature value table, and then obtains the human blood pressure measurement value according to the human blood pressure detection signal and the blood pressure detection model after the updated coefficient . 7. The self-calibration method for blood pressure measurement according to claim 6, characterized in that: the blood pressure acquisition unit collects the body volume pulse wave through the photoelectric sensor as the human body blood pressure detection signal. 8. The self-calibration method for blood pressure measurement according to claim 6, characterized in that: the human body posture acquisition unit collects the body movements generated during the movement of the human body through a three-axis acceleration sensor, a three-axis geomagnetic sensor and a three-axis gyroscope sensor. Acceleration data, magnetic field data generated during human motion, and angular velocity data generated during human motion are used as human motion signals. 9. The self-calibration method for blood pressure measurement according to claim 6, wherein the control unit outputs the currently detected blood pressure value through an output unit. 10. The self-calibration method for blood pressure measurement according to claim 6, characterized in that: the control unit performs data interaction with a cloud server, and stores the human body posture recognition model and feature value table in the cloud server.