Torque estimation method and device, vehicle, storage medium and product

Abstract

The invention discloses a torque estimation method and device, a vehicle, a storage medium and a product, and relates to the technical field of motor control, and the method comprises the steps: obtaining real-time operation parameters of a target motor, and constructing a target feature vector based on the real-time operation parameters; the target feature vector is input into a target width learning model, the estimated output torque of the target motor under the real-time operation parameters is obtained, and the target width learning model takes the feature vector constructed on the basis of the operation parameters of the motor at one moment as model input data to obtain the estimated output torque of the target motor under the real-time operation parameters. Label data constructed based on the real output torque of the motor at the moment is used as a model to train labels. The accuracy of motor output torque estimation can be improved.

Description

Technical Field

[0001] The present application relates to the field of motor control technology, and in particular to a torque estimation method, device, vehicle, storage medium and product. Background Art

[0002] With the rapid development and widespread application of electric vehicles, the optimization and precise control of the motor as the core power component of electric vehicles plays a vital role in improving the power performance, energy efficiency and safety of the whole vehicle. Among them, accurate estimation of the motor output torque is one of the key technologies to ensure that the vehicle can maintain good performance under various working conditions.

[0003] At present, most common motor torque estimation methods rely on the physical model and working characteristics of the motor, and achieve torque estimation by constructing a torque calculation formula based on parameters such as current and voltage. This method may have good estimation accuracy under specific working conditions (such as constant speed driving), but for dynamic working conditions such as acceleration, braking and climbing, the method of calculating the motor output torque according to a fixed calculation formula cannot accurately calculate the various losses under dynamic working conditions, making it difficult to accurately reflect the actual torque output of the motor, resulting in torque estimation errors, which in turn affects the vehicle's power performance and energy efficiency, and even causes motor overload and damage.

[0004] Therefore, how to improve the accuracy of motor output torque estimation is a problem that needs to be solved urgently. Summary of the invention

[0005] The main purpose of this application is to provide a torque estimation method, device, vehicle, storage medium and product, aiming to improve the accuracy of motor output torque estimation.

[0006] To achieve the above object, the present application provides a torque estimation method, which comprises:

[0007] Acquire real-time operating parameters of the target motor, and construct a target feature vector based on the real-time operating parameters;

[0008] The target feature vector is input into a target width learning model to obtain an estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by using a feature vector constructed based on the operating parameters of the motor at a moment as model input data and label data constructed based on the actual output torque of the motor at the moment as model training data.

[0009] In one embodiment, the method further comprises:

[0010] Acquire an initial sample data set, wherein the initial sample data set includes a plurality of first sample data, and one first sample data includes an operating parameter and a real output torque of the motor at a moment;

[0011] Normalizing each piece of the first sample data to obtain a target sample data set;

[0012] The initial width learning model is trained based on the target sample data set to obtain the target width learning model.

[0013] In one embodiment, the first sample data includes operating parameters of multiple parameter types, and the step of normalizing each piece of the first sample data to obtain a target sample data set includes:

[0014] For any target parameter type in each of the parameter types, determine a maximum operating parameter and a minimum operating parameter among the operating parameters of the target parameter type in each of the first sample data, and determine a first difference between the maximum operating parameter and the minimum operating parameter;

[0015] Based on the minimum operating parameters and the first difference values ​​corresponding to the parameter types, respectively, normalizing the operating parameters in the first sample data to obtain the second sample data corresponding to the first sample data;

[0016] A target sample data set is constructed based on each of the second sample data.

[0017] In one embodiment, the step of normalizing the operating parameters in each of the first sample data based on the minimum operating parameters and the first difference corresponding to each of the parameter types to obtain the second sample data corresponding to each of the first sample data includes:

[0018] For any one target sample data in each of the first sample data, determine a second difference between a first operating parameter of the target parameter type in the target sample data and a minimum operating parameter corresponding to the target parameter type, and divide the second difference by the first difference corresponding to the target parameter type to obtain a second operating parameter of the target parameter type;

[0019] Based on the second operating parameter of each parameter type in the target sample data and the actual output torque, second sample data corresponding to the target sample data are constructed.

[0020] In one embodiment, the step of training the initial width learning model based on the target sample data set to obtain the target width learning model includes:

[0021] For any third sample data in each of the second sample data in the target sample data set, construct a feature vector based on a third operating parameter in the third sample data, and construct label data based on a third real operating parameter in the third sample data;

[0022] The feature vector corresponding to the third sample data is used as model input data, and the label data corresponding to the third sample data is used as a model training label, and the initial width learning model is trained to obtain the target width learning model.

[0023] In one embodiment, the step of training the initial width learning model to obtain the target width learning model includes:

[0024] Inputting each of the model input data into an initial width learning model to obtain each target output weight and a plurality of feature nodes and a plurality of enhancement nodes of each of the model input data;

[0025] Determine the predicted output torque corresponding to each of the model input data based on the target output weights, characteristic nodes and enhanced nodes corresponding to each of the model input data;

[0026] Determining the loss value between the predicted input torque and the model training label corresponding to each of the model input data respectively;

[0027] Determine the minimum value among the loss values, and use the target output weight corresponding to the minimum value as the output weight of the target width learning model.

[0028] In one embodiment, the step of obtaining each target output weight and each of the plurality of feature nodes and the plurality of enhancement nodes of the model input data comprises:

[0029] For any target model input data in each of the model input data, nonlinear processing is performed on the target model input data by using the initial width learning model to obtain a plurality of feature nodes;

[0030] Performing nonlinear processing on each of the characteristic nodes to obtain a plurality of enhanced nodes;

[0031] A target output weight is determined based at least on each of the feature nodes and each of the enhancement nodes.

[0032] In one embodiment, the parameter types of the real-time operating parameters include bus voltage, motor temperature, direct-axis current, quadrature-axis current and motor speed.

[0033] In addition, to achieve the above-mentioned purpose, the present application also provides a torque estimation device, the torque estimation device comprising:

[0034] An acquisition module, used for acquiring real-time operating parameters of a target motor, and constructing a target feature vector based on the real-time operating parameters;

[0035] An estimation module is used to input the target feature vector into a target width learning model to obtain the estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by model training using a feature vector constructed based on the operating parameters of the motor at a moment as model input data and label data constructed based on the actual output torque of the motor at the moment as model training data.

[0036] In addition, to achieve the above-mentioned purpose, the present application also provides a storage medium, which is a computer-readable storage medium, and a program for implementing the torque estimation method is stored on the computer-readable storage medium. The program for implementing the torque estimation method is executed by a processor to implement the steps of the torque estimation method as described above.

[0037] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, including a computer program, which implements the steps of the torque estimation method as described above when executed by a processor.

[0038] The present application provides a torque estimation method, which first obtains the real-time operating parameters of the target motor, constructs a target feature vector based on the real-time operating parameters, and then inputs the target feature vector into a pre-trained target width learning model to obtain the estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by model input data using a feature vector constructed based on the operating parameters of the motor at a moment, and label data constructed based on the actual output torque of the motor at that moment as model training label training.

[0039] In summary, the present application trains the width learning model by pre-training the feature vector constructed based on the operating parameters of the motor at a certain moment as the model input data, and using the label data constructed based on the actual output torque of the motor at that moment as the model training label, to obtain a target width learning model for predicting the motor output torque, so that the target width learning model can learn the actual torque output of the motor under different operating parameters during the training process, that is, the target width learning model can learn the torque output of the motor under various dynamic conditions. Therefore, compared with the traditional method of estimating the motor output torque based on a fixed calculation formula, the present application estimates the motor output torque based on real-time operating parameters through the target width learning model, thereby improving the accuracy of the motor output torque estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0042] Figure 1 This is a schematic diagram of the flow chart of the first embodiment of the torque estimation method of the present application;

[0043] Figure 2 A schematic diagram of a torque estimation process involved in an embodiment of a torque estimation method of the present application;

[0044] Figure 3 A schematic diagram of a training process of a width learning model involved in an embodiment of a torque estimation method of the present application;

[0045] Figure 4 This is a schematic diagram of the module structure of the torque estimation device of the present application;

[0046] Figure 5 Schematic diagram of the device structure of the hardware operating environment involved in the torque estimation method in the embodiment of the present application.

[0047] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0048] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0049] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0050] The main solution of the present application is: to obtain the real-time operating parameters of the target motor, and to construct a target feature vector based on the real-time operating parameters; to input the target feature vector into a target width learning model to obtain the estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by model input data using a feature vector constructed based on the operating parameters of the motor at a moment, and label data constructed based on the actual output torque of the motor at the moment as model training data.

[0051] At present, most common motor torque estimation methods rely on the physical model and working characteristics of the motor, and achieve torque estimation by constructing a torque calculation formula based on parameters such as current and voltage. This method may have good estimation accuracy under specific working conditions (such as constant speed driving), but for dynamic working conditions such as acceleration, braking and climbing, the method of calculating the motor output torque according to a fixed calculation formula cannot accurately calculate the various losses under dynamic working conditions, making it difficult to accurately reflect the actual torque output of the motor, resulting in torque estimation errors, which in turn affects the vehicle's power performance and energy efficiency, and even causes motor overload and damage.

[0052] The present application trains the width learning model in advance using a feature vector constructed based on the operating parameters of the motor at a certain moment as model input data, and using label data constructed based on the actual output torque of the motor at that moment as model training labels to obtain a target width learning model for predicting the motor output torque, so that the target width learning model can learn the actual torque output of the motor under different operating parameters during the training process, that is, the target width learning model can learn the torque output of the motor under various dynamic working conditions. Therefore, compared with the traditional method of estimating the motor output torque based on a fixed calculation formula, the present application estimates the motor output torque based on real-time operating parameters using a target width learning model, thereby improving the accuracy of the motor output torque estimation.

[0053] It should be noted that the execution subject of the method in each embodiment of the torque estimation method of the present application may be a torque estimation system, or a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or a vehicle capable of realizing the above functions, etc., and this embodiment does not specifically limit this. The following takes the torque estimation system as an example to illustrate this embodiment and the following embodiments.

[0054] Based on this, the present application proposes a torque estimation method of the first embodiment, please refer to Figure 1 , the torque estimation method comprises steps S10 to S20:

[0055] Step S10, acquiring real-time operating parameters of the target motor, and constructing a target feature vector based on the real-time operating parameters;

[0056] It should be noted that the motor currently being estimated for output torque is referred to as the target motor.

[0057] The operating parameters of the target motor (hereinafter referred to as the real-time operating parameters for distinction) are acquired in real time, and then a feature vector (hereinafter referred to as the target feature vector for distinction) is constructed based on the real-time operating parameters.

[0058] In this embodiment, the parameter types of the real-time operating parameters include bus voltage, motor temperature, direct-axis current, quadrature-axis current and motor speed.

[0059] It should be noted that the real-time operating parameters of the target motor include bus voltage value, motor temperature value, d-axis current value (ie, direct-axis current value), q-axis current value (ie, quadrature-axis current value) and motor speed value.

[0060] In a feasible implementation, the step of constructing a target feature vector based on the real-time operating parameters includes: performing data preprocessing on each item of data in the real-time operating parameters to obtain a feature quantity corresponding to each item of data; and combining each of the feature quantities to obtain a target feature vector.

[0061] Step S20, input the target feature vector into the target width learning model to obtain the estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by model input data using a feature vector constructed based on the operating parameters of the motor at a moment and label data constructed based on the actual output torque of the motor at the moment as model training data.

[0062] It should be noted that the target width learning model is obtained by training the feature vector constructed based on the operating parameters of the motor at a certain moment as the model input data, and the label data constructed based on the actual output torque of the motor at that moment as the model training label. Among them, the operating parameters of the motor at a specific moment also include bus voltage value, motor temperature value, direct-axis current value, quadrature-axis current value and motor speed value. That is, the parameter types of the operating parameters in a first sample data include bus voltage, motor temperature, direct-axis current, quadrature-axis current and motor speed. In addition, using the width learning model as a training model has the advantages of high accuracy and low load rate.

[0063] The target feature vector is input into the target width learning model to obtain the output torque of the target motor under the implemented operating parameters (hereinafter referred to as the estimated output torque for distinction). The estimated output torque is the prediction result obtained by the target width learning model based on the target feature vector.

[0064] The embodiment of the present application trains the width learning model in advance using a feature vector constructed based on the operating parameters of the motor at a certain moment as model input data, and using label data constructed based on the actual output torque of the motor at that moment as model training labels to obtain a target width learning model for predicting the motor output torque, so that the target width learning model can learn the actual torque output of the motor under different operating parameters during the training process, that is, the target width learning model can learn the torque output of the motor under various dynamic conditions. Therefore, compared with the traditional method of estimating the motor output torque based on a fixed calculation formula, the embodiment of the present application estimates the motor output torque based on real-time operating parameters using the target width learning model, thereby improving the accuracy of the motor output torque estimation.

[0065] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. On this basis, the torque estimation method also includes:

[0066] Step A10, obtaining an initial sample data set, wherein the initial sample data set includes a plurality of first sample data, and one first sample data includes an operating parameter and a real output torque of the motor at a moment;

[0067] The operating parameters and actual output torque of the motor at different times are obtained, a plurality of sample data (hereinafter referred to as first sample data for distinction) are constructed, and a sample data set (hereinafter referred to as initial sample data set for distinction) is constructed based on the plurality of first sample data.

[0068] It should be noted that the embodiment of the present application does not limit the model of the motor in the initial sample data set, and the operating parameters and true output torque of motors of different models can be collected as sample data. In addition, each first sample data includes the operating parameters and true output torque of the motor under different working conditions. Each first sample data includes the true output torque of the operating parameters of the motor at a certain moment under one working condition.

[0069] Exemplarily, each first sample data in the initial sample data set is constructed by collecting the operating parameters and real output torque of the permanent magnet synchronous motor under different control states. Specifically, the permanent magnet synchronous motor is placed in an experimental bench, the bus voltage and current data of the motor are collected through various sensors, and the real-time torque and speed of the motor are accurately collected using a dynamometer. The bench test is repeated in this way to collect the operating parameters and real output torque of the motor under different working conditions, and multiple first sample data are determined based on the collected data to construct an initial sample data set based on each first sample data.

[0070] The initial sample data set is represented as X, and the expression of X is X=[U,TEM ,i d ,i q ,N|T e ], where U is the bus voltage, T EM is the motor temperature, i d is the motor d-axis current, i q is the motor q-axis current, N is the motor speed, [U,T EM ,i d ,i q ,N] as the input value of the training data sample set; T e is the actual output torque of the motor, which is the output value of the training data sample set.

[0071] Step A20, normalizing each piece of the first sample data to obtain a target sample data set;

[0072] A normalization algorithm is used to perform data preprocessing on each first sample data in the initial sample data set to obtain a processed sample data set (hereinafter referred to as a target sample data set for distinction).

[0073] In this embodiment, the first sample data includes operating parameters of multiple parameter types, and the step A20 may include:

[0074] Step A201, for any target parameter type in each of the parameter types, determining a maximum operating parameter and a minimum operating parameter among the operating parameters of the target parameter type in each of the first sample data, and determining a first difference between the maximum operating parameter and the minimum operating parameter;

[0075] It should be noted that the operating parameters in each first sample data include line voltage, motor temperature, direct axis current, quadrature axis current and motor speed, that is, the first sample data includes operating parameters of multiple parameter types. Any parameter type in each parameter type is called a target parameter type for distinction.

[0076] For any parameter type in each parameter type, that is, the above-mentioned target parameter type, determine the operating parameters of the target parameter type among all the operating parameters in each first sample data, and determine the maximum value (hereinafter referred to as the maximum operating parameter for distinction) and the minimum value (hereinafter referred to as the minimum operating parameter for distinction) among the operating parameters of the target parameter type, and then determine the difference between the maximum operating parameter and the minimum operating parameter corresponding to the target parameter type (hereinafter referred to as the first difference for distinction). In this way, the minimum operating parameter and the first difference corresponding to each parameter type are obtained.

[0077] Exemplarily, the parameter types of the operating parameters include bus voltage, motor temperature, direct-axis current, quadrature-axis current, and motor speed. Taking bus voltage as an example, the operating parameters whose parameter type is bus voltage in each first sample data are determined, that is, the bus voltage value in each first sample data, the maximum value (that is, the maximum bus voltage value) and the minimum value (that is, the minimum bus voltage value) in each bus voltage value are determined, and the first difference obtained by subtracting the maximum value from the minimum value is calculated, that is, the first bus voltage difference. In this way, each parameter type corresponds to a maximum operating parameter, a minimum operating parameter, and a first difference.

[0078] Step A202, based on the minimum operating parameter and the first difference value corresponding to each parameter type, respectively normalize the operating parameters in each of the first sample data to obtain second sample data corresponding to each of the first sample data;

[0079] Based on the minimum operating parameters and the first difference corresponding to each parameter type, the operating parameters in each first sample data are normalized to obtain normalized sample data (hereinafter referred to as second sample data for distinction). That is, each first sample data is normalized to obtain the second sample data corresponding to each first sample data.

[0080] In this embodiment, the step A202 may include:

[0081] Step A2021: for any one target sample data in each of the first sample data, determine a second difference between a first operating parameter of the target parameter type in the target sample data and a minimum operating parameter corresponding to the target parameter type, and divide the second difference by the first difference corresponding to the target parameter type to obtain a second operating parameter of the target parameter type;

[0082] It should be noted that any one piece of sample data among the first sample data is referred to as target sample data.

[0083] For any one of the first sample data, i.e., the above-mentioned target sample data, determine the difference between the operating parameter of any parameter type (i.e., the target parameter type) in the target sample data (hereinafter referred to as the first operating parameter for distinction) and the minimum operating parameter corresponding to the target parameter type (hereinafter referred to as the second difference for distinction), and then calculate the result of dividing the second difference by the first difference corresponding to the current target parameter type, i.e., the second operating parameter. It can be understood that the parameter type of the second operating parameter corresponding to the operating parameter of the target parameter type is also the target parameter type.

[0084] Exemplarily, taking bus voltage as an example, the second difference between the bus voltage value in each first sample data and the minimum bus voltage value, that is, the second bus voltage difference, is determined, and then the second bus voltage difference is divided by the first bus voltage difference. The result value is the normalized operating parameter, that is, the second operating parameter whose parameter type is bus voltage.

[0085] Step A2022: construct second sample data corresponding to the target sample data based on the second operating parameters of each parameter type in the target sample data and the actual output torque.

[0086] Based on the second operating parameters of each parameter type in the target sample data and the true output torque in the target sample data, the second sample data after normalization is constructed. In this way, the second sample data corresponding to each first sample data is obtained.

[0087] Exemplarily, the operating parameters in each first sample data in the initial sample data set are normalized. Specifically, the first sample data is represented as x a , where a is a sample identifier. For example, when the number of first sample data is 10, the value range of a is 1 to 10. The second operating parameter of any parameter type in each second sample data is represented as x a ′,x a The expression of ′ is as follows:

[0088]

[0089] Among them, x min is the minimum operating parameter, x max is the maximum operating parameter.

[0090] Step A203: construct a target sample data set based on each of the second sample data.

[0091] After each piece of first sample data is normalized according to the above normalization processing method, a plurality of pieces of second sample data are obtained, and then a target sample data set is constructed based on each piece of second sample data.

[0092] Step A30: training the initial width learning model based on the target sample data set to obtain the target width learning model.

[0093] Based on the target sample data set obtained after normalization processing, a preset width learning module (hereinafter referred to as the initial width learning model for distinction) is trained to obtain a target width learning model, wherein the initial width learning model refers to a width learning model that has not yet been trained.

[0094] In this way, after obtaining a data set containing motor operating parameters and motor output torque under different working conditions, the embodiment of the present application normalizes each sample data in the data set, and controls each data in the sample data within the range of 0 to 1. This not only makes it easier to reflect the characteristics of each sample data and the differences between each sample data, thereby improving the accuracy of subsequent model training results (i.e., target width learning model), but also reduces the amount of data calculation in the subsequent model training process, thereby improving the efficiency of model training.

[0095] Based on the first and second embodiments of the present application, in the third embodiment of the present application, the same or similar contents as those of the above-mentioned first and second embodiments can be referred to the above introduction and will not be repeated later.

[0096] On this basis, the step A30 may include:

[0097] Step A301, for any third sample data in each of the second sample data in the target sample data set, construct a feature vector based on a third operating parameter in the third sample data, and construct label data based on a third real operating parameter in the third sample data;

[0098] Any sample data in each second sample data in the target sample data set is called the third sample data, and a feature vector is constructed based on the operation parameter in the third sample data (hereinafter referred to as the third operation parameter for distinction), and label data is constructed based on the real operation parameter in the third sample data (hereinafter referred to as the third real operation parameter for distinction). That is, the corresponding feature vector and label data are constructed for each second sample data.

[0099] Step A302: Use the feature vector corresponding to the third sample data as model input data, use the label data corresponding to the third sample data as model training labels, and train the initial width learning model to obtain the target width learning model.

[0100] The feature vector corresponding to the third sample data is used as the model input data, and the label data corresponding to the third sample data is used as the model training label, and the initial width learning model is trained to obtain the target width learning model. That is, the feature vector corresponding to each second sample data is used as the model input data, and the label data corresponding to each second sample data is used as the model training label.

[0101] In this embodiment, step A302 may include:

[0102] Step B10, inputting each of the model input data into an initial width learning model to obtain each target output weight and a plurality of feature nodes and a plurality of enhancement nodes of each of the model input data;

[0103] The model input data is input into the initial width learning model to obtain the output weight (hereinafter referred to as the target output weight) used to guide the prediction of the motor output torque and multiple feature nodes and multiple enhanced nodes determined based on the model input data. Among them, the output weight is a type of model parameter. The feature node is a mapping feature generated from the model input data. The enhanced node is a node obtained by further processing the feature node. They increase the width and complexity of the network, thereby improving the expressiveness of the model.

[0104] In this embodiment, the step B10 may include:

[0105] Step B101, for any target model input data in each of the model input data, nonlinear processing is performed on the target model input data by using the initial width learning model to obtain a plurality of feature nodes;

[0106] For any model input data in each model input data (hereinafter referred to as target model input data for distinction), after the target model input data is input into the initial width learning model, the target model input data is nonlinearly processed by the initial width learning model to obtain multiple feature nodes.

[0107] For example, the feature node is represented as Z i ,i ranges from 1 to n, where n is the number of feature nodes. The embodiment of the present application does not limit the number of feature nodes. Feature Node Z i The expression of Z is: i =φ i (x a 'W ei +β ei ),i=1,2,…,n, where φ i is a linear or nonlinear activation function, W ei is the random weight of the model to be trained, β ei is the bias of the model to be trained. And the activation function, random weight and bias used in the nonlinear processing of each model input data are different. Then, a feature node group is constructed based on each feature node, and the feature node group is represented by Z n , Z n The expression of Z n =[Z1,Z2,…,Z n ], the number of feature node groups is n.

[0108] Step B102, performing nonlinear processing on each of the characteristic nodes to obtain a plurality of enhanced nodes;

[0109] Each characteristic node is processed nonlinearly to obtain multiple enhanced nodes.

[0110] For example, the enhanced node is represented as H j ,j has a value range of 1 to m, where m is the number of enhanced nodes. The embodiment of the present application does not limit the number of enhanced nodes. j The expression is: H j =ξ j (Z n W hj +β hj ),j=1,2,3,…m, where ξ j is a nonlinear activation function, W hj is the random weight of the model to be trained, β hj is the bias of the model to be trained. And the activation function, random weight and bias used in the nonlinear processing of each feature node group are different. Then, based on each enhanced node, an enhanced node group is constructed, and the enhanced node group is represented as H m , H m The expression is H m =[H1,H2,…,H m ], the number of enhanced node groups is m.

[0111] Step B103, determining a target output weight based at least on each of the feature nodes and each of the enhancement nodes.

[0112] In a specific implementation, the target output weight is determined based on each feature node, each enhancement node, the number of samples in the target sample data set, and the third sample data.

[0113] Exemplarily, the target output weight is represented as W m , W m The calculation formula is: m =(A T A+λI a ) -1 A T Y a ,in, A + represents the pseudo-inverse of A, where A is the matrix composed of each feature node and each enhancement node, and I a is the identity matrix of the model input data, Y a is the model training label. λ represents the number of second sample data, that is, the number of sample data in the target sample data set.

[0114] Step B20, determining the predicted output torque corresponding to each of the model input data based on the target output weights, characteristic nodes and enhanced nodes corresponding to each of the model input data;

[0115] Based on the target output weight, multiple feature nodes and multiple enhancement nodes corresponding to each model input data, the prediction result corresponding to the model input data, that is, the predicted output torque, is determined.

[0116] Exemplarily, the predicted output torque is expressed as The calculation formula is like Figure 2 The figure shows a schematic diagram of the torque estimation process. The initial sample data set X is normalized to obtain the target sample data set X′. Then, for each sample data in the target sample data set, multiple feature nodes and multiple enhancement nodes corresponding to each sample data are determined. Then, an output weight W is determined based on each feature node and each enhancement node. m Then, based on the output weight, each characteristic node and each enhanced node of the sample data, the predicted output torque corresponding to each sample data is determined.

[0117] Step B30, respectively determining the loss value between the predicted input torque corresponding to each of the model input data and the model training label;

[0118] Substitute the predicted input torque and model training label corresponding to each model input data into the preset loss function to obtain the loss value.

[0119] It should be noted that the loss function pre-set in the embodiment of the present application is RMSE (Root Mean Square Error).

[0120] The predicted output torque and the model training label corresponding to each second sample data are substituted into the preset loss function to obtain the loss value corresponding to each second sample data.

[0121] Exemplarily, the expression of the preset loss function is as follows:

[0122]

[0123] Among them, Y a is the predicted output torque corresponding to the second sample data, The model training label corresponding to the second sample data.

[0124] Step B40, determining the minimum value among the loss values, and using the target output weight corresponding to the minimum value as the output weight of the target width learning model.

[0125] After determining the loss value corresponding to each second sample data, the second sample data corresponding to the minimum value among the loss values ​​is determined, and the target output weight corresponding to the second sample data with the minimum loss value is used as the output weight of the target width learning model.

[0126] For example, according to the regression evaluation index, the weight W corresponding to the minimum RMSE (loss value) is selected. m The output layer weights of the model are learned as target width.

[0127] In one possible implementation, if Figure 3 The figure shows a schematic diagram of the training process of the width learning model. First, the actual output torque of the motor's operating parameters under different working conditions is collected, and an initial sample data set is constructed based on the collected multiple first sample data; the initial sample data set is normalized to obtain a target sample data set; for each second sample data in the target sample data set, a feature vector and label data are constructed; the feature vector of each second sample data is used as model input data, and the label data of each second sample data is used as a model training label; the model input data is input into the model to be trained to obtain the target output weight and predicted output torque; the loss value between the predicted output torque corresponding to each second sample data and the model training label is calculated; the target output weight corresponding to the second sample data with the smallest loss value is used as the output weight of the target width learning model.

[0128] In this way, the embodiment of the present application estimates the motor output torque by training the width learning model, so that the trained target width learning model can learn the actual torque output under different motor operating parameters from a large amount of sample data. Through the synergistic effect of the feature nodes and the enhancement nodes in the width learning model, the multi-dimensional features in the motor operating parameters are effectively captured, and the accuracy of the motor output torque estimation is improved. The width learning model can realize the torque estimation of the motor under different working conditions. Compared with the formula method, the torque estimation method based on width learning does not rely too much on the physical model of the motor, thereby reducing the complexity of the motor output torque estimation algorithm.

[0129] The present application also provides a torque estimation device, please refer to Figure 4 , the torque estimation device comprises:

[0130] An acquisition module 10 is used to acquire real-time operating parameters of a target motor and construct a target feature vector based on the real-time operating parameters;

[0131] The estimation module 20 is used to input the target feature vector into the target width learning model to obtain the estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by model input data using a feature vector constructed based on the operating parameters of the motor at a moment and label data constructed based on the actual output torque of the motor at the moment as model training data.

[0132] Optionally, the torque estimation device further comprises a training module, wherein the training module is used for:

[0133] Acquire an initial sample data set, wherein the initial sample data set includes a plurality of first sample data, and one first sample data includes an operating parameter and a real output torque of the motor at a moment;

[0134] Normalizing each piece of the first sample data to obtain a target sample data set;

[0135] The initial width learning model is trained based on the target sample data set to obtain the target width learning model.

[0136] Optionally, the first sample data includes operating parameters of multiple parameter types, and the training module is further used for:

[0137] For any target parameter type in each of the parameter types, determine a maximum operating parameter and a minimum operating parameter among the operating parameters of the target parameter type in each of the first sample data, and determine a first difference between the maximum operating parameter and the minimum operating parameter;

[0138] Based on the minimum operating parameters and the first difference values ​​corresponding to the parameter types, respectively, normalizing the operating parameters in the first sample data to obtain the second sample data corresponding to the first sample data;

[0139] A target sample data set is constructed based on each of the second sample data.

[0140] Optionally, the training module is further used to:

[0141] For any one target sample data in each of the first sample data, determine a second difference between a first operating parameter of the target parameter type in the target sample data and a minimum operating parameter corresponding to the target parameter type, and divide the second difference by the first difference corresponding to the target parameter type to obtain a second operating parameter of the target parameter type;

[0142] Based on the second operating parameter of each parameter type in the target sample data and the actual output torque, second sample data corresponding to the target sample data are constructed.

[0143] Optionally, the training module is further used to:

[0144] For any third sample data in each of the second sample data in the target sample data set, construct a feature vector based on a third operating parameter in the third sample data, and construct label data based on a third real operating parameter in the third sample data;

[0145] The feature vector corresponding to the third sample data is used as model input data, and the label data corresponding to the third sample data is used as a model training label, and the initial width learning model is trained to obtain the target width learning model.

[0146] Optionally, the training module is further used to:

[0147] Inputting each of the model input data into an initial width learning model to obtain each target output weight and a plurality of feature nodes and a plurality of enhancement nodes of each of the model input data;

[0148] Determine the predicted output torque corresponding to each of the model input data based on the target output weights, characteristic nodes and enhanced nodes corresponding to each of the model input data;

[0149] Determining the loss value between the predicted input torque and the model training label corresponding to each of the model input data respectively;

[0150] Determine the minimum value among the loss values, and use the target output weight corresponding to the minimum value as the output weight of the target width learning model.

[0151] Optionally, the training module is further used to:

[0152] For any target model input data in each of the model input data, nonlinear processing is performed on the target model input data by using the initial width learning model to obtain a plurality of feature nodes;

[0153] Performing nonlinear processing on each of the characteristic nodes to obtain a plurality of enhanced nodes;

[0154] A target output weight is determined based at least on each of the feature nodes and each of the enhancement nodes.

[0155] Optionally, the parameter types of the real-time operating parameters include bus voltage, motor temperature, direct-axis current, quadrature-axis current and motor speed.

[0156] The torque estimation device provided in the embodiment of the present application adopts the torque estimation method in the above embodiment, which can solve the technical problem of how to improve the accuracy of the motor output torque estimation. Compared with the prior art, the beneficial effects of the torque estimation device provided in the embodiment of the present application are the same as the beneficial effects of the torque estimation method provided in the above embodiment, and the other technical features in the torque estimation device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0157] The present application provides a vehicle, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the torque estimation method in the above-mentioned embodiment one.

[0158] Reference below Figure 5 , which shows a schematic structural diagram of a vehicle suitable for implementing an embodiment of the present application. Figure 5 The vehicle shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0159] like Figure 5 As shown, the vehicle may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read On ly Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. Various programs and data required for vehicle operation are also stored in RAM1004. The processing device 1001, ROM1002, and RAM1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems may be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 may allow the vehicle to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a vehicle with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have instead.

[0160] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0161] The vehicle provided by the present application adopts the torque estimation method in the above embodiment, which can solve the technical problem of how to improve the accuracy of the motor output torque estimation. Compared with the prior art, the beneficial effects of the vehicle provided by the present application are the same as the beneficial effects of the torque estimation method provided by the above embodiment, and the other technical features in the vehicle are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0162] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0163] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0164] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the torque estimation method in the above-mentioned embodiment.

[0165] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM: Erasable Programmab le Read On ly Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read On ly Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0166] The computer-readable storage medium may be included in the vehicle, or may exist independently without being installed in the vehicle.

[0167] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by a vehicle, the vehicle: obtains the real-time operating parameters of the target motor and constructs a target feature vector based on the real-time operating parameters; inputs the target feature vector into a target width learning model to obtain an estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by model input data using a feature vector constructed based on the operating parameters of the motor at a moment and label data constructed based on the actual output torque of the motor at the moment as model training data.

[0168] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0169] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0170] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0171] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned torque estimation method, and can solve the technical problem of how to improve the accuracy of motor output torque estimation. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the present application are the same as the beneficial effects of the torque estimation method provided in the above-mentioned embodiment, and will not be repeated here.

[0172] An embodiment of the present application provides a computer program product, including a computer program, which implements the steps of the torque estimation method as described above when executed by a processor.

[0173] The computer program product provided in this application can improve the accuracy of motor output torque estimation. Compared with the prior art, the beneficial effects of the computer program product provided in the embodiment of this application are the same as the beneficial effects of the torque estimation method provided in the above embodiment, which will not be repeated here.

[0174] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent processing scope of the present application.

Claims

1. A torque estimation method, characterized in that: The torque estimation method comprises: Acquire real-time operating parameters of the target motor, and construct a target feature vector based on the real-time operating parameters; The target feature vector is input into a target width learning model to obtain an estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by using a feature vector constructed based on the operating parameters of the motor at a moment as model input data and label data constructed based on the actual output torque of the motor at the moment as model training data.

2. The method according to claim 1, characterized in that The method further comprises: Acquire an initial sample data set, wherein the initial sample data set includes a plurality of first sample data, and one first sample data includes an operating parameter and a real output torque of the motor at a moment; Normalizing each piece of the first sample data to obtain a target sample data set; The initial width learning model is trained based on the target sample data set to obtain the target width learning model.

3. The method according to claim 2, characterized in that The first sample data includes operating parameters of multiple parameter types, and the step of normalizing each piece of the first sample data to obtain a target sample data set includes: For any target parameter type in each of the parameter types, determine a maximum operating parameter and a minimum operating parameter among the operating parameters of the target parameter type in each of the first sample data, and determine a first difference between the maximum operating parameter and the minimum operating parameter; Based on the minimum operating parameters and the first difference values ​​corresponding to the parameter types, respectively, normalizing the operating parameters in the first sample data to obtain the second sample data corresponding to the first sample data; A target sample data set is constructed based on each of the second sample data.

4. The method according to claim 3, characterized in that The step of normalizing the operating parameters in each of the first sample data based on the minimum operating parameters and the first difference corresponding to each of the parameter types to obtain the second sample data corresponding to each of the first sample data includes: For any one target sample data in each of the first sample data, determine a second difference between a first operating parameter of the target parameter type in the target sample data and a minimum operating parameter corresponding to the target parameter type, and divide the second difference by the first difference corresponding to the target parameter type to obtain a second operating parameter of the target parameter type; Based on the second operating parameter of each parameter type in the target sample data and the actual output torque, second sample data corresponding to the target sample data are constructed.

5. The method according to claim 3, characterized in that The step of training the initial width learning model based on the target sample data set to obtain the target width learning model includes: For any one third sample data in each of the second sample data in the target sample data set, construct a feature vector based on a third operating parameter in the third sample data, and construct label data based on a third real operating parameter in the third sample data; The feature vector corresponding to the third sample data is used as model input data, and the label data corresponding to the third sample data is used as a model training label, and the initial width learning model is trained to obtain the target width learning model.

6. The method according to claim 5, characterized in that The step of training the initial width learning model to obtain the target width learning model includes: Inputting each of the model input data into an initial width learning model to obtain each target output weight and a plurality of feature nodes and a plurality of enhancement nodes of each of the model input data; Determine the predicted output torque corresponding to each of the model input data based on the target output weights, characteristic nodes and enhanced nodes corresponding to each of the model input data; Determining the loss value between the predicted input torque and the model training label corresponding to each of the model input data respectively; Determine the minimum value among the loss values, and use the target output weight corresponding to the minimum value as the output weight of the target width learning model.

7. The method according to claim 6, characterized in that The step of obtaining each target output weight and each of the model input data's respective multiple feature nodes and multiple enhancement nodes comprises: For any target model input data in each of the model input data, nonlinear processing is performed on the target model input data by using the initial width learning model to obtain a plurality of feature nodes; Performing nonlinear processing on each of the characteristic nodes to obtain a plurality of enhanced nodes; A target output weight is determined based at least on each of the feature nodes and each of the enhancement nodes.

8. The method according to any one of claims 1 to 7, characterized in that The parameter types of the real-time operating parameters include bus voltage, motor temperature, direct-axis current, quadrature-axis current and motor speed.

9. A torque estimation device, characterized in that: The torque estimation device comprises: An acquisition module, used for acquiring real-time operating parameters of a target motor, and constructing a target feature vector based on the real-time operating parameters; An estimation module is used to input the target feature vector into a target width learning model to obtain the estimated output torque of the target motor under the real-time operating parameters, wherein the target width learning model is obtained by model training using a feature vector constructed based on the operating parameters of the motor at a moment as model input data and label data constructed based on the actual output torque of the motor at the moment as model training data.

10. A vehicle, characterized in that: The vehicle comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the torque estimation method according to any one of claims 1 to 8.

11. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the torque estimation method according to any one of claims 1 to 8 are implemented.

12. A computer program product, characterized in that The computer program product comprises a computer program, which implements the steps of the torque estimation method according to any one of claims 1 to 8 when executed by a processor.

Images