CN117634025A - Vehicle motion state prediction method, device, equipment and readable storage medium - Google Patents

Vehicle motion state prediction method, device, equipment and readable storage medium Download PDF

Info

Publication number
CN117634025A
CN117634025A CN202311549404.1A CN202311549404A CN117634025A CN 117634025 A CN117634025 A CN 117634025A CN 202311549404 A CN202311549404 A CN 202311549404A CN 117634025 A CN117634025 A CN 117634025A
Authority
CN
China
Prior art keywords
motion state
vehicle motion
prediction
data
correction
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
Application number
CN202311549404.1A
Other languages
Chinese (zh)
Inventor
刘凤阳
冯坦
陈镇
何胜勇
李鹏远
赵紫薇
陈苗苗
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Dongfeng Commercial Vehicle Co Ltd
Original Assignee
Dongfeng Commercial Vehicle Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Dongfeng Commercial Vehicle Co Ltd filed Critical Dongfeng Commercial Vehicle Co Ltd
Priority to CN202311549404.1A priority Critical patent/CN117634025A/en
Publication of CN117634025A publication Critical patent/CN117634025A/en
Pending legal-status Critical Current

Links

Classifications

    • 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

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

A vehicle motion state prediction method, device, equipment and readable storage medium, the vehicle motion state prediction method includes: acquiring vehicle motion state data, and acquiring first prediction data through a circulating gate neural network by using the normalized vehicle motion state data; obtaining correction parameters and weight parameters of the attention selection layer through a driving style classifier; correcting the first prediction data by using the correction parameter and the weight parameter through the attention selection layer to obtain second prediction data; and carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data to obtain third prediction data, and taking the third prediction data as a final vehicle motion state prediction result. According to the vehicle motion state prediction method and device, the vehicle motion state prediction is carried out through the machine learning algorithm, the model construction is simple, the calculated amount is small, the vehicle motion state prediction method and device can be embedded into the ECU of the vehicle to carry out real-time prediction, the cooperation of a sensor and a high-precision map is not needed, and all vehicle operation conditions can be covered.

Description

Vehicle motion state prediction method, device, equipment and readable storage medium
Technical Field
The present disclosure relates to the field of vehicle state prediction technologies, and in particular, to a vehicle motion state prediction method, device, equipment, and readable storage medium.
Background
At present, a method of predicting the running state of a vehicle is often adopted, however, most of the current vehicle dynamics models are commercial software, the vehicle dynamics models are complex to construct and large in calculation amount, cannot be embedded into an ECU (Electronic Control Unit) of the vehicle to perform real-time prediction, and further require the cooperation of a sensor and a high-precision map, so that all the running conditions of the vehicle cannot be covered.
Disclosure of Invention
The application provides a vehicle motion state prediction method, device, equipment and readable storage medium, and aims to solve the technical problems that the running state of a vehicle is predicted at present, most of commonly adopted vehicle dynamics models are commercial software, the vehicle dynamics models are complex in construction and large in calculation amount, cannot be embedded into an ECU (electronic control unit) of the vehicle for real-time prediction, and the sensor and a high-precision map are required to be matched, so that all the running working conditions of the vehicle cannot be covered.
In a first aspect, an embodiment of the present application provides a vehicle motion state prediction method, including:
acquiring vehicle motion state data, and carrying out normalization processing to obtain normalized vehicle motion state data;
obtaining first prediction data through a circulating gate neural network by using the normalized vehicle motion state data;
obtaining correction parameters and weight parameters of the attention selection layer through a driving style classifier by using the normalized vehicle motion state data;
correcting the first prediction data by using the correction parameter and the weight parameter through the attention selection layer to obtain second prediction data;
and carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data to obtain third prediction data, and taking the third prediction data as a final vehicle motion state prediction result.
Optionally, the vehicle motion state data includes a vehicle speed, a brake pedal stroke, an accelerator pedal stroke, an engine speed and an engine power, and the obtaining the vehicle motion state data and performing normalization processing to obtain normalized vehicle motion state data includes:
acquiring the speed, the travel of a brake pedal, the travel of an accelerator pedal, the rotation speed of an engine and the power of the engine of a vehicle;
the normalized vehicle speed, brake pedal travel, accelerator pedal travel, engine speed and engine power are calculated by the formula one:
Y＝(X-Xmin)/(Xmax-Xmin)；
wherein Y represents normalized vehicle motion data, X represents acquired vehicle motion state data, xmax represents maximum value of vehicle motion state data, and Xmin represents minimum value of vehicle motion state data.
Optionally, the loop gate neural network includes an update gate, a reset gate, and a hidden layer, and the obtaining the first prediction data through the loop gate neural network using the normalized vehicle motion state data includes:
and according to the normalized vehicle motion state data and the output of the hidden layer at one moment, calculating to obtain the output of the update door through a formula II, wherein the formula II is as follows:
Z t ＝σ(U z *X t +W z *H t-1 +bz)；
and calculating to obtain the output of the reset gate through a formula III according to the normalized vehicle motion state data and the output of the hidden layer at one moment, wherein the formula III is as follows:
R t ＝σ(U r *X t +W r *H t-1 +br)；
according to the output of the update gate and the output of the reset gate, calculating to obtain first prediction data of the hidden layer through a formula IV, wherein the formula IV is as follows:
H t ＝(1-Z t )*H t-1 +Z t *H’ t ，H’ t ＝tanh(U h *X t +W h *R t +bh)；
wherein Z is t To update the gate output, U z To update the input weight parameters of the gate, X t For normalized vehicle motion state data, W z To conceal the weight parameters of the layer in the update gate, H t-1 For the output of the hidden layer at the time immediately preceding the time t, bz is the correction parameter of the update gate, σ () is the activation function σ (x) =1/(1+e) -x ) E is a natural base, R t To reset the gate output, U r For the input weight parameter of the reset gate, wr is the weight parameter of the hidden layer in the reset gate, br is the correction parameter of the reset gate, H t For the output of the hidden layer at time t, i.e. the first predicted data, H' t For the output of the activated hidden layer at time t, tanh () is the activation function tanh (x) = (e) x -e -x )/(e x +e -x )，U h To input weight parameters of hidden layer, W h To reset the weight parameters of the gate in the hidden layer, bh is the correction parameters of the hidden layer.
Optionally, the correcting, by the attention selection layer, the first prediction data using the correction parameter and the weight parameter to obtain the second prediction data includes:
and correcting the first prediction data through a fifth formula by using the correction parameters and the weight parameters through the attention selection layer to obtain second prediction data, wherein the fifth formula is as follows:
SA＝H t *Wdcc+bdcc；
wherein SA is second predicted data, H t For the first prediction data, wdcc is a weight parameter, and bdcc is a correction parameter.
Optionally, the performing nonlinear function correction and kalman filter fusion correction on the second prediction data to obtain third prediction data includes:
and carrying out nonlinear function correction on the second prediction data based on the feedforward neural network to obtain a correction result of the nonlinear function, wherein the nonlinear function correction formula is as follows:
F t ＝ReLU(SA)；
and carrying out Kalman filtering fusion correction according to a correction result of the nonlinear function and an actual value of the vehicle motion state data to obtain third prediction data, wherein the formula six is as follows:
P t ＝(1-K t )*P t - ；
wherein F is t As a result of the correction of the nonlinear function, reLU () is a nonlinear function ReLU (x) =max (0, x), SA is second prediction data,fusing the corrected predicted value, i.e. the third predicted data, for Kalman filtering>K is a predicted value before Kalman filtering fusion correction t For Kalman gain, Z t For the actual value of the vehicle motion state data at the moment t, W is the weight parameter of the linear link, B is the correction parameter of the linear link, R is the noise of the actual process, Q is the noise of the prediction process, and>for a priori covariance matrix, P t For the posterior covariance matrix at time t, P t-1 The posterior covariance matrix at the time immediately preceding time t.
In a second aspect, embodiments of the present application provide a vehicle motion state prediction apparatus, including:
the acquisition module is used for acquiring the vehicle motion state data and carrying out normalization processing to obtain normalized vehicle motion state data;
the first prediction module is used for obtaining first prediction data through a circulating gate nerve network by using the normalized vehicle motion state data;
the classification module is used for obtaining correction parameters and weight parameters of the attention selection layer through the driving style classifier by using the normalized vehicle motion state data;
the second prediction module is used for correcting the first prediction data by using the correction parameter and the weight parameter through the attention selection layer to obtain second prediction data;
the correction module is used for carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data to obtain third prediction data, and taking the third prediction data as a final vehicle motion state prediction result
Optionally, the second prediction module is configured to:
and correcting the first prediction data through a fifth formula by using the correction parameters and the weight parameters through the attention selection layer to obtain second prediction data, wherein the fifth formula is as follows:
SA＝H t *Wdcc+bdcc；
wherein SA is second predicted data, H t For the first prediction data, wdcc is a weight parameter, and bdcc is a correction parameter.
Optionally, the correction module is configured to:
and carrying out nonlinear function correction on the second prediction data based on the feedforward neural network to obtain a correction result of the nonlinear function, wherein the nonlinear function correction formula is as follows:
F t ＝ReLU(SA)；
and carrying out Kalman filtering fusion correction according to a correction result of the nonlinear function and an actual value of the vehicle motion state data to obtain third prediction data, wherein the formula six is as follows:
P t ＝(1-K t )*P t - ；
wherein F is t As a result of the correction of the nonlinear function, reLU () is a nonlinear function ReLU (x) =max (0, x), SA is second prediction data,fusing the corrected predicted value, i.e. the third predicted data, for Kalman filtering>K is a predicted value before Kalman filtering fusion correction t For Kalman gain, Z t For the actual value of the vehicle motion state data at the moment t, W is the weight parameter of the linear link, B is the correction parameter of the linear link, R is the noise of the actual process, Q is the noise of the prediction process, and>for a priori covariance matrix, P t For the posterior covariance matrix at time t, P t-1 The posterior covariance matrix at the time immediately preceding time t.
In a third aspect, embodiments of the present application provide a vehicle motion state prediction apparatus, including a processor, a memory, and a vehicle motion state prediction program stored on the memory and executable by the processor, wherein the vehicle motion state prediction program, when executed by the processor, implements the steps of the vehicle motion state prediction method as described above.
In a fourth aspect, embodiments of the present application provide a readable storage medium having a vehicle motion state prediction program stored thereon, wherein the vehicle motion state prediction program, when executed by a processor, implements the steps of the vehicle motion state prediction method as described above.
The beneficial effects that technical scheme that this application embodiment provided include:
in the embodiment of the application, vehicle motion state data are acquired, and normalization processing is carried out to obtain normalized vehicle motion state data; obtaining first prediction data through a circulating gate neural network by using the normalized vehicle motion state data; obtaining correction parameters and weight parameters of the attention selection layer through a driving style classifier by using the normalized vehicle motion state data; correcting the first prediction data by using the correction parameter and the weight parameter through the attention selection layer to obtain second prediction data; and carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data to obtain third prediction data, and taking the third prediction data as a final vehicle motion state prediction result. According to the method, the device and the system, the real-time motion state data of the vehicle are obtained, the first prediction data of the motion state of the vehicle is obtained through the circulating gate neural network after normalization processing, and the first prediction data is corrected through the attention selection layer, nonlinear correction and Kalman filtering fusion correction are further carried out, the final motion state prediction result of the vehicle is obtained, the motion state prediction of the vehicle is carried out through the machine learning algorithm, the model construction is simple, the calculated amount is small, the accuracy and the stability are high, the real-time prediction can be carried out by embedding the model into the ECU of the vehicle, the cooperation of a sensor and a high-precision map is not needed, and all the vehicle operation conditions can be covered.
Drawings
FIG. 1 is a schematic flow chart of an embodiment of a method for predicting a vehicle motion state according to the present application;
FIG. 2 is a schematic diagram of a refinement flow chart of step S10 in FIG. 1 of the present application;
FIG. 3 is a schematic diagram of a refinement flow chart of step S20 in FIG. 1 of the present application;
FIG. 4 is a schematic diagram of a refinement flow chart of step S50 in FIG. 1 of the present application;
FIG. 5 is a schematic functional block diagram of an embodiment of a vehicle motion state prediction apparatus according to the present application;
fig. 6 is a schematic hardware configuration diagram of a vehicle motion state prediction apparatus according to an embodiment of the present application.
Detailed Description
In order to make the present application solution better understood by those skilled in the art, the following description will clearly and completely describe the technical solution in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
For the purpose of making the objects, technical solutions and advantages of the present application more apparent, the embodiments of the present application will be described in further detail below with reference to the accompanying drawings.
In a first aspect, embodiments of the present application provide a vehicle motion state prediction method.
In an embodiment, referring to fig. 1, fig. 1 is a flowchart illustrating an embodiment of a vehicle motion state prediction method according to the present application, and as shown in fig. 1, the vehicle motion state prediction method includes:
step S10, vehicle motion state data are obtained, normalization processing is carried out, and normalized vehicle motion state data are obtained.
In this embodiment, real-time motion state data of the vehicle may be obtained through various vehicle-mounted sensors, where the real-time motion state data of the vehicle may include a vehicle speed, a brake pedal stroke, an accelerator pedal stroke, an engine speed, an engine power, and the like, and in order to facilitate processing of the neural network model, normalization processing is further performed, so as to obtain normalized vehicle motion state data.
And step S20, obtaining first prediction data through a circulating gate nerve network by using the normalized vehicle motion state data.
In this embodiment, normalized vehicle motion state data is input into a pre-trained cyclic gate neural network model, and first prediction data is obtained through the cyclic gate neural network.
And step S30, obtaining correction parameters and weight parameters of the attention selection layer through a driving style classifier by using the normalized vehicle motion state data.
In this embodiment, different driving styles may affect the prediction of the motion state of the vehicle, and the driving style classifier is also a pre-trained neural network model, and outputs correction parameters and weight parameters of the corresponding attention selection layer for different driving styles, so as to correct the first prediction data.
Step S40, the first prediction data is corrected by the attention selection layer by using the correction parameter and the weight parameter to obtain second prediction data.
In this embodiment, the attention selection layer corrects the first prediction data by using the correction parameter and the weight parameter to obtain the second prediction data, that is, corrects the first prediction data for different driving styles.
And S50, carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data to obtain third prediction data, and taking the third prediction data as a final vehicle motion state prediction result.
In this embodiment, the nonlinear correction may use a linear rectification function, which is also called a ReLU function (Rectified Linear Unit), which is an activation function commonly used in deep learning and neural networks, and the ReLU function has the characteristics of simple and efficient calculation, nonlinearity, sparsity of activation, convexity, and the like, has better gradient stability, and can improve generalization capability and expression capability of the model. The Kalman filtering fusion correction is a state estimation method, the current state is corrected according to the difference between a measured value (namely an actual value) and a predicted value, a Kalman filtering state correction equation generally consists of a prediction step and an updating step, the Kalman filtering state correction equation is an iterative process, the prediction and updating steps are carried out at each moment, and the optimal state estimation result is finally obtained by continuously correcting the state estimation value. And carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data, so that the accuracy and stability of a final vehicle motion state prediction result are improved.
In this embodiment, real-time motion state data of a vehicle is obtained, after normalization processing, first prediction data of the motion state of the vehicle is obtained through a pre-trained circulating gate neural network, and considering different driving styles, the motion state prediction of the vehicle is affected, a attention selection layer is introduced to correct corresponding correction parameters and weight parameters for different driving styles, further, the accuracy and stability of the prediction are improved, nonlinear correction and Kalman filtering fusion correction are carried out, a final motion state prediction result of the vehicle is obtained, the motion state prediction of the vehicle is carried out through a machine learning algorithm of the application, the model construction is simple, the calculated amount is small, the accuracy and stability are high, the real-time prediction can be carried out by embedding the vehicle into an ECU of the vehicle, the cooperation of a sensor and a high-precision map is not needed, and all vehicle operation conditions can be covered.
Further, in an embodiment, the vehicle motion state data includes a vehicle speed, a brake pedal stroke, an accelerator pedal stroke, an engine speed and an engine power, referring to fig. 2, fig. 2 is a detailed flowchart of step S10 in fig. 1 of the present application, and as shown in fig. 2, step S10 includes:
step S101, acquiring vehicle speed, brake pedal stroke, accelerator pedal stroke, engine speed and engine power;
step S102, the normalized vehicle speed, brake pedal stroke, accelerator pedal stroke, engine speed and engine power are obtained through calculation according to a formula I, wherein the formula I is as follows:
Y＝(X-Xmin)/(Xmax-Xmin)；
wherein Y represents normalized vehicle motion data, X represents acquired vehicle motion state data, xmax represents maximum value of vehicle motion state data, and Xmin represents minimum value of vehicle motion state data.
In this embodiment, real-time motion state data of the vehicle may be obtained through various vehicle-mounted sensors, where the real-time motion state data of the vehicle includes, but is not limited to, a vehicle speed, a brake pedal stroke, an accelerator pedal stroke, an engine speed, an engine power, and the like, and in order to facilitate processing of the neural network model, normalization processing needs to be performed on the obtained real-time motion state data of the vehicle. The normalization is a way of simplifying computation, namely, an expression with dimension is transformed into an expression without dimension to be a scalar, and the normalization simultaneously enables the absolute value of the numerical value of a physical system to be a relation of a certain relative value. And (3) calculating and processing each item of vehicle motion state data acquired from the sensor through a formula I to obtain normalized each item of vehicle motion data, so that the neural network model is convenient to process efficiently and conveniently.
Further, in an embodiment, the loop gate neural network includes an update gate, a reset gate and a hidden layer, referring to fig. 3, fig. 3 is a detailed flow chart of step S20 in fig. 1 of the present application, as shown in fig. 3, step S20 includes:
step S201, according to the normalized vehicle motion state data and the output of the hidden layer at one moment, calculating to obtain the output of the update gate through a formula II, wherein the formula II is as follows:
Z t ＝σ(U z *X t +W z *H t-1 +bz)；
step S202, according to the normalized vehicle motion state data and the output of the hidden layer at one moment, calculating through a formula III to obtain the output of the reset gate, wherein the formula III is as follows:
R t ＝σ(U r *X t +W r *H t-1 +br)；
step S203, according to the output of the update gate and the output of the reset gate, calculating to obtain the first prediction data of the hidden layer according to a formula four:
H t ＝(1-Z t )*H t-1 +Z t *H’ t ，H’ t ＝tanh(U h *X t +W h *R t +bh)；
wherein Z is t To update the gate output, U z To update the input weight parameters of the gate, X t For normalized vehicle motion state data, W z To conceal the weight parameters of the layer in the update gate, H t-1 For the output of the hidden layer at the time immediately preceding the time t, bz is the correction parameter of the update gate, σ () is the activation function σ (x) =1/(1+e) -x ) E is a natural base, R t To reset the gate output, U r For the input weight parameter of the reset gate, wr is the weight parameter of the hidden layer in the reset gate, br is the correction parameter of the reset gate, H t For the output of the hidden layer at time t, i.e. the first predicted data, H' t For the output of the activated hidden layer at time t, tanh () is the activation function tanh (x) = (e) x -e -x )/(e x +e -x )，U h To input weight parameters of hidden layer, W h To reset the weight parameters of the gate in the hidden layer, bh is the correction parameters of the hidden layer.
In this embodiment, the cyclic gate neural network adopts a GRU (Gate Recurrent Unit, cyclic gate unit), which is a special RNN (Recurrent Neural Network, cyclic neural network), and the GRU combines the forgetting gate and the input gate to a single update gate, and because there is one gate control less, parameters of the model are reduced, and cell states and hidden states are combined, so that the model is simpler, and the vehicle state prediction method of the present application can be embedded into an ECU of a vehicle to perform real-time prediction. The historical vehicle motion state data can be normalized and time-sequence constructed in advance to obtain a training data set, the training data set is used for training the circulating gate neural network, when the circulating gate neural network converges, a trained circulating gate neural network model is obtained, and all weight parameters and correction parameters of the updating gate, the resetting gate and the hidden layer are obtained at the same time, including the input weight parameter U of the updating gate z Weight parameter W of hidden layer in update gate z Update the correction parameters bz of the gate, resetInput weight parameter U of door r Weight parameter Wr of hidden layer in reset gate, correction parameter br of reset gate, input weight parameter U of hidden layer h Resetting the weight parameter W of the gate in the hidden layer h And a correction parameter bh of the hidden layer. When training the circulating gate neural network, the predicted time length can be set, namely the circulating gate neural network is set to be used for predicting the vehicle motion state after how much time length.
Further, in an embodiment, step S40 includes:
and correcting the first prediction data through a fifth formula by using the correction parameters and the weight parameters through the attention selection layer to obtain second prediction data, wherein the fifth formula is as follows:
SA＝H t *Wdcc+bdcc；
wherein SA is second predicted data, H t For the first prediction data, wdcc is a weight parameter, and bdcc is a correction parameter.
In this embodiment, the correction parameters and the weight parameters are obtained by using normalized vehicle motion state data through a driving style classifier, and the first prediction data is corrected by using the correction parameters and the weight parameters through the attention selection layer, that is, the first prediction data output by the circulating gate neural network is corrected for different driving styles, so that the obtained second prediction data is more accurate.
Further, referring to fig. 4, fig. 4 is a detailed flow chart of step S50 in fig. 1 of the present application, and as shown in fig. 4, step S50 includes:
step S501, performing nonlinear function correction on the second prediction data based on the feedforward neural network to obtain a correction result of the nonlinear function, where the formula of the nonlinear function correction is as follows:
F t ＝ReLU(SA)；
step S502, according to the correction result of the nonlinear function and the actual value of the vehicle motion state data, carrying out Kalman filtering fusion correction through a formula six to obtain third prediction data, wherein the formula six is as follows:
P t ＝(1-K t )*P t - ；
wherein F is t As a result of the correction of the nonlinear function, reLU () is a nonlinear function ReLU (x) =max (0, x), SA is second prediction data,fusing the corrected predicted value, i.e. the third predicted data, for Kalman filtering>K is a predicted value before Kalman filtering fusion correction t For Kalman gain, Z t For the actual value of the vehicle motion state data at the moment t, W is the weight parameter of the linear link, B is the correction parameter of the linear link, R is the noise of the actual process, Q is the noise of the prediction process, and>for a priori covariance matrix, P t For the posterior covariance matrix at time t, P t-1 The posterior covariance matrix at the time immediately preceding time t.
In this embodiment, the linear rectification function is also called ReLU function (Rectified Linear Unit), which is an activation function commonly used in deep learning and neural networks, and if the input is greater than 0, the value provided as the input is directly returned, and if the input is 0 or less, the value is returned to 0, so that the prediction model can have better gradient stability through nonlinear function correction, and the generalization capability and expression capability of the prediction model can be improved. The Kalman filtering fusion correction is a state estimation method, the current state is corrected according to the difference between the measured value (i.e. actual value) and the predicted value, the Kalman filtering state correction equation generally consists of a prediction step and an updating step, the Kalman filtering state correction equation is an iterative process, the prediction and updating steps are carried out at each moment, and the most significant state is finally obtained by continuously correcting the state estimation valueThe iterative process can effectively utilize the system model and the measurement information to improve the accuracy and stability of state estimation. And carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data, so that the accuracy and stability of a final vehicle motion state prediction result are improved. Wherein the actual value Z of the vehicle motion state data at the time t t The method can be obtained through a vehicle-mounted sensor, the weight parameter W of the linear link and the correction parameter B of the linear link are obtained after training a neural network model, system disturbance and error at each moment, namely noise, are divided into noise of an actual process and noise of a prediction process, and a covariance matrix represents error covariance between a system state estimated value and a real state and is divided into a priori covariance matrix and a posterior covariance matrix.
In a second aspect, embodiments of the present application further provide a vehicle motion state prediction apparatus.
In one embodiment, referring to fig. 5, fig. 5 is a schematic functional block diagram of an embodiment of a vehicle motion state prediction device according to the present application, and as shown in fig. 5, the vehicle motion state prediction device includes:
the acquiring module 10 is configured to acquire vehicle motion state data, and perform normalization processing to obtain normalized vehicle motion state data;
a first prediction module 20, configured to obtain first prediction data through a cyclic gate neural network using the normalized vehicle motion state data;
the classification module 30 is configured to obtain a correction parameter and a weight parameter of the attention selection layer through a driving style classifier using the normalized vehicle motion state data;
a second prediction module 40, configured to modify, through the attention selection layer, the first prediction data using the modification parameter and the weight parameter to obtain second prediction data;
and the correction module 50 is configured to perform nonlinear correction and kalman filter fusion correction on the second prediction data to obtain third prediction data, and take the third prediction data as a final vehicle motion state prediction result.
Further, in one embodiment, the vehicle motion state data includes a vehicle speed, a brake pedal stroke, an accelerator pedal stroke, an engine speed, and an engine power, and the obtaining module 10 is configured to:
acquiring the speed, the travel of a brake pedal, the travel of an accelerator pedal, the rotation speed of an engine and the power of the engine of a vehicle;
the normalized vehicle speed, brake pedal travel, accelerator pedal travel, engine speed and engine power are calculated by the formula one:
Y＝(X-Xmin)/(Xmax-Xmin)；
wherein Y represents normalized vehicle motion data, X represents acquired vehicle motion state data, xmax represents maximum value of vehicle motion state data, and Xmin represents minimum value of vehicle motion state data.
Further, in an embodiment, the cyclic gate neural network includes an update gate, a reset gate, and a hidden layer, the first prediction module 20 is configured to:
and according to the normalized vehicle motion state data and the output of the hidden layer at one moment, calculating to obtain the output of the update door through a formula II, wherein the formula II is as follows:
Z t ＝σ(U z *X t +W z *H t-1 +bz)；
and calculating to obtain the output of the reset gate through a formula III according to the normalized vehicle motion state data and the output of the hidden layer at one moment, wherein the formula III is as follows:
R t ＝σ(U r *X t +W r *H t-1 +br)；
according to the output of the update gate and the output of the reset gate, calculating to obtain first prediction data of the hidden layer through a formula IV, wherein the formula IV is as follows:
H t ＝(1-Z t )*H t-1 +Z t *H’ t ，H’ t ＝tanh(U h *X t +W h *R t +bh)；
wherein Z is t To updateGate output, U z To update the input weight parameters of the gate, X t For normalized vehicle motion state data, W z To conceal the weight parameters of the layer in the update gate, H t-1 For the output of the hidden layer at the time immediately preceding the time t, bz is the correction parameter of the update gate, σ () is the activation function σ (x) =1/(1+e) -x ) E is a natural base, R t To reset the gate output, U r For the input weight parameter of the reset gate, wr is the weight parameter of the hidden layer in the reset gate, br is the correction parameter of the reset gate, H t For the output of the hidden layer at time t, i.e. the first predicted data, H' t For the output of the activated hidden layer at time t, tanh () is the activation function tanh (x) = (e) x -e -x )/(e x +e -x )，U h To input weight parameters of hidden layer, W h To reset the weight parameters of the gate in the hidden layer, bh is the correction parameters of the hidden layer.
Further, in an embodiment, the second prediction module 40 is configured to:
and correcting the first prediction data through a fifth formula by using the correction parameters and the weight parameters through the attention selection layer to obtain second prediction data, wherein the fifth formula is as follows:
SA＝H t *Wdcc+bdcc；
wherein SA is second predicted data, H t For the first prediction data, wdcc is a weight parameter, and bdcc is a correction parameter.
Further, in an embodiment, the correction module 50 is configured to:
and carrying out nonlinear function correction on the second prediction data based on the feedforward neural network to obtain a correction result of the nonlinear function, wherein the nonlinear function correction formula is as follows:
F t ＝ReLU(SA)；
and carrying out Kalman filtering fusion correction according to a correction result of the nonlinear function and an actual value of the vehicle motion state data to obtain third prediction data, wherein the formula six is as follows:
P t ＝(1-K t )*P t - ；
wherein F is t As a result of the correction of the nonlinear function, reLU () is a nonlinear function ReLU (x) =max (0, x), SA is second prediction data,fusing the corrected predicted value, i.e. the third predicted data, for Kalman filtering>K is a predicted value before Kalman filtering fusion correction t For Kalman gain, Z t For the actual value of the vehicle motion state data at the moment t, W is the weight parameter of the linear link, B is the correction parameter of the linear link, R is the noise of the actual process, Q is the noise of the prediction process, and>for a priori covariance matrix, P t For the posterior covariance matrix at time t, P t-1 The posterior covariance matrix at the time immediately preceding time t.
The function implementation of each module in the vehicle motion state prediction device corresponds to each step in the vehicle motion state prediction method embodiment, and the function and implementation process thereof are not described in detail herein.
In a third aspect, embodiments of the present application provide a vehicle motion state prediction apparatus.
Referring to fig. 6, fig. 6 is a schematic hardware configuration diagram of a vehicle motion state prediction apparatus according to an embodiment of the present application. In an embodiment of the application, the vehicle motion state prediction device may include a processor, a memory, a communication interface, and a communication bus.
The communication bus may be of any type for implementing the processor, memory, and communication interface interconnections.
The communication interfaces include input/output (I/O) interfaces, physical interfaces, logical interfaces, and the like for implementing interconnections of devices inside the vehicle motion state prediction apparatus, and interfaces for implementing interconnections of the vehicle motion state prediction apparatus with other apparatuses (e.g., other computing apparatuses or user apparatuses). The physical interface may be an ethernet interface, a fiber optic interface, an ATM interface, etc.; the user device may be a Display, a Keyboard (Keyboard), or the like.
The memory may be various types of storage media such as random access memory (randomaccess memory, RAM), read-only memory (ROM), nonvolatile RAM (non-volatileRAM, NVRAM), flash memory, optical memory, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (electrically erasable PROM, EEPROM), and the like.
The processor may be a general-purpose processor, and the general-purpose processor may call a vehicle motion state prediction program stored in the memory and execute the vehicle motion state prediction method provided in the embodiment of the present application. For example, the general purpose processor may be a central processing unit (central processing unit, CPU). The method executed when the vehicle motion state prediction program is called may refer to various embodiments of the vehicle motion state prediction method of the present application, and will not be described herein.
Those skilled in the art will appreciate that the hardware configuration shown in fig. 6 is not limiting of the application and may include more or fewer components than shown, or may combine certain components, or a different arrangement of components.
In a fourth aspect, embodiments of the present application also provide a readable storage medium.
The present application stores a vehicle motion state prediction program on a readable storage medium, wherein the vehicle motion state prediction program, when executed by a processor, implements the steps of the vehicle motion state prediction method as described above.
The method implemented when the vehicle motion state prediction program is executed may refer to various embodiments of the vehicle motion state prediction method of the present application, and will not be described herein.
It should be noted that, the foregoing embodiment numbers are merely for describing the embodiments, and do not represent the advantages and disadvantages of the embodiments.
The terms "comprising" and "having" and any variations thereof in the description and claims of the present application and in the foregoing drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus. The terms "first," "second," and "third," etc. are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order, and are not limited to the fact that "first," "second," and "third" are not identical.
In the description of embodiments of the present application, "exemplary," "such as," or "for example," etc., are used to indicate an example, instance, or illustration. Any embodiment or design described herein as "exemplary," "such as" or "for example" is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "such as" or "for example," etc., is intended to present related concepts in a concrete fashion.
In the description of the embodiments of the present application, unless otherwise indicated, "/" means or, for example, a/B may represent a or B; the text "and/or" is merely an association relation describing the associated object, and indicates that three relations may exist, for example, a and/or B may indicate: the three cases where a exists alone, a and B exist together, and B exists alone, and in addition, in the description of the embodiments of the present application, "plural" means two or more than two.
In some of the processes described in the embodiments of the present application, a plurality of operations or steps occurring in a particular order are included, but it should be understood that these operations or steps may be performed out of the order in which they occur in the embodiments of the present application or in parallel, the sequence numbers of the operations merely serve to distinguish between the various operations, and the sequence numbers themselves do not represent any order of execution. In addition, the processes may include more or fewer operations, and the operations or steps may be performed in sequence or in parallel, and the operations or steps may be combined.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising several instructions for causing a terminal device to perform the method described in the various embodiments of the present application.
The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the claims, and all equivalent structures or equivalent processes using the descriptions and drawings of the present application, or direct or indirect application in other related technical fields are included in the scope of the claims of the present application.

Claims (10)

1. A vehicle motion state prediction method, characterized by comprising:
acquiring vehicle motion state data, and carrying out normalization processing to obtain normalized vehicle motion state data;
obtaining first prediction data through a circulating gate neural network by using the normalized vehicle motion state data;
obtaining correction parameters and weight parameters of the attention selection layer through a driving style classifier by using the normalized vehicle motion state data;
correcting the first prediction data by using the correction parameter and the weight parameter through the attention selection layer to obtain second prediction data;
and carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data to obtain third prediction data, and taking the third prediction data as a final vehicle motion state prediction result.
2. The vehicle motion state prediction method according to claim 1, wherein the vehicle motion state data includes a vehicle speed, a brake pedal stroke, an accelerator pedal stroke, an engine speed, and an engine power, the obtaining the vehicle motion state data, and performing normalization processing to obtain the normalized vehicle motion state data includes:
acquiring the speed, the travel of a brake pedal, the travel of an accelerator pedal, the rotation speed of an engine and the power of the engine of a vehicle;
the normalized vehicle speed, brake pedal travel, accelerator pedal travel, engine speed and engine power are calculated by the formula one:
Y＝(X-Xmin)/(Xmax-Xmin)；
wherein Y represents normalized vehicle motion data, X represents acquired vehicle motion state data, xmax represents maximum value of vehicle motion state data, and Xmin represents minimum value of vehicle motion state data.
3. The vehicle motion state prediction method according to claim 1, wherein the cyclic gate neural network includes an update gate, a reset gate, and a hidden layer, and the obtaining the first prediction data through the cyclic gate neural network using the normalized vehicle motion state data includes:
and according to the normalized vehicle motion state data and the output of the hidden layer at one moment, calculating to obtain the output of the update door through a formula II, wherein the formula II is as follows:
Z t ＝σ(U z *X t +W z *H t-1 +bz)；
and calculating to obtain the output of the reset gate through a formula III according to the normalized vehicle motion state data and the output of the hidden layer at one moment, wherein the formula III is as follows:
R t ＝σ(U r *X t +W r *H t-1 +br)；
according to the output of the update gate and the output of the reset gate, calculating to obtain first prediction data of the hidden layer through a formula IV, wherein the formula IV is as follows:
H t ＝(1-Z t )*H t-1 +Z t *H’ t ，H’ t ＝tanh(U h *X t +W h *R t +bh)；
wherein Z is t To update the gate output, U z To update the input weight parameters of the gate, X t For normalized vehicle motion state data, W z To conceal the weight parameters of the layer in the update gate, H t-1 For the output of the hidden layer at the time immediately preceding the time t, bz is the correction parameter of the update gate, σ () is the activation function σ (x) =1/(1+e) -x ) E is a natural base, R t To reset the gate output, U r For the input weight parameter of the reset gate, wr is the weight parameter of the hidden layer in the reset gate, br is the correction parameter of the reset gate, H t For the output of the hidden layer at time t, i.e. the first predicted data, H' t For the output of the activated hidden layer at time t, tanh () is the activation function tanh (x) = (e) x -e -x )/(e x +e -x )，U h To input weight parameters of hidden layer, W h To reset the weight parameters of the gate in the hidden layer, bh is the correction parameters of the hidden layer.
4. The vehicle motion state prediction method according to claim 1, wherein the correcting the first prediction data using the correction parameter and the weight parameter by the attention selection layer to obtain the second prediction data includes:
and correcting the first prediction data through a fifth formula by using the correction parameters and the weight parameters through the attention selection layer to obtain second prediction data, wherein the fifth formula is as follows:
SA＝H t *Wdcc+bdcc；
wherein SA is second predicted data, H t For the first predicted data, wdcc is the rightThe heavy parameter, bdcc, is the correction parameter.
5. The vehicle motion state prediction method according to claim 1, wherein the performing nonlinear function correction and kalman filter fusion correction on the second prediction data to obtain third prediction data includes:
and carrying out nonlinear function correction on the second prediction data based on the feedforward neural network to obtain a correction result of the nonlinear function, wherein the nonlinear function correction formula is as follows:
F t ＝ReLU(SA)；
and carrying out Kalman filtering fusion correction according to a correction result of the nonlinear function and an actual value of the vehicle motion state data to obtain third prediction data, wherein the formula six is as follows:
K t ＝P t - *(P t - +R)，P t - ＝W 2 *P t-1 +Q，P t ＝(1-K t )*P t - ；
wherein F is t As a result of the correction of the nonlinear function, reLU () is a nonlinear function ReLU (x) =max (0, x), SA is second prediction data,fusing the corrected predicted value, i.e. the third predicted data, for Kalman filtering>K is a predicted value before Kalman filtering fusion correction t For Kalman gain, Z t For the actual value of the vehicle motion state data at the moment t, W is the weight parameter of the linear link, B is the correction parameter of the linear link, R is the noise of the actual process, Q is the noise of the prediction process, P t - For a priori covariance matrix, P t For the posterior covariance matrix at time t, P t-1 At time tA time-of-day posterior covariance matrix.
6. A vehicle motion state prediction apparatus, characterized by comprising:
the acquisition module is used for acquiring the vehicle motion state data and carrying out normalization processing to obtain normalized vehicle motion state data;
the first prediction module is used for obtaining first prediction data through a circulating gate nerve network by using the normalized vehicle motion state data;
the classification module is used for obtaining correction parameters and weight parameters of the attention selection layer through the driving style classifier by using the normalized vehicle motion state data;
the second prediction module is used for correcting the first prediction data by using the correction parameter and the weight parameter through the attention selection layer to obtain second prediction data;
and the correction module is used for carrying out nonlinear correction and Kalman filtering fusion correction on the second prediction data to obtain third prediction data, and taking the third prediction data as a final vehicle motion state prediction result.
7. The vehicle motion state prediction apparatus according to claim 6, characterized in that the second prediction module is configured to:
and correcting the first prediction data through a fifth formula by using the correction parameters and the weight parameters through the attention selection layer to obtain second prediction data, wherein the fifth formula is as follows:
SA＝H t *Wdcc+bdcc；
wherein SA is second predicted data, H t For the first prediction data, wdcc is a weight parameter, and bdcc is a correction parameter.
8. The vehicle motion state prediction apparatus according to claim 6, wherein the correction module is configured to:
and carrying out nonlinear function correction on the second prediction data based on the feedforward neural network to obtain a correction result of the nonlinear function, wherein the nonlinear function correction formula is as follows:
F t ＝ReLU(SA)；
and carrying out Kalman filtering fusion correction according to a correction result of the nonlinear function and an actual value of the vehicle motion state data to obtain third prediction data, wherein the formula six is as follows:
K t ＝P t - *(P t - +R)，P t - ＝W 2 *P t-1 +Q，P t ＝(1-K t )*P t - ；
wherein F is t As a result of the correction of the nonlinear function, reLU () is a nonlinear function ReLU (x) =max (0, x), SA is second prediction data,fusing the corrected predicted value, i.e. the third predicted data, for Kalman filtering>K is a predicted value before Kalman filtering fusion correction t For Kalman gain, Z t For the actual value of the vehicle motion state data at the moment t, W is the weight parameter of the linear link, B is the correction parameter of the linear link, R is the noise of the actual process, Q is the noise of the prediction process, P t - For a priori covariance matrix, P t For the posterior covariance matrix at time t, P t-1 The posterior covariance matrix at the time immediately preceding time t.
9. A vehicle motion state prediction apparatus, characterized in that it comprises a processor, a memory, and a vehicle motion state prediction program stored on the memory and executable by the processor, wherein the vehicle motion state prediction program, when executed by the processor, implements the steps of the vehicle motion state prediction method according to any one of claims 1 to 5.
10. A readable storage medium, wherein a vehicle motion state prediction program is stored on the readable storage medium, wherein the vehicle motion state prediction program, when executed by a processor, implements the steps of the vehicle motion state prediction method according to any one of claims 1 to 5.
CN202311549404.1A 2023-11-17 2023-11-17 Vehicle motion state prediction method, device, equipment and readable storage medium Pending CN117634025A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311549404.1A CN117634025A (en) 2023-11-17 2023-11-17 Vehicle motion state prediction method, device, equipment and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311549404.1A CN117634025A (en) 2023-11-17 2023-11-17 Vehicle motion state prediction method, device, equipment and readable storage medium

Publications (1)

Publication Number Publication Date
CN117634025A true CN117634025A (en) 2024-03-01

Family

ID=90034918

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311549404.1A Pending CN117634025A (en) 2023-11-17 2023-11-17 Vehicle motion state prediction method, device, equipment and readable storage medium

Country Status (1)

Country Link
CN (1) CN117634025A (en)

Similar Documents

Publication Publication Date Title
CN111708355B (en) Multi-unmanned aerial vehicle action decision method and device based on reinforcement learning
Zeng et al. A survey on visual navigation for artificial agents with deep reinforcement learning
Gao et al. Long short-term memory-based deep recurrent neural networks for target tracking
Li et al. A novel nonlinear optimization method for fitting a noisy Gaussian activation function
US20220363259A1 (en) Method for generating lane changing decision-making model, method for lane changing decision-making of unmanned vehicle and electronic device
Degris et al. Off-policy actor-critic
Momennejad et al. Predicting the future with multi-scale successor representations
CN111027686B (en) Landslide displacement prediction method, device and equipment
CN110459051B (en) Road section feature model training method and device, terminal equipment and storage medium
Tarel et al. Using robust estimation algorithms for tracking explicit curves
CN109726676B (en) Planning method for automatic driving system
JP2862153B2 (en) Rule-based computer system, method of calling rules executed by the rule-based computer system, and incremental function learning method executed by the rule-based computer system
CN112381303A (en) Task index data prediction method and system
CN112381320A (en) Community personnel flow prediction method, system, storage medium and equipment
CN111950633A (en) Neural network training method, neural network target detection method, neural network training device, neural network target detection device and storage medium
CN114694379B (en) Traffic flow prediction method and system based on self-adaptive dynamic graph convolution
Senn et al. Reducing the computational effort of optimal process controllers for continuous state spaces by using incremental learning and post-decision state formulations
Geist et al. A brief survey of parametric value function approximation
US20210166131A1 (en) Training spectral inference neural networks using bilevel optimization
CN117634025A (en) Vehicle motion state prediction method, device, equipment and readable storage medium
CN115805947A (en) Personalized vehicle trajectory prediction method, system, equipment and storage medium
Martín H et al. The k NN-TD reinforcement learning algorithm
US11561326B1 (en) System and method for generating accurate hyperlocal nowcasts
CN112766462A (en) Data processing method, device and computer readable storage medium
CN111931994A (en) Short-term load and photovoltaic power prediction method, system, equipment and medium thereof

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