Motor torque prediction model training method and device, motor torque prediction method and device and vehicle

Abstract

The invention provides a motor torque prediction model training method, a motor torque prediction method, a motor torque prediction device and a vehicle, relates to the technical field of vehicle control, and aims to solve the technical problem of insufficient model precision caused by difficulty in acquiring high-quality real-time data and accurate label data in the prior art. And sharing the theoretical state data to the real state data to obtain mapping state data, predicting a target motor torque based on the mapping state data and the real state data, and updating a weight parameter of the motor torque prediction model based on the theoretical motor torque and the target motor torque. According to the method, the distribution deviation between theoretical state data and real state data is reduced, and the stability of data transmission is ensured, so that the prediction performance of a motor torque prediction model in a target field is improved, accurate prediction and real-time monitoring of the motor torque are realized, and comprehensive support is provided for performance optimization, driving safety and energy management of a new energy vehicle.

Description

Technical Field

[0001] The present invention relates to the field of vehicle control technology, and in particular to a motor torque prediction model training method, a prediction method, a device and a vehicle. Background Art

[0002] In vehicle motor torque prediction, a complex neural network model is used to capture the complex relationship between multiple factors such as motor torque, vehicle speed, driving mode and road conditions, so as to achieve efficient prediction and optimal control of motor torque. The complex neural network model is obtained by training with a large amount of real-time data to achieve high-precision prediction and real-time dynamic adjustment of motor torque. However, there are certain defects in training the model. For example, it is difficult to obtain high-quality real-time data and accurate label data, which leads to insufficient model accuracy and affects the accuracy of the predicted data. Summary of the Invention

[0003] In view of this, an object of the present invention is to provide a motor torque prediction model training method, prediction method, device and vehicle to improve the prediction accuracy of motor torque.

[0004] In a first aspect, the present application provides a motor torque prediction model training method, comprising: Acquire source domain data and target domain data of the vehicle; wherein the source domain data includes theoretical state data and theoretical motor torque corresponding to the theoretical state data, and the target domain data includes real state data corresponding to the theoretical motor torque; Based on the source domain data and the target domain data, a motor torque prediction model is trained. The training operation includes: Determine the characteristic transmission constraint factor based on theoretical state data and actual state data; Based on the characteristic transmission constraint factor, the theoretical state data is shared with the real state data to obtain the mapped state data; Predicting target motor torque based on mapped state data and real state data; Based on the theoretical motor torque and the target motor torque, the weight parameters of the motor torque prediction model are updated.

[0005] Optionally, the source domain data and the target domain data of the motor are obtained, including: Collecting different operating state data of the vehicle under laboratory conditions and the motor torque corresponding to the different operating state data; constructing source domain data based on the different operating state data of the vehicle under laboratory conditions and the motor torque corresponding to the different operating state data; Collect data on different operating states of the vehicle under actual operating conditions; construct target domain data based on the data on different operating states of the vehicle under actual operating conditions.

[0006] Optionally, the feature transmission constraint factor includes a cost constraint matrix and an alignment constraint matrix; determining the feature transmission constraint factor based on the theoretical state data and the actual state data includes: Determine a cost constraint matrix based on the theoretical potential eigenvectors of the theoretical state data and the real potential eigenvectors of the real state data; An alignment constraint matrix is ​​determined based on theoretical statistical values ​​of the theoretical state data and real statistical values ​​of the real state data.

[0007] Optionally, the alignment constraint matrix includes a mean constraint factor and a variance constraint factor; determining the alignment constraint matrix based on theoretical statistical values ​​of the theoretical state data and actual statistical values ​​of the actual state data includes: Determine a mean constraint factor based on a theoretical mean of theoretical state data and a true mean of true state data; A variance constraint factor is determined based on the theoretical variance of the theoretical state data and the actual variance of the actual state data.

[0008] Optionally, based on the characteristic transmission constraint factor, the theoretical state data is shared with the real state data to obtain the mapped state data, including: Determine feature mapping factors based on the cost constraint matrix and the alignment constraint matrix; Mapping state data is determined based on the characteristic mapping factors and the theoretical potential eigenvectors of the theoretical state data.

[0009] Optionally, based on the cost constraint matrix and the alignment constraint matrix, a feature mapping factor is determined, including: Determine the Frobenius norm between the cost constraint matrix and the alignment constraint matrix; Based on the Frobenius norm, the feature mapping factor is determined.

[0010] Optionally, predicting the target motor torque based on the mapped state data and the real state data includes: Perform additive aggregation processing on the mapped state data and the real state data to obtain the comprehensive state data; Based on the comprehensive status data, the target motor torque is predicted.

[0011] In a second aspect, the present application provides a motor torque prediction method, comprising: Get the current status data of the vehicle; Based on the current state data, a motor torque prediction model is used to predict the current motor torque of the vehicle; wherein the motor torque prediction model is trained using the above-mentioned motor torque prediction model training method.

[0012] In a third aspect, the present application further provides a motor torque prediction model training device, comprising: A data acquisition unit, configured to acquire source domain data and target domain data of the vehicle; wherein the source domain data includes theoretical state data and theoretical motor torque corresponding to the theoretical state data, and the target domain data includes real state data corresponding to the theoretical motor torque; The model training unit is used to perform training operations on the motor torque prediction model based on source domain data and target domain data; wherein the training operations include: determining the characteristic transmission constraint factor based on theoretical state data and real state data; sharing the theoretical state data with the real state data to obtain mapped state data based on the characteristic transmission constraint factor; predicting the target motor torque based on the mapped state data and the real state data; and updating the weight parameters of the motor torque prediction model based on the theoretical motor torque and the target motor torque.

[0013] In a fourth aspect, the present application further provides a motor torque prediction device, comprising: A state acquisition unit, used to acquire the current state data of the vehicle; The torque prediction unit is used to predict the current motor torque of the vehicle based on the current state data using a motor torque prediction model; wherein the motor torque prediction model is trained using the above-mentioned motor torque prediction model training method.

[0014] In a fifth aspect, the present application also provides a vehicle comprising the above-mentioned motor torque prediction device.

[0015] In a sixth aspect, the present application also provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a processor, the above-mentioned motor torque prediction model training method or the above-mentioned motor torque prediction method is implemented.

[0016] The embodiments of the present invention provide a motor torque prediction model training method, prediction method, device and vehicle, which share theoretical state data with real state data to obtain mapping state data based on a characteristic transmission constraint factor, predict the target motor torque based on the mapping state data and the real state data, and update the weight parameters of the motor torque prediction model based on the theoretical motor torque and the target motor torque to reduce the distribution deviation between the theoretical state data and the real state data, ensure the stability of data transmission, thereby improving the prediction performance of the motor torque prediction model in the target field, realizing accurate prediction and real-time monitoring of the motor torque, and providing comprehensive support for performance optimization, driving safety and energy management of new energy vehicles.

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A schematic diagram of a flow chart of a motor torque prediction model training method provided by an embodiment of the present invention is shown; Figure 2 A schematic diagram showing a flow chart of another motor torque prediction model training method provided by an embodiment of the present invention; Figure 3 A schematic flow chart of a motor torque prediction method provided by an embodiment of the present invention is shown; Figure 4 A schematic structural diagram of a motor torque prediction model training device provided by an embodiment of the present invention is shown; Figure 5 A schematic structural diagram of a motor torque prediction device provided by an embodiment of the present invention is shown; Figure 6 A schematic structural diagram of an electronic device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0021] In order to facilitate those skilled in the art to better understand this application, the technical terms involved in this application are briefly introduced below.

[0022] Source domain data is used to reflect the theoretical state data of the vehicle under laboratory conditions. In this application, the source domain data includes different operating state data and the motor torque corresponding to the different operating state data.

[0023] The target domain data is used to reflect the real state data of the vehicle under actual operating conditions. In this application, the target domain data includes different working condition state data, etc.

[0024] The feature transmission constraint factor is used to constrain the volatility during the transmission process between the source domain data and the target domain data. In this application, the feature transmission constraint factor includes a cost constraint matrix and an alignment constraint matrix. The alignment constraint matrix includes a mean constraint factor and a variance constraint factor.

[0025] The motor torque prediction model is a deep learning model that predicts the motor torque by learning the mapping relationship between source domain data, target domain data and target motor torque. In this application, the motor torque prediction model includes a multi-channel MLP (Multi-Layer Perceptron) and a long short-term memory network model (Long Short-Term Memory, LSTM) connected in sequence. After determining the feature transmission constraint factor based on the theoretical state data of the source domain data and the real state data of the target domain data through the multi-channel MLP, the theoretical state data is shared with the real state data based on the feature transmission constraint factor to obtain mapping state data, and the comprehensive state data is determined based on the mapping state data and the real state data and then input into the LSTM model. The target motor torque is predicted based on the comprehensive state data through the LSTM model.

[0026] After introducing the technical terms involved in this application, the technical solutions provided by this application are described in detail.

[0027] The present invention provides a method for training a motor torque prediction model. Figure 1 As shown, the motor torque prediction model training method provided in the embodiment of the present application includes at least the following steps: Step 110 , obtaining source domain data and target domain data of the vehicle; wherein the source domain data includes theoretical state data and theoretical motor torque corresponding to the theoretical state data, and the target domain data includes real state data corresponding to the theoretical motor torque.

[0028] In the embodiment of the present application, when acquiring the source domain data and target domain data of the vehicle, the following methods may be used but are not limited to: Collecting different operating state data of the vehicle under laboratory conditions and the motor torque corresponding to the different operating state data; constructing source domain data based on the different operating state data of the vehicle under laboratory conditions and the motor torque corresponding to the different operating state data; Collect data on different operating states of the vehicle under actual operating conditions; construct target domain data based on the data on different operating states of the vehicle under actual operating conditions.

[0029] Specifically, the battery discharge voltage data and motor speed data of the vehicle, as well as the motor torque data corresponding to the motor speed data, are collected under laboratory conditions. The motor torque data is used as the label data. Based on the battery discharge voltage data, motor speed data and label data under laboratory conditions, the source domain data is constructed, namely S = [X s1 , X s2 , Y], where X s1 Represents the battery discharge voltage theoretical state data in the source domain data, X s2 represents the theoretical state data of the motor speed in the source domain data, and Y represents the label data; Collect the battery discharge voltage data and motor speed data of the vehicle under actual operating conditions, and construct the target domain data based on the battery discharge voltage data and motor speed data under actual operating conditions, that is, T=[X t1 , X t2 ], where X t1 represents the actual state data of battery discharge voltage in the target domain data, X t2 Represents the actual state data of the motor speed in the target domain data.

[0030] Step 120: Perform a training operation on the motor torque prediction model based on the source domain data and the target domain data; wherein the training operation includes: determining a characteristic transmission constraint factor based on the theoretical state data and the actual state data; sharing the theoretical state data with the actual state data to obtain mapped state data based on the characteristic transmission constraint factor; predicting the target motor torque based on the mapped state data and the actual state data; and updating the weight parameters of the motor torque prediction model based on the theoretical motor torque and the target motor torque.

[0031] Since the data distribution difference between the source domain data and the target domain data will lead to limited accuracy of the motor torque prediction model, in an embodiment of the present application, the source domain data and the target domain data are processed based on the characteristic transmission constraint factor to reduce the data distribution difference between different domains, thereby improving the prediction performance of the motor torque prediction model in the target domain.

[0032] In the embodiments of this application, Figure 2 As shown in the figure, based on the source domain data and the target domain data, the training operation process of the motor torque prediction model is as follows: Step a: Determine the characteristic transmission constraint factor based on theoretical state data and actual state data.

[0033] In a specific embodiment, the feature transmission constraint factor includes a cost constraint matrix and an alignment constraint matrix, wherein the cost constraint matrix is ​​determined based on the theoretical potential feature vector of the theoretical state data and the real potential feature vector of the real state data; the alignment constraint matrix is ​​determined based on the theoretical statistical value of the theoretical state data and the real statistical value of the real state data; wherein the alignment constraint matrix includes a mean constraint factor and a variance constraint factor; wherein the mean constraint factor is determined based on the theoretical mean of the theoretical state data and the real mean of the real state data; and the variance constraint factor is determined based on the theoretical variance of the theoretical state data and the real variance of the real state data.

[0034] In a specific embodiment, the theoretical latent eigenvector and the true latent eigenvector can be obtained in the following manner: the mean and variance of the theoretical state data and the true state data are extracted respectively through a multi-channel MLP (Multi-Layer Perceptron) to obtain the theoretical mean data and variance data as well as the true mean data and variance data, and then the theoretical mean data and variance data as well as the true mean data and variance data are sampled respectively to obtain the potential eigenvector corresponding to each channel.

[0035] For example, the first theoretical state data in the source domain data is processed by multi-channel MLP Perform mean and variance extraction to obtain the first theoretical mean and the first theoretical variance ; The second theoretical state data in the source domain data Perform mean and variance extraction to obtain the second theoretical mean and the second theoretical variance ; For the first theoretical state data in the source domain data and the second theoretical state data Perform mean and variance extraction to obtain the third theoretical mean and the third theoretical variance value ; According to the theoretical mean data and theoretical variance numbers of the source domain data, sampling is performed to obtain the theoretical potential feature vector corresponding to each channel ,in, In this application, the multi-channel MLP is three-channel. The theoretical state data in the source domain data are respectively averaged, variance and sampled by the three-channel MLP to obtain the first theoretical potential feature vector of the first channel , the second theoretical latent eigenvector of the second channel and the third theoretical latent eigenvector of the third channel ; The first true state data in the target domain data is trained by multi-channel MLP Perform mean and variance extraction to obtain the first true mean and the first true variance value ; The second real state data in the source domain data Perform mean and variance extraction to obtain the second true mean and the second true variance value ; The first real state data in the source domain data and the second real state data Perform mean and variance extraction to obtain the third true mean and the third true variance value ; According to the true mean data and true variance numbers of the source domain data, the true potential feature vector corresponding to each channel is obtained. ,in, In this application, the three-channel MLP is used to perform mean, variance and sampling on the real state data in the target domain data, and the first real potential feature vector of the first channel is obtained. , the second true latent feature vector of the second channel and the third true latent feature vector of the third channel .

[0036] Furthermore, the weight of data transmission is adjusted through the cost constraint matrix so that the data can better adapt to the data distribution of the target field during the transmission process, as follows: The cost constraint matrix is ​​obtained by the similarity between the theoretical potential feature vector and the real potential feature vector under the same channel. The expression of the cost constraint matrix is:

[0037] Where, is the cost constraint matrix, .

[0038] It should be noted that Represents the first cost matrix corresponding to the first theoretical latent eigenvector of the first channel of the source domain data and the first true latent eigenvector of the first channel of the target domain data, a second cost matrix representing a second theoretical latent eigenvector of the second channel of the source domain data and a second true latent eigenvector of the second channel of the target domain data; A third cost matrix corresponding to a third theoretical latent eigenvector of the third channel of the source domain data and a third true latent eigenvector of the third channel of the target domain data is represented.

[0039] Furthermore, the data transmission process is restricted by the mean constraint factor to ensure consistency during the data transmission process and reduce the data distribution offset between the source domain data and the target domain data. The details are as follows: By constraining the theoretical mean data and the actual mean data under the same channel, the mean constraint factor is obtained, where the mean constraint factor expression is:

[0040] Where, is the mean constraint factor, ; , is the mean constraint factor corresponding to the first theoretical mean data of the first channel of the source domain data and the first mean data of the first channel of the target domain data; is the mean constraint factor corresponding to the second theoretical mean data of the second channel of the source domain data and the second mean data of the second channel of the target domain data; is the mean constraint factor corresponding to the third theoretical mean data of the third channel of the source domain data and the third mean data of the third channel of the target domain data; Represents the positive real number space with m columns and n rows; The sum of the elements in each row of the i-th mean constraint factor is equal to the theoretical mean of the i-th channel of the source domain data scalar value of ; The sum of each column element of the i-th mean constraint factor is equal to the true mean of the i-th channel of the target domain data A scalar value of ; T is the transpose.

[0041] Furthermore, the volatility of data transmission is controlled by the variance constraint factor to ensure the stability of data transmission, as follows: By constraining the theoretical variance data and the actual variance data under the same channel, the variance constraint factor is obtained, where the variance constraint factor expression is:

[0042] Where, is the variance constraint factor, =1,2,3; , The variance constraint factor corresponding to the first theoretical variance data of the first channel of the source domain data and the first variance data of the first channel of the target domain data; The variance constraint factor corresponding to the second theoretical variance data of the second channel of the source domain data and the second variance data of the second channel of the target domain data; is the variance constraint factor corresponding to the third theoretical variance data of the third channel of the source domain data and the third third variance data of the third channel of the target domain data; Represents the positive real number space with m columns and n rows; The sum of the elements in each row of the i-th variance constraint factor is equal to the theoretical variance of the i-th channel of the source domain data scalar value of ; The sum of each column element of the i-th variance constraint factor is equal to the true variance of the i-th channel of the target domain data A scalar value of ; T is the transpose.

[0043] Step b: Based on the characteristic transmission constraint factor, the theoretical state data is shared with the real state data to obtain the mapped state data.

[0044] In a specific embodiment, the feature mapping factors are determined based on the cost constraint matrix and the alignment constraint matrix; and the mapping state data are determined based on the feature mapping factors and the theoretical potential feature vectors of the theoretical state data.

[0045] Furthermore, the Frobenius norm between the cost constraint matrix and the alignment constraint matrix can be determined; based on the Frobenius norm, the feature mapping factor is determined, where the feature mapping factor expression is:

[0046] Where, is the feature mapping factor, is the Frobenius norm, is the cost constraint matrix, is the mean constraint factor, The source domain data The first theoretical mean data of the channel and the target domain data The mean constraint factor corresponding to the first mean data of the channel, is the variance constraint factor, The source domain data The first theoretical variance data of the channel and the target domain data The variance constraint factor corresponding to the first variance data of the channel.

[0047] Furthermore, the data transmission process is optimized by the feature mapping factor, thereby improving the generalization ability of the model. Specifically, based on the feature mapping factor and the theoretical potential feature vector of the theoretical state data, the mapping state data is determined, where the mapping state data expression is:

[0048] Where, is the mapping state data, A is the mapping relationship, is the theoretical potential eigenvector, is the feature mapping factor.

[0049] It should be noted that Mapping state data is obtained by sharing the theoretical state data of the i-th channel to the real state data.

[0050] Step c: predicting the target motor torque based on the mapped state data and the real state data.

[0051] In a specific embodiment, the mapped state data and the real state data are added and aggregated to obtain comprehensive state data; and the target motor torque is predicted based on the comprehensive state data.

[0052] In order to better adapt to changes in data distribution in the target field and improve the prediction accuracy and robustness of the motor torque prediction model, in the embodiment of the present application, the mapped state data and the real state data are spliced ​​and aggregated by multi-channel MLP to obtain comprehensive state data, where the comprehensive state data expression is:

[0053] Where, is the comprehensive status data, To share the theoretical state data of the first channel to the real state data to obtain the first mapped state data, is the first true latent feature vector of the first channel, To share the theoretical state data of the second channel to the real state data to obtain the second mapped state data, is the second true latent feature vector of the second channel, To share the theoretical state data of the third channel to the real state data to obtain the third mapped state data, is the third true latent feature vector of the third channel.

[0054] In this application, by sharing the source domain data with the target domain data, the data distribution differences between different domains are effectively reduced; the target motor torque is predicted based on the comprehensive state data through LSTM, thereby improving the prediction performance of the motor torque prediction model in the target domain and realizing accurate prediction and real-time monitoring of the motor torque.

[0055] Step d: Based on the theoretical motor torque and the target motor torque, update the weight parameters of the motor torque prediction model.

[0056] In a specific embodiment, based on the error between the theoretical motor torque and the target motor torque, the weight parameters of the motor torque prediction model are updated until the error reaches a preset error threshold, and the updating of the weight parameters of the motor torque prediction model is stopped. The weight parameters of the motor torque prediction model updated when the training operation is last performed are used to obtain the motor torque prediction model.

[0057] In an embodiment of the present application, the theoretical state data is shared with the real state data based on the characteristic transmission constraint factor to obtain the mapped state data, the target motor torque is predicted based on the mapped state data and the real state data, and the weight parameters of the motor torque prediction model are updated based on the theoretical motor torque and the target motor torque to reduce the distribution deviation between the theoretical state data and the real state data, ensure the stability of data transmission, thereby improving the prediction performance of the motor torque prediction model in the target field, realizing accurate prediction and real-time monitoring of the motor torque, and providing comprehensive support for vehicle performance optimization, driving safety and energy management.

[0058] Based on the above embodiments, the present invention provides a method for predicting motor torque. Figure 3 As shown, the motor torque prediction method provided by the embodiment of the present application includes at least the following steps: Step 210: Obtain current status data of the vehicle; Step 220 : Based on the current state data, the motor torque prediction model is used to predict the current motor torque of the vehicle; wherein the motor torque prediction model is trained using the above-mentioned motor torque prediction model training method.

[0059] In an embodiment of the present application, the vehicle's current state data is input into a trained motor torque prediction model to predict the vehicle's current motor torque, thereby achieving accurate prediction and real-time monitoring of the vehicle's motor torque, providing comprehensive support for vehicle performance optimization, driving safety and energy management.

[0060] Based on the above embodiments, the present application provides a motor torque prediction model training device, see Figure 4 As shown, the motor torque prediction model training device provided in the embodiment of the present application includes at least: The data acquisition unit 310 is configured to acquire source domain data and target domain data of the vehicle; wherein the source domain data includes theoretical state data and theoretical motor torque corresponding to the theoretical state data, and the target domain data includes actual state data corresponding to the theoretical motor torque; The model training unit 320 is used to perform training operations on the motor torque prediction model based on the source domain data and the target domain data; wherein the training operations include: determining the characteristic transmission constraint factor based on the theoretical state data and the actual state data; sharing the theoretical state data with the actual state data to obtain the mapped state data based on the characteristic transmission constraint factor; predicting the target motor torque based on the mapped state data and the actual state data; and updating the weight parameters of the motor torque prediction model based on the theoretical motor torque and the target motor torque.

[0061] In an optional embodiment, the data acquisition unit 310 is further configured to: Collecting different operating state data of the vehicle under laboratory conditions and the motor torque corresponding to the different operating state data; constructing source domain data based on the different operating state data of the vehicle under laboratory conditions and the motor torque corresponding to the different operating state data; Collect data on different operating states of the vehicle under actual operating conditions; construct target domain data based on the data on different operating states of the vehicle under actual operating conditions.

[0062] In an optional embodiment, the feature transmission constraint factor includes a cost constraint matrix and / or an alignment constraint matrix; the model training unit 320 is further configured to: determining a cost constraint matrix based on the theoretical latent eigenvectors of the theoretical state data and the actual latent eigenvectors of the actual state data, and / or; An alignment constraint matrix is ​​determined based on the theoretical mean and the theoretical variance of the theoretical state data and the true mean and the true variance of the true state data.

[0063] In an optional embodiment, the alignment constraint matrix includes a mean constraint factor and / or a variance constraint factor; the model training unit 320 is further configured to: determining a mean constraint factor based on a theoretical mean of the theoretical state data and a true mean of the true state data, and / or; A variance constraint factor is determined based on the theoretical variance of the theoretical state data and the actual variance of the actual state data.

[0064] In an optional embodiment, the model training unit 320 is further configured to: Determine feature mapping factors based on the cost constraint matrix and the alignment constraint matrix; Based on the feature mapping factor, the source domain data is shared with the target domain data to obtain the source domain mapping state data.

[0065] In an optional embodiment, the model training unit 320 is further configured to: Based on the mean constraint factor and the variance constraint factor, a first error constraint factor is obtained; Obtaining a second error constraint factor based on the cost constraint matrix and the first error constraint factor; Based on the objective function, the second error constraint factor is optimized and the feature mapping factor is determined.

[0066] In an optional embodiment, the model training unit 320 is further configured to: Perform concatenation, addition, and aggregation processing on the mapped state data and the real state data to obtain the comprehensive state data; Based on the comprehensive state data, the motor torque prediction model is used to perform prediction processing to obtain the target motor torque.

[0067] Based on the above embodiments, the present invention provides a motor torque prediction device. Figure 5 As shown, the motor torque prediction device provided in the embodiment of the present application includes at least: A state acquisition unit 410 is used to acquire the current state data of the vehicle; The torque prediction unit 420 is used to predict the current motor torque of the vehicle based on the current state data using a motor torque prediction model; wherein the motor torque prediction model is trained using the above-mentioned motor torque prediction model training method.

[0068] Based on the above embodiments, an embodiment of the present application provides a vehicle, and the vehicle provided by the embodiment of the present application includes the above-mentioned motor torque prediction device.

[0069] After introducing the motor torque prediction model training method, training device, prediction method, prediction device and vehicle provided in the embodiments of the present application, the electronic device provided in the embodiments of the present application is briefly introduced.

[0070] See Figure 6 As shown, the electronic device 500 provided in the embodiment of the present application includes at least a processor 501, a memory 502, and a computer program stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program, the motor torque prediction model training method or the motor torque prediction method provided in the embodiment of the present application is implemented.

[0071] The electronic device 500 provided in the embodiment of the present application may further include a bus 503 connecting different components (including the processor 501 and the memory 502). The bus 503 represents one or more of several types of bus structures, including a memory bus, a peripheral bus, a local bus, etc.

[0072] Memory 502 may include a readable storage medium in the form of volatile memory, such as random access memory (RAM) 5021 and / or cache memory 5022, and may further include read-only memory (ROM) 5023. Memory 502 may also include a program tool 5025 having a set (at least one) of program modules 5024. Program modules 5024 include, but are not limited to, an operating subsystem, one or more application programs, other program modules, and program data. Each of these examples, or some combination thereof, may include an implementation of a network environment.

[0073] The processor 501 may be a processing element or a collective term for multiple processing elements. For example, the processor 501 may be a central processing unit (CPU) or one or more integrated circuits configured to implement the motor torque prediction model training method or the motor torque prediction method provided in the embodiments of the present application. Specifically, the processor 501 may be a general-purpose processor, including but not limited to a CPU, an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc.

[0074] The electronic device 500 can communicate with one or more external devices 504 (e.g., keyboard, remote control, etc.), and can also communicate with one or more devices that enable a user to interact with the electronic device 500 (e.g., mobile phone, computer, etc.), and / or communicate with a device that enables the electronic device 500 to communicate with one or more other electronic devices 500 (e.g., router, modem, etc.). Such communication can be performed through an input / output (I / O) interface 505. In addition, the electronic device 500 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN) and / or public network, such as the Internet) through a network adapter 506. Figure 6 As shown, the network adapter 506 communicates with other modules of the electronic device 500 via the bus 503. Figure 6 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 500, including but not limited to microcode, device drivers, redundant processors, external disk drive arrays, disk array (Redundant Arrays of Independent Disks, RAID) subsystems, tape drives, and data backup storage subsystems.

[0075] It should be noted that Figure 6 The electronic device 500 shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0076] The following describes the computer-readable storage medium provided in an embodiment of the present application. The computer-readable storage medium provided in an embodiment of the present application stores computer instructions, which, when executed by a processor, implement the motor torque prediction model training method or motor torque prediction method provided in an embodiment of the present application. Specifically, the computer instructions can be built-in or installed in a processor, so that the processor can implement the motor torque prediction model training method or motor torque prediction method provided in an embodiment of the present application by executing the built-in or installed computer instructions.

[0077] In addition, the motor torque prediction model training method or motor torque prediction method provided in the embodiment of the present application can also be implemented as a computer program product, which includes program code. When the program code runs on a processor, it implements the motor torque prediction model training method or motor torque prediction method provided in the embodiment of the present application.

[0078] The computer program product provided in the embodiments of the present application may adopt one or more computer-readable storage media, and the computer-readable storage medium may be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any suitable combination of the above. Specifically, more specific examples of computer-readable storage media (a non-exhaustive list) include an electrical connection with one or more wires, a portable disk, a hard disk, RAM, ROM, Erasable Programmable Read Only Memory (EPROM), optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.

[0079] The computer program product provided in the embodiments of the present application may be a CD-ROM and include program code, and may also be run on an electronic device such as a computer. However, the computer program product provided in the embodiments of the present application is not limited thereto. In the embodiments of the present application, the computer-readable storage medium may be any tangible medium that contains or stores program code, and the program code may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0080] It should be noted that although several units or subunits of the device are mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, depending on the embodiment of the application, the features and functions of two or more units described above can be embodied in a single unit. Conversely, the features and functions of a single unit described above can be further divided and embodied by multiple units.

[0081] Furthermore, although the operations of the method of the present application are described in a particular order in the accompanying drawings, this does not require or imply that the operations must be performed in this particular order, or that all illustrated operations must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0082] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0083] Obviously, those skilled in the art may make various changes and modifications to the embodiments of the present application without departing from the spirit and scope of the embodiments of the present application. Thus, if these modifications and variations of the embodiments of the present application fall within the scope of the claims of the present application and their equivalents, the present application also intends to include such modifications and variations.

Claims

1. A motor torque prediction model training method, characterized in that: include: Acquire source domain data and target domain data of the vehicle; wherein the source domain data includes theoretical state data and theoretical motor torque corresponding to the theoretical state data, and the target domain data includes real state data corresponding to the theoretical motor torque; Based on the source domain data and the target domain data, a training operation is performed on the motor torque prediction model; wherein the training operation includes: determining a characteristic transmission constraint factor based on the theoretical state data and the actual state data; Based on the characteristic transmission constraint factor, sharing the theoretical state data with the real state data to obtain mapped state data; predicting a target motor torque based on the mapped state data and the real state data; Based on the theoretical motor torque and the target motor torque, weight parameters of the motor torque prediction model are updated.

2. The motor torque prediction model training method according to claim 1, characterized in that: Obtain the source domain data and target domain data of the motor, including: collecting different operating state data of the vehicle under laboratory conditions and motor torques corresponding to the different operating state data; constructing the source domain data based on the different operating state data of the vehicle under laboratory conditions and the motor torques corresponding to the different operating state data; Collecting different operating status data of the vehicle under actual operating conditions; constructing the target domain data based on the different operating status data of the vehicle under actual operating conditions.

3. The motor torque prediction model training method according to claim 1, characterized in that: The feature transfer constraint factors include a cost constraint matrix and an alignment constraint matrix; Determining a characteristic transmission constraint factor based on the theoretical state data and the actual state data includes: determining the cost constraint matrix based on the theoretical potential eigenvectors of the theoretical state data and the real potential eigenvectors of the real state data; The alignment constraint matrix is ​​determined based on the theoretical statistical values ​​of the theoretical state data and the real statistical values ​​of the real state data.

4. The motor torque prediction model training method according to claim 3, characterized in that: The alignment constraint matrix includes a mean constraint factor and a variance constraint factor; Determining the alignment constraint matrix based on the theoretical statistical values ​​of the theoretical state data and the real statistical values ​​of the real state data includes: determining the mean constraint factor based on the theoretical mean of the theoretical state data and the real mean of the real state data; The variance constraint factor is determined based on the theoretical variance of the theoretical state data and the actual variance of the actual state data.

5. The motor torque prediction model training method according to claim 4, characterized in that: Based on the characteristic transmission constraint factor, the theoretical state data is shared with the real state data to obtain mapped state data, including: determining a feature mapping factor based on the cost constraint matrix and the alignment constraint matrix; The mapping state data is determined based on the feature mapping factors and the theoretical potential feature vectors of the theoretical state data.

6. The motor torque prediction model training method according to claim 5, characterized in that: Determining a feature mapping factor based on the cost constraint matrix and the alignment constraint matrix includes: determining a Frobenius norm between the cost constraint matrix and the alignment constraint matrix; The feature mapping factor is determined based on the Frobenius norm.

7. The motor torque prediction model training method according to any one of claims 1 to 6, characterized in that: Predicting a target motor torque based on the mapped state data and the real state data includes: Performing additive aggregation processing on the mapped state data and the real state data to obtain comprehensive state data; The target motor torque is predicted based on the integrated state data.

8. A motor torque prediction method, characterized in that: include: Get the current status data of the vehicle; Based on the current state data, a motor torque prediction model is used to predict the current motor torque of the vehicle; wherein the motor torque prediction model is trained using the motor torque prediction model training method according to any one of claims 1 to 7.

9. A motor torque prediction model training device, characterized in that: include: a data acquisition unit, configured to acquire source domain data and target domain data of the vehicle; wherein the source domain data includes theoretical state data and theoretical motor torque corresponding to the theoretical state data, and the target domain data includes real state data corresponding to the theoretical motor torque; A model training unit is used to perform a training operation on a motor torque prediction model based on the source domain data and the target domain data; wherein the training operation includes: determining a characteristic transmission constraint factor based on the theoretical state data and the real state data; based on the characteristic transmission constraint factor, sharing the theoretical state data with the real state data to obtain mapping state data; based on the mapping state data and the real state data, predicting a target motor torque; and updating a weight parameter of the motor torque prediction model based on the theoretical motor torque and the target motor torque.

10. A motor torque prediction device, characterized in that: include: A state acquisition unit, used to acquire the current state data of the vehicle; A torque prediction unit is used to predict the current motor torque of the vehicle based on the current state data using a motor torque prediction model; wherein the motor torque prediction model is trained using the motor torque prediction model training method according to any one of claims 1 to 7.

11. A vehicle, characterized in that: It includes the motor torque prediction device as claimed in claim 10.

12. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the motor torque prediction model training method according to any one of claims 1 to 7 or the motor torque prediction method according to claim 8 is implemented.

Images