CN116184470A - Patrol car fusion positioning method and device based on data driving and relevant medium thereof - Google Patents
Patrol car fusion positioning method and device based on data driving and relevant medium thereof Download PDFInfo
- Publication number
- CN116184470A CN116184470A CN202310205349.8A CN202310205349A CN116184470A CN 116184470 A CN116184470 A CN 116184470A CN 202310205349 A CN202310205349 A CN 202310205349A CN 116184470 A CN116184470 A CN 116184470A
- Authority
- CN
- China
- Prior art keywords
- data
- positioning
- neural network
- patrol car
- network model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000004927 fusion Effects 0.000 title claims abstract description 27
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 66
- 238000012549 training Methods 0.000 claims abstract description 33
- 230000002159 abnormal effect Effects 0.000 claims abstract description 25
- 238000007781 pre-processing Methods 0.000 claims abstract description 19
- 230000006870 function Effects 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 10
- 238000005259 measurement Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 8
- 230000001186 cumulative effect Effects 0.000 claims description 6
- 238000006073 displacement reaction Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000003062 neural network model Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 description 7
- 230000009471 action Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 239000011800 void material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
- G01S19/38—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system
- G01S19/39—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system the satellite radio beacon positioning system transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
- G01S19/42—Determining position
- G01S19/45—Determining position by combining measurements of signals from the satellite radio beacon positioning system with a supplementary measurement
- G01S19/47—Determining position by combining measurements of signals from the satellite radio beacon positioning system with a supplementary measurement the supplementary measurement being an inertial measurement, e.g. tightly coupled inertial
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computer Networks & Wireless Communication (AREA)
- Control Of Position Or Direction (AREA)
Abstract
The invention discloses a patrol car fusion positioning method and device based on data driving and a relevant medium thereof, wherein the method comprises the following steps: respectively carrying out data preprocessing on the first positioning data and the first pose data to obtain second positioning data and second pose data for training; taking the second positioning data as a target vector, taking the second pose data as an input vector, and training the hollow convolutional neural network model to obtain a trained hollow convolutional neural network model; judging whether the positioning signals of the positioning units on the patrol car are abnormal or not, and performing positioning prediction by using the trained cavity convolutional neural network model according to the judging result to obtain a positioning prediction result. According to the invention, the trained cavity convolutional neural network model is utilized to predict the positioning data when the positioning signal of the patrol car is abnormal, so that the positioning accuracy and robustness of the patrol car are greatly improved.
Description
Technical Field
The invention relates to the technical field of positioning of computing intelligent equipment, in particular to a patrol car fusion positioning method and device based on data driving and a related medium thereof.
Background
At present, the intelligent patrol car becomes new equipment for digital management of urban property, and the digital management level of the city is greatly improved. However, as the vehicle-mounted GPS module of the intelligent patrol car has no direct open interface, and GPS data can not meet the positioning real-time requirement when being called from a cloud platform; thus, the intelligent patrol car currently positioning reception is realized by a GPS module in an intelligent reasoning box mounted on the intelligent patrol car.
And the GPS module in the intelligent reasoning box is easy to be blocked by tree yin, bridge hole, tunnel, high building and the like when the urban street is driven, so that the GPS positioning accuracy can be greatly reduced and even the situation that the GPS module cannot be used can be caused. In addition, the positioning receiver has hot start and cold start time of tens of seconds or even longer when being started, so that the positioning data of the intelligent patrol car is lost within a period of time after the intelligent patrol car is started; the lack of positioning data will cause the patrol line of the intelligent patrol car to break down or other more serious faults.
Disclosure of Invention
The embodiment of the invention provides a patrol car fusion positioning method and device based on data driving and a related medium thereof, aiming at solving the problem that the intelligent patrol car in the prior art is lack of positioning data due to shielding by objects and overlong starting time.
In a first aspect, an embodiment of the present invention provides a patrol car fusion positioning method based on data driving, including:
respectively carrying out data preprocessing on the first positioning data and the first pose data to obtain second positioning data and second pose data for training; the first positioning data and the first pose data are obtained through the positioning unit and an inertia measuring unit on the patrol car respectively when positioning signals of the positioning unit on the patrol car are normal;
taking the second positioning data as a target vector, taking the second pose data as an input vector, and training the hollow convolutional neural network model to obtain a trained hollow convolutional neural network model;
judging whether the positioning signals of the positioning units on the patrol car are abnormal or not, if not, continuing to train the cavity convolutional neural network model; if yes, acquiring pose data when the positioning signal is abnormal through the inertial measurement unit, and preprocessing the data to obtain target pose data for prediction;
and inputting the target pose data into the trained cavity convolutional neural network model for positioning prediction to obtain a positioning prediction result.
In a second aspect, an embodiment of the present invention provides a patrol car fusion positioning device based on data driving, including:
the data processing unit is used for respectively carrying out data preprocessing on the first positioning data and the first pose data to obtain second positioning data and second pose data for training; the first positioning data and the first pose data are obtained through the positioning unit and an inertia measuring unit on the patrol car respectively when positioning signals of the positioning unit on the patrol car are normal;
the model training unit is used for taking the second positioning data as a target vector, taking the second pose data as an input vector, and training the hollow convolutional neural network model to obtain a trained hollow convolutional neural network model;
the data judging unit is used for judging whether the positioning signals of the positioning units on the patrol car are abnormal or not, and if not, the training of the cavity convolutional neural network model is continuously maintained; if yes, acquiring pose data when the positioning signal is abnormal through the inertial measurement unit, and preprocessing the data to obtain target pose data for prediction;
and the data output unit is used for inputting the target pose data into the trained cavity convolutional neural network model to perform positioning prediction so as to obtain a positioning prediction result.
In a third aspect, an embodiment of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, where the processor implements the data-driven patrol car fusion positioning method according to the first aspect when executing the computer program.
In a fourth aspect, an embodiment of the present invention provides a computer readable storage medium, where the computer readable storage medium stores a computer program, where the computer program when executed by a processor implements the data-driven patrol car fusion positioning method of the first aspect.
The embodiment of the invention provides a patrol car fusion positioning method based on data driving, which comprises the following steps: respectively carrying out data preprocessing on the first positioning data and the first pose data to obtain second positioning data and second pose data for training; taking the second positioning data as a target vector, taking the second pose data as an input vector, and training the hollow convolutional neural network model to obtain a trained hollow convolutional neural network model; judging whether the positioning signals of the positioning units on the patrol car are abnormal or not, and performing positioning prediction by using the trained cavity convolutional neural network model according to the judging result to obtain a positioning prediction result. According to the invention, the trained cavity convolutional neural network model is utilized to predict the positioning data when the positioning signal of the patrol car is abnormal, so that the positioning accuracy and robustness of the patrol car are greatly improved.
The embodiment of the invention also provides a patrol car fusion positioning device based on data driving, computer equipment and a storage medium, and the patrol car fusion positioning device has the same beneficial effects.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, 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 flow chart of a patrol car fusion positioning method based on data driving according to an embodiment of the present invention;
fig. 2 is another flow chart of a patrol car fusion positioning method based on data driving according to an embodiment of the present invention;
FIG. 3 is a diagram of a positioning system according to an embodiment of the present invention;
FIG. 4 is a flowchart of a model of a convolutional neural network of a cavity provided in an embodiment of the present invention;
fig. 5 is a schematic block diagram of a patrol car fusion positioning device based on data driving according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be understood that the terms "comprises" and "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
It should be further understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.
Referring to fig. 1, fig. 1 is a schematic flow chart of a patrol car fusion positioning method based on data driving according to an embodiment of the present invention, which specifically includes: steps S101 to S104.
S101, respectively carrying out data preprocessing on first positioning data and first pose data to obtain second positioning data and second pose data for training; the first positioning data and the first pose data are obtained through the positioning unit and an inertia measuring unit on the patrol car respectively when positioning signals of the positioning unit on the patrol car are normal;
s102, taking the second positioning data as a target vector, taking the second pose data as an input vector, and training the hollow convolutional neural network model to obtain a trained hollow convolutional neural network model;
s103, judging whether positioning signals of a positioning unit on the patrol car are abnormal, if not, continuing to train the cavity convolutional neural network model; if yes, acquiring pose data when the positioning signal is abnormal through the inertial measurement unit, and preprocessing the data to obtain target pose data for prediction;
s104, inputting the target pose data into the trained cavity convolutional neural network model for positioning prediction, and obtaining a positioning prediction result.
In step S101, when a positioning signal of a positioning unit on the patrol car is normal, the positioning unit (GPS, the same applies hereinafter) acquires the first positioning data, and the inertial measurement unit (IMU, the same applies hereinafter) acquires the first pose data; the first positioning data and the first pose data can be used for training the cavity convolutional neural network model only after data preprocessing is carried out on the first positioning data and the first pose data; finally, the first positioning data and the first pose data are subjected to data preprocessing to obtain the second positioning data and the second pose data. It should be noted that the first positioning data and the first pose data are historical data of the patrol car.
As shown in fig. 2 and 3, in an embodiment, the step S101 includes:
performing enhancement processing on the first positioning data by using a Neville interpolation method to obtain the second positioning data; and filtering the first pose data by utilizing a sliding weighting method to obtain the second pose data.
Further, a register with a memory space of N is selected; when detecting that new first pose data is input, placing the new first pose data into the last position of the register, and sequentially moving the rest data in the register forwards by one position; all data lambda in the register are written as follows i Weight ω corresponding thereto i After multiplication, the weighted sum is obtained:
the filtered output is calculated according to the following formula and used as second pose data
In this embodiment, the positioning accuracy of the GPS is lower because the frequency of receiving data is lower than that of the IMU; therefore, it is necessary to improve the positioning accuracy of the GPS by data enhancement processing, where the first positioning data is interpolated and supplemented by the Neville interpolation method. Firstly, determining the number n of inserted data points between adjacent data points acquired by the GPS according to the following formula:
wherein, (x) 0 ,y 0 ) And (x) 1 ,y 1 ) Is the coordinate point of the adjacent data points acquired by the GPS.
The value (x) can be obtained by Neville interpolation 0 ,y 0 ) And (x) 1 ,y 1 ) A point in between:
finally, n interpolation points (namely the second positioning data) are obtained through iteration according to the following formula:
further, in the running process of the patrol car, a sliding weighting method is adopted to carry out filtering processing on the first pose data. Firstly, selecting a register with a memory space of N, when new first pose data is input, placing the new first pose data into the last bit of the register, sequentially moving the rest data in the register forwards by one bit, and then inputting all data lambda in the register i Weight ω corresponding thereto i And after multiplication, obtaining weighted sum to obtain a filtered output result, and then calculating the filtered output result to obtain the second pose data.
Specifically, when new first pose data is input, the register data sequentially slides forward by one bit, and the new first pose data enters the last bit of the register. In the running process of the patrol car, the reliability of early sampled data is low, so that the pose signal of the IMU is processed by adopting sliding weighted filtering; each time when a new first pose data is input, discarding the earliest first pose data, and sequentially assigning the rest data to the previous data, wherein the previous data has smaller weight and the new data has larger weight; and finally, calculating the weighted average value of N new data in the buffer memory area, and outputting the weighted average value as data of the node.
In step S102, after the target vector and the input vector are obtained, the hole convolutional neural network model may be trained, so as to obtain the trained hole convolutional neural network model; and when the patrol car positioning signal is abnormal, performing positioning prediction by using the trained cavity convolutional neural network model.
In one embodiment, the step S102 includes:
inputting the input vector into the cavity convolutional neural network model to obtain an output value; substituting the output value and the target vector into a loss function, and calculating to obtain an error term; judging whether the error term is in an allowable range, if not, updating a weight matrix, and repeating the steps until the error term reaches the allowable range; if yes, outputting the trained cavity convolutional neural network model.
In this embodiment, the input vector is the second pose data in the x and y directions, and the target vector is the relative displacement between two adjacent points of the second positioning data on the x and y axes. In the training process, the input vector is input into the cavity convolutional neural network model, and the output value is obtained; substituting the output value and the target vector into a loss function to obtain an error term of the output value and the target vector; if the error term is not in the allowable range, updating the weight matrix; and finally, repeating the steps until the error term reaches the allowable range, and outputting the trained cavity convolutional neural network model. It should be noted that, when the patrol car positioning signal is normal, the training data (i.e., the second pose data and the second positioning data) are updated online, and the training data of the cavity convolutional neural network model is updated in real time as the data is updated during the patrol car traveling process.
Referring to fig. 4, in an embodiment, a batch of normalization layers is inserted before the convolutional layer of the hole convolutional neural network model, and the expression of the batch of normalization layers is as follows:
wherein y is i An ith output representing the batch normalization layer; x is x i Representing an ith input; μ represents the average value of the batch normalization layer inputs; o' represents variance; gamma represents a scaling factor; beta represents an offset coefficient.
In the embodiment, the problem existing in the image segmentation field is solved by the initial proposal of the cavity convolution, and the cavity is injected into the convolution layer of the traditional convolution neural network, so that the receptive field can be improved on the premise of keeping the parameters of the convolution layer unchanged, and the operation amount is reduced; the cavity convolution rate is different, the receptive field of convolution operation is different, and in order to eliminate error accumulation of singular samples, the batch standardization layer is inserted before the convolution layer of the cavity convolution neural network model, so that training speed and prediction accuracy are effectively improved.
In an embodiment, a gaussian error linear unit activation function is inserted after the convolution layer of the cavity convolutional neural network model, and the expression of the gaussian error linear unit activation function is as follows:
GELU(x i )＝x i Pr[X≤x i ]＝x i Φ(x i )
wherein Pr [ X ] is less than or equal to X i ]Indicating that the random variable X is less than X i Is a cumulative probability of (1); phi (x) i ) The cumulative function representing the standard probability density corresponds to x i An input value; x is x i Representing the ith output of the convolutional layer.
In this embodiment, the gaussian error linear unit activation function is inserted after the convolutional layer of the cavity convolutional neural network model, and the nonlinear variation of the gaussian error linear unit activation function is a random regularized transformation mode which accords with expectations, and is a high-performance neural network activation function, so that the model learning capability can be effectively improved.
Further, five convolution blocks are connected in series after the Gaussian error linear unit activation function module of the cavity convolution neural network model, and the cavity convolution rate calculation formula of each convolution layer is as follows:
wherein j represents the number of layers of the convolution layer; d, d j The void convolution rate of the j-th layer is represented;
it should be noted that the number of convolution blocks may be increased or decreased, which is not limited herein; and the convolution block can be connected with a flat layer and a linear layer, and finally the cavity convolution neural network model is output.
In step S103, determining whether a positioning signal of a positioning unit on the patrol car is abnormal, and training the cavity convolutional neural network model by using positioning data of a GPS and pose data of an IMU when the patrol car positioning signal is normal; when the patrol car positioning signal is abnormal, the positioning of the patrol car is predicted through the trained cavity convolutional neural network model and the target pose data.
In step S104, the target pose data is input to the trained hole convolutional neural network model to perform positioning prediction, a positioning prediction result is obtained, and the positioning prediction result is uploaded to a cloud platform.
In an embodiment, the step S104 is configured to:
the positioning prediction result is obtained through calculation according to the following formula:
x＝GPS x0 +∑x i
y＝GPS y0 +∑y i
wherein x and y respectively represent an abscissa and an ordinate of the positioning prediction result; GPS (Global positioning System) x0 、GPS y0 Representing the foremost position of the patrol car positioning signal before abnormalityThe latter constant; x is x i 、y i Representing the relative displacement of the abscissa and the ordinate, respectively.
In this embodiment, based on the trained model of the hollow convolutional neural network, the relative displacement x in the x-axis and y-axis directions is calculated i 、y i Predicting x from abnormal positioning signals i And y i Accumulating and adding the accumulated value Sigma x i 、∑y i And adding the result with the last fixed value before the abnormality of the positioning signal of the patrol car, wherein the obtained result is the positioning prediction result.
In summary, the method uses the cavity convolutional neural network model to integrate the two positioning modes of the positioning unit and the inertial measurement unit, and when the patrol car positioning signal is normal, the cavity convolutional neural network model is trained by using the second position data and the second pose data; and when the patrol car positioning signal is abnormal, predicting the data of the patrol car through the trained cavity convolutional neural network model and the second pose data.
Referring to fig. 5, fig. 5 is a schematic block diagram of a data-driven patrol car fusion positioning device according to an embodiment of the present invention, where a data-driven patrol car fusion positioning device 500 includes:
a data processing unit 501, configured to perform data preprocessing on the first positioning data and the first pose data, to obtain second positioning data and second pose data for training; the first positioning data and the first pose data are obtained through the positioning unit and an inertia measuring unit on the patrol car respectively when positioning signals of the positioning unit on the patrol car are normal;
the model training unit 502 is configured to use the second positioning data as a target vector, and the second pose data as an input vector, and train the hole convolutional neural network model to obtain a trained hole convolutional neural network model;
a data judging unit 503, configured to judge whether a positioning signal of a positioning unit on the patrol car is abnormal, and if not, continue to keep training the cavity convolutional neural network model; if yes, acquiring pose data when the positioning signal is abnormal through the inertial measurement unit, and preprocessing the data to obtain target pose data for prediction;
and the data output unit 504 is used for inputting the target pose data into the trained cavity convolutional neural network model for positioning prediction to obtain a positioning prediction result.
In this embodiment, first, the data processing unit 501 performs data preprocessing on the first positioning data and the first pose data, to obtain second positioning data and second pose data for training; the first positioning data and the first pose data are obtained through the positioning unit and an inertia measuring unit on the patrol car respectively when positioning signals of the positioning unit on the patrol car are normal; secondly, the model training unit 502 takes the second positioning data as a target vector, takes the second pose data as an input vector, and trains the hollow convolutional neural network model to obtain a trained hollow convolutional neural network model; then, the data judging unit 503 judges whether the positioning signal of the positioning unit on the patrol car is abnormal, if not, the training of the cavity convolutional neural network model is continuously maintained; if yes, acquiring pose data when the positioning signal is abnormal through the inertial measurement unit, and preprocessing the data to obtain target pose data for prediction; finally, the data output unit 504 inputs the target pose data to the trained hole convolutional neural network model for positioning prediction, so as to obtain a positioning prediction result.
In an embodiment, the data processing unit comprises:
the enhancement unit is used for enhancing the first positioning data by using a Neville interpolation method to obtain the second positioning data;
and the filtering unit is used for filtering the first pose data by utilizing a sliding weighting method to obtain the second pose data.
In an embodiment, the filtering unit includes:
a selecting unit, configured to select a register with a memory space N;
the displacement unit is used for placing the new first pose data into the last position of the register when detecting that the new first pose data is input, and sequentially moving the rest data in the register forwards by one position;
a multiplication unit for multiplying all data lambda in the register according to the following formula i Weight ω corresponding thereto i After multiplication, the weighted sum is obtained:
a pose unit for calculating the filtered output according to the following formula and using the filtered output as second pose data
In one embodiment, the convolutional layer of the hole convolutional neural network model is preceded by a batch of normalization layers, and the expression of the batch of normalization layers is as follows:
wherein y is i An ith output representing the batch normalization layer; x is x i Representing an ith input; μ represents the average value of the batch normalization layer inputs; o' represents variance; gamma represents a scaling factor; beta represents an offset coefficient.
In an embodiment, a gaussian error linear unit activation function is inserted after the convolution layer of the cavity convolutional neural network model, and the expression of the gaussian error linear unit activation function is as follows:
GELU(x i )＝x i Pr[X≤x i ]＝x i Φ(x i )
wherein Pr [ X ] is less than or equal to X i ]Indicating that the random variable X is less than X i Is a cumulative probability of (1); phi (x) i ) The cumulative function representing the standard probability density corresponds to x i An input value; x is x i Representing the ith output of the convolutional layer.
In an embodiment, the model training unit comprises:
the output unit is used for inputting the input vector into the cavity convolutional neural network model to obtain an output value;
the calculating unit is used for substituting the loss function according to the output value and the target vector, and calculating to obtain an error term;
the judging unit is used for judging whether the error item is in an allowable range, if not, updating the weight matrix, and repeating the steps until the error item reaches the allowable range; if yes, outputting the trained cavity convolutional neural network model.
In an embodiment, the data output unit is configured to:
the positioning prediction result is obtained through calculation according to the following formula:
x＝GPS x0 +∑x i
y＝GPS y0 +∑y i
wherein x and y respectively represent an abscissa and an ordinate of the positioning prediction result; GPS (Global positioning System) x0 、GPS y0 Representing the last constant value before the abnormal positioning signal of the patrol car; x is x i 、y i Representing the relative displacement of the abscissa and the ordinate, respectively.
Since the embodiments of the apparatus portion and the embodiments of the method portion correspond to each other, the embodiments of the apparatus portion are referred to the description of the embodiments of the method portion, and are not repeated herein.
The embodiment of the present invention also provides a computer readable storage medium having a computer program stored thereon, which when executed can implement the steps provided in the above embodiment. The storage medium may include: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The embodiment of the invention also provides a computer device, which can comprise a memory and a processor, wherein the memory stores a computer program, and the processor can realize the steps provided by the embodiment when calling the computer program in the memory. Of course, the computer device may also include various network interfaces, power supplies, and the like.
In the description, each embodiment is described in a progressive manner, and each embodiment is mainly described by the differences from other embodiments, so that the same similar parts among the embodiments are mutually referred. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section. It should be noted that it would be obvious to those skilled in the art that various improvements and modifications can be made to the present application without departing from the principles of the present application, and such improvements and modifications fall within the scope of the claims of the present application.
It should also be noted that in this specification, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
Claims (10)
1. The patrol car fusion positioning method based on data driving is characterized by comprising the following steps of:
respectively carrying out data preprocessing on the first positioning data and the first pose data to obtain second positioning data and second pose data for training; the first positioning data and the first pose data are obtained through the positioning unit and an inertia measuring unit on the patrol car respectively when positioning signals of the positioning unit on the patrol car are normal;
taking the second positioning data as a target vector, taking the second pose data as an input vector, and training the hollow convolutional neural network model to obtain a trained hollow convolutional neural network model;
judging whether the positioning signals of the positioning units on the patrol car are abnormal or not, if not, continuing to train the cavity convolutional neural network model; if yes, acquiring pose data when the positioning signal is abnormal through the inertial measurement unit, and preprocessing the data to obtain target pose data for prediction;
and inputting the target pose data into the trained cavity convolutional neural network model for positioning prediction to obtain a positioning prediction result.
2. The data-driven patrol car fusion positioning method according to claim 1, wherein the performing data preprocessing on the first positioning data and the first pose data to obtain second positioning data and second pose data for training includes:
performing enhancement processing on the first positioning data by using a Neville interpolation method to obtain the second positioning data;
and filtering the first pose data by utilizing a sliding weighting method to obtain the second pose data.
3. The data-driven patrol car fusion positioning method according to claim 2, wherein the filtering the first pose data by using a sliding weighting method to obtain the second pose data comprises:
selecting a register with a memory space of N;
when detecting that new first pose data is input, placing the new first pose data into the last position of the register, and sequentially moving the rest data in the register forwards by one position;
all data lambda in the register are written as follows i Weight ω corresponding thereto i After multiplication, the weighted sum is obtained:
the filtered output is calculated according to the following formula and used as second pose data
4. The data-driven patrol car fusion positioning method according to claim 1, wherein a batch of standardization layer is inserted in front of a convolution layer of the cavity convolution neural network model, and an expression of the batch of standardization layer is as follows:
wherein y is i An ith output representing the batch normalization layer; x is x i Representing an ith input; μ represents the average value of the batch normalization layer inputs; o' represents variance; gamma represents a scaling factor; beta represents an offset coefficient.
5. The data-driven patrol car fusion positioning method according to claim 1, wherein a gaussian error linear unit activation function is inserted after a convolution layer of the cavity convolutional neural network model, and an expression of the gaussian error linear unit activation function is as follows:
GELU(x i )＝x i Pr[X≤x i ]＝x i Φ(x i )
wherein pr [ X is less than or equal to X ] i ]Indicating that the random variable X is less than X i Is a cumulative probability of (1); phi (x) i ) The cumulative function representing the standard probability density corresponds to x i An input value; x is x i Representing the ith output of the convolutional layer.
6. The data-driven patrol car fusion positioning method according to claim 1, wherein the training the hole convolutional neural network model with the second positioning data as a target vector and the second pose data as an input vector to obtain a trained hole convolutional neural network model comprises:
inputting the input vector into the cavity convolutional neural network model to obtain an output value;
substituting the output value and the target vector into a loss function, and calculating to obtain an error term;
judging whether the error term is in an allowable range, if not, updating a weight matrix, and repeating the steps until the error term reaches the allowable range; if yes, outputting the trained cavity convolutional neural network model.
7. The data-driven patrol car fusion positioning method according to claim 1, wherein the step of inputting the target pose data to the trained cavity convolutional neural network model to perform positioning prediction to obtain a positioning prediction result comprises:
the positioning prediction result is obtained through calculation according to the following formula:
x＝GPS x0 +∑x i
y＝GPS y0 +∑y i
wherein x and y respectively represent an abscissa and an ordinate of the positioning prediction result; GPS (Global positioning System) x0 、GPS y0 Representing the last constant value before the abnormal positioning signal of the patrol car; x is x i 、y i Representing the relative displacement of the abscissa and the ordinate, respectively.
8. Patrol car fuses positioner based on data drive, its characterized in that includes:
the data processing unit is used for respectively carrying out data preprocessing on the first positioning data and the first pose data to obtain second positioning data and second pose data for training; the first positioning data and the first pose data are obtained through the positioning unit and an inertia measuring unit on the patrol car respectively when positioning signals of the positioning unit on the patrol car are normal;
the model training unit is used for taking the second positioning data as a target vector, taking the second pose data as an input vector, and training the hollow convolutional neural network model to obtain a trained hollow convolutional neural network model;
the data judging unit is used for judging whether the positioning signals of the positioning units on the patrol car are abnormal or not, and if not, the training of the cavity convolutional neural network model is continuously maintained; if yes, acquiring pose data when the positioning signal is abnormal through the inertial measurement unit, and preprocessing the data to obtain target pose data for prediction;
and the data output unit is used for inputting the target pose data into the trained cavity convolutional neural network model to perform positioning prediction so as to obtain a positioning prediction result.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the data-driven based patrol car fusion positioning method according to any one of claims 1 to 7 when the computer program is executed.
10. A computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and when executed by a processor, the computer program implements the data-driven based patrol car fusion positioning method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310205349.8A CN116184470A (en) | 2023-03-06 | 2023-03-06 | Patrol car fusion positioning method and device based on data driving and relevant medium thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310205349.8A CN116184470A (en) | 2023-03-06 | 2023-03-06 | Patrol car fusion positioning method and device based on data driving and relevant medium thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116184470A true CN116184470A (en) | 2023-05-30 |
Family
ID=86450525
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310205349.8A Pending CN116184470A (en) | 2023-03-06 | 2023-03-06 | Patrol car fusion positioning method and device based on data driving and relevant medium thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116184470A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118706019A (en) * | 2024-08-28 | 2024-09-27 | 铁正检测科技有限公司 | Method and system for automatically monitoring tunnel lining deformation based on laser tracking |
-
2023
- 2023-03-06 CN CN202310205349.8A patent/CN116184470A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118706019A (en) * | 2024-08-28 | 2024-09-27 | 铁正检测科技有限公司 | Method and system for automatically monitoring tunnel lining deformation based on laser tracking |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112289034A (en) | Deep neural network robust traffic prediction method based on multi-mode space-time data | |
| CN108267123A (en) | A kind of double-current vehicle-mounted pedestrian vehicle Forecasting Methodology based on bounding box and range prediction | |
| CN110309537B (en) | Intelligent health prediction method and system for aircraft | |
| CN115100247B (en) | Method for predicting ship track step by step based on AIS dynamic information | |
| CN116342657A (en) | TCN-GRU ship track prediction method, system, equipment and medium based on coding-decoding structure | |
| CN114255616A (en) | Unpowered ship track prediction method, device, equipment and storage medium | |
| CN110889530A (en) | Destination prediction method based on recurrent neural network and server | |
| CN116476863A (en) | Automatic driving transverse and longitudinal integrated decision-making method based on deep reinforcement learning | |
| CN112578419B (en) | GPS data reconstruction method based on GRU network and Kalman filtering | |
| CN116184470A (en) | Patrol car fusion positioning method and device based on data driving and relevant medium thereof | |
| CN116495007A (en) | Vehicle track deep learning prediction method considering physical constraint | |
| CN116306790A (en) | Offshore ship track real-time prediction method, system, equipment and medium based on CNN-GRU and attention mechanism | |
| CN115457764B (en) | Road section traffic density estimation method, device and medium based on vehicle track data | |
| CN116166642A (en) | Spatio-temporal data filling method, system, equipment and medium based on guide information | |
| CN113665576B (en) | Method, device, equipment and medium for predicting running condition of vehicle | |
| CN118013167A (en) | Vehicle track data processing method and device and electronic equipment | |
| CN116861262B (en) | Perception model training method and device, electronic equipment and storage medium | |
| CN108981733B (en) | Speed prediction method of electric vehicle charging navigation system | |
| CN117576148A (en) | Target tracking method, device, electronic equipment and storage medium | |
| CN114399901A (en) | Method and equipment for controlling traffic system | |
| CN118052347B (en) | Travel time estimation method and system based on travel track sequence of floating car | |
| CN116972837B (en) | Self-adaptive vehicle-mounted combined navigation positioning method and related equipment | |
| CN116542528B (en) | Traffic demand satisfaction analysis method and device | |
| CN118664607B (en) | Method and system for predicting motion trail of upper limb rehabilitation robot | |
| CN115062270B (en) | Maximum likelihood-based missing value filling sensor data processing method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |