Performance test method for electric drive controller

Abstract

The present invention discloses a performance test method for an electric drive controller, which belongs to the field of performance testing, and includes: receiving a target control instruction, and determining a control execution window based on the target control instruction; executing the target control instruction through the electric drive controller, and collecting the timing data of the control parameters in the control execution window; forming an instruction execution matrix based on the timing data and the control execution window; obtaining the heterogeneous parameter matching analysis results and the parameter self-matching analysis results according to the instruction execution matrix; integrating the heterogeneous parameter matching analysis results and the parameter self-matching analysis results, and executing the control performance results of the target control instruction online. The present application solves the technical problem that the existing electric drive controller performance test method is limited to offline evaluation, resulting in the inability to accurately obtain the dynamic control performance of the electric drive controller in real time, and achieves the technical effect of accurately evaluating the dynamic control performance of the electric drive controller in real time through the timing data of the electric drive controller.

Description

Technical Field

[0001] The present invention relates to the field of performance testing, and in particular to a performance testing method for an electric drive controller. Background Art

[0002] The electric drive controller is the core component of the electric drive system. The quality of its performance directly affects the control quality of the electric drive system. In order to ensure the efficient and stable operation of the electric drive system, the control performance of the electric drive controller needs to be comprehensively and accurately tested and evaluated. In the prior art, the performance test method of the electric drive controller mainly focuses on offline evaluation, calibrating the electric drive controller under specific working conditions, and judging its performance by analyzing the input and output parameters of the electric drive controller. However, offline testing cannot truly reflect the dynamic response characteristics of the electric drive controller in actual operation, and it is difficult to simulate complex and changeable control scenarios, which leads to deviations between the test results and the actual control performance of the electric drive controller, and cannot provide a reliable basis. In addition, the performance data obtained by offline testing is often static and intermittent, lacks time-series correlation, and cannot characterize the continuous process of the electric drive controller dynamically executing instructions, making it difficult to timely discover and locate control deviations, which restricts the control accuracy and stability of the electric drive system. Therefore, the existing performance test method of the electric drive controller is limited to offline evaluation, and it is impossible to accurately obtain the dynamic control performance of the electric drive controller in real time. Summary of the invention

[0003] The present application provides a performance testing method for an electric drive controller, aiming to solve the technical problem that the existing electric drive controller performance testing method is limited to offline evaluation, resulting in the inability to accurately obtain the dynamic control performance of the electric drive controller in real time.

[0004] The present application discloses a performance testing method for an electric drive controller, the method comprising: receiving a target control instruction, determining a control execution window based on the target control instruction, wherein the control execution window comprises N control nodes, where N is a positive integer; executing the target control instruction through the electric drive controller, collecting time series data of M control parameters in the control execution window, wherein M is a positive integer; forming an instruction execution matrix based on the time series data and the control execution window, wherein, in the instruction execution matrix, rows correspond to control parameters, and columns correspond to control parameter values; performing a matrix longitudinal analysis according to the instruction execution matrix, respectively obtaining matching degrees of multiple control parameter values ​​under the N control nodes, and obtaining N heterogeneous parameter matching analysis results; performing a matrix transverse analysis according to the instruction execution matrix, obtaining matching degrees of adjacent control parameter values ​​of each control parameter under each control node, and obtaining M parameter self-matching analysis results; integrating the N heterogeneous parameter matching analysis results with the M parameter self-matching analysis results, and obtaining online the control performance result of the electric drive controller executing the target control instruction.

[0005] The technical solution provided in this application has at least the following technical effects or advantages:

[0006] By receiving the target control instruction, a control execution window containing N control nodes is determined, providing a time range and sampling points for the subsequent collection of the electric drive controller execution data; the electric drive controller executes the target control instruction, collects the timing data of M control parameters in the control execution window, and obtains the original data reflecting the dynamic execution effect of the electric drive controller; an instruction execution matrix is ​​established based on the timing data and the control execution window, wherein in the instruction execution matrix, the rows correspond to the control parameters and the columns correspond to the control parameter values, which is convenient for the subsequent parameter matching degree analysis; the instruction execution matrix is ​​analyzed vertically to obtain the matching degree between different control parameter values ​​at each column of the matrix, that is, at each control node, thereby obtaining N parameters reflecting the execution effect of different control nodes. The invention discloses a technical scheme for online testing of the dynamic control performance of an electric drive controller, which solves the technical problem that the existing electric drive controller performance testing method is limited to offline evaluation, resulting in the inability to accurately obtain the dynamic control performance of the electric drive controller in real time, and achieves the technical effect of accurately evaluating the dynamic control performance of the electric drive controller in real time through the timing data of the electric drive controller.

[0007] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A flow chart of a performance testing method for an electric drive controller is provided for an embodiment of the present application;

[0009] Figure 2 A schematic diagram for constructing a first parameter input matrix in a performance testing method for an electric drive controller is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0010] The overall idea of ​​the technical solution provided by this application is as follows:

[0011] An embodiment of the present application provides a performance testing method for an electric drive controller. During the process of the electric drive controller executing a target control instruction, the dynamic control performance of the electric drive controller is tested online by dynamically collecting timing data and constructing an instruction execution matrix, combined with vertical and horizontal analysis of the matrix.

[0012] Specifically, first, the target control instruction is received, and the control execution window containing N control nodes is determined based on the target control instruction to specify the time range and sampling point of data collection. Then, in the process of the electric drive controller executing the target control instruction, the timing data of M control parameters are collected, and the instruction execution matrix is ​​constructed based on the timing data and the control execution window. Next, the constructed instruction execution matrix is ​​analyzed vertically and horizontally respectively. The vertical analysis obtains the matching degree of different parameters under the same control node, and obtains N hetero-parameter matching results. The horizontal analysis obtains the matching degree of the same parameter between adjacent control nodes, and obtains M parameter self-matching results. Subsequently, the two types of analysis results are integrated to form a quantitative indicator reflecting the comprehensive dynamic control performance of the electric drive controller.

[0013] In summary, this application makes full use of the timing data generated by the electric drive controller, combines matrix construction and multi-dimensional analysis, and establishes a dynamic, real-time, and quantitative performance test system for the electric drive controller, breaking through the limitations of existing static offline tests. It can accurately present the dynamic control effect of the electric drive controller online, providing a precise and reliable basis for system control.

[0014] After introducing the basic principles of the present application, the non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the drawings in the specification.

[0015] like Figure 1 As shown, an embodiment of the present application provides a performance testing method for an electric drive controller, the method comprising:

[0016] S1: receiving a target control instruction, and determining a control execution window based on the target control instruction, wherein the control execution window includes N control nodes, where N is a positive integer;

[0017] Specifically, the target control instruction refers to the control request issued by the driver based on driving needs (such as braking, acceleration, etc.), which is generated by the vehicle controller after analysis, reflecting the driver's expected control target for the vehicle. After receiving the driver's request, the vehicle controller comprehensively analyzes the current driving conditions and generates a target control instruction corresponding to the driver's request.

[0018] After generating the target control instruction, the vehicle controller determines the corresponding control execution window according to the control instruction of the target control instruction. The control execution window refers to the time range in which the electric drive controller executes the target control instruction, which specifies the time domain boundary for subsequent data acquisition and performance analysis. The span of the control execution window will also vary for target control instructions of different types and parameters. For example, for a braking instruction, the greater the braking depth, the longer the execution time may be; for an acceleration instruction, the higher the target speed, the longer the execution time. The control execution window contains N control nodes, each of which represents a time sampling point. N is a positive integer, and its size depends on the time span of the control execution window and the required time resolution. Through the control execution window, the continuous control execution process is discretized into N control nodes, which can not only reduce the complexity of data processing, but also ensure the time resolution of sampling, and provide a time benchmark for analyzing the dynamic response performance of the electric drive controller.

[0019] After determining the control execution window, the vehicle controller sends the target control instruction and its corresponding control execution window to the electric drive controller. The electric drive controller performs specific control actions within the specified control execution window according to the received target control instruction to realize the driving control of the vehicle.

[0020] By receiving the target control command generated by the driver and determining the corresponding control execution window based on the command type and parameters, the time range and discrete sampling point regulations are provided for the execution of the electric drive controller, which is the basis for realizing the online test of the performance of the electric drive controller. At the same time, the size N of the control execution window can be adjusted according to actual needs to balance the calculation complexity and analysis accuracy.

[0021] S2: Execute the target control instruction through the electric drive controller, and collect the time series data of M control parameters in the control execution window, where M is a positive integer;

[0022] Specifically, after determining the control execution window, the electric drive controller starts to execute the received target control instruction. In the process of executing the target control instruction, the electric drive controller monitors and records the various parameters related to the execution of the target control instruction in real time according to the control execution window, and collects the time series data of M control parameters in the control execution window, such as motor speed, current, voltage, etc. M is a positive integer, which represents the number of selected control parameters, and its specific value can be determined according to the evaluation requirements and the complexity of the control system. The obtained time series data is a series of values ​​of the control parameters collected and recorded in time sequence within the control execution window. By obtaining the values ​​of the control parameters at different times, the dynamic process of the electric drive controller executing the target control instruction can be reflected. The frequency of collecting time series data is determined by the division of the control nodes, and each control node corresponds to a sampling time.

[0023] The acquisition of the timing data of the control parameters is achieved through the built-in sensors and data recording modules of the electric drive controller. These sensors monitor the changes of the control parameters in real time and transmit the acquired data to the data recording module for storage. The data recording module can be the memory inside the electric drive controller or an external data recording device connected to it.

[0024] By collecting the time series data of M control parameters in the control execution window, the original data reflecting the dynamic execution effect of the electric drive controller is obtained, forming the premise for testing the electric drive controller and laying the data foundation for subsequent control performance analysis.

[0025] S3: Building an instruction execution matrix based on the timing data and the control execution window, wherein in the instruction execution matrix, rows correspond to control parameters and columns correspond to control parameter values;

[0026] Specifically, after acquiring the timing data of M control parameters, the collected timing data and the information of the control execution window are fused to construct an instruction execution matrix to form a structured data representation for subsequent analysis and calculation.

[0027] The instruction execution matrix is ​​a two-dimensional array used to store and represent the parameter data of the target control instruction during execution. The number of rows in the matrix is ​​equal to the number of selected control parameters M, and each row corresponds to a control parameter; the number of columns in the matrix is ​​equal to the number of control nodes N in the control execution window, and each column corresponds to a control node, indicating the value of each control parameter at the control node.

[0028] First, according to the number of control parameters M, determine the number of rows of the matrix, and assign a row index to each control parameter in sequence. Then, according to the number of control nodes N of the control execution window, determine the number of columns of the matrix, and assign a column index to each control node in time sequence. Next, traverse the time series data and fill the values ​​of each control parameter at each sampling time into the corresponding position of the matrix. At this point, a complete instruction execution matrix is ​​obtained, in which each element represents the specific value of a control parameter at a certain control node.

[0029] By constructing the instruction execution matrix, the originally scattered time series data is integrated according to the two dimensions of parameters and time to form a structured data representation, which not only facilitates data storage and access, but also facilitates subsequent data analysis. By processing the rows and columns of the matrix separately, a comprehensive analysis of different control parameters and different time nodes can be achieved, thereby comprehensively evaluating the dynamic control performance of the electric drive controller.

[0030] In addition, the number of rows M and columns N of the instruction execution matrix can be adjusted according to actual needs. Increasing the number of rows can include more control parameters and improve the comprehensiveness of the analysis; increasing the number of columns can improve the time resolution and capture more subtle dynamic changes. By reasonably setting the values ​​of M and N, a balance can be achieved between analysis accuracy and computational efficiency.

[0031] S4: performing matrix longitudinal analysis according to the instruction execution matrix, respectively obtaining the matching degrees of multiple control parameter values ​​under N control nodes, and obtaining N different parameter matching analysis results;

[0032] Specifically, after constructing the instruction execution matrix, the matrix is ​​analyzed vertically, focusing on the matching relationship between different control parameter values ​​at each control node to evaluate the control performance of the electric drive controller at different time nodes. The vertical analysis refers to extracting the data at each control node column by column according to the column direction of the instruction execution matrix, and analyzing each column.

[0033] For each column of the instruction execution matrix, extract the values ​​of all control parameters at the control node corresponding to the column to form a parameter value vector. Then, calculate the matching degree between different control parameter values ​​in the parameter value vector to obtain a heterogeneous parameter matching analysis result, which reflects the coordinated control ability of the electric drive controller at a certain control node for different control parameters. Among them, the higher the matching degree, the better the control performance of the electric drive controller at each control node according to the requirements of the target control instruction; conversely, the lower the matching degree, the worse the control performance of the electric drive controller at the control node. The calculation of the matching degree depends on the characteristics of the control parameters and the analysis requirements. For example, calculate the Euclidean distance between the parameter value vector at each control node and the ideal parameter vector of the target control instruction at the control node, and then map the distance value to the matching degree. The smaller the distance, the higher the matching degree. Calculate the cosine value of the angle between the parameter value vector at each control node and the ideal parameter vector, and map the cosine value to the matching degree. The closer the cosine value is to 1, the higher the matching degree.

[0034] By performing a longitudinal analysis on the instruction execution matrix and calculating the matching degree between different control parameter values ​​at each control node, N different parameter matching analysis results are obtained, which reflect the parameter coordination control capability of the electric drive controller at different time nodes. This is used to evaluate the dynamic response characteristics and execution effect of the controller, and provide a basis for online testing of the performance of the electric drive controller.

[0035] S5: Perform matrix lateral analysis according to the instruction execution matrix to obtain the matching degree of each control parameter between adjacent control parameter values ​​under each control node, and obtain M parameter self-matching analysis results;

[0036] Specifically, while analyzing the instruction execution matrix vertically, we also analyze the instruction execution matrix horizontally, focusing on the value change trend of each control parameter between different control nodes. The horizontal analysis is to extract the value sequence of each control parameter at all control nodes row by row according to the row direction of the instruction execution matrix, and analyze each row.

[0037] For each row of the matrix, extract the value of the control parameter corresponding to the row at each control node to form a parameter value sequence. Then, calculate the matching degree between the parameter values ​​at adjacent control nodes in the parameter value sequence to obtain a parameter self-matching analysis result, which reflects the continuous control capability of the electric drive controller for a certain control parameter in the process of executing the target control instruction. Among them, the higher the matching degree, the more the electric drive controller can adjust the value of the control parameter smoothly and coherently according to the requirements of the target control instruction, achieve good control continuity, and have better control performance; conversely, the lower the matching degree, the more the electric drive controller has fluctuations or jumps when adjusting the control parameter, and the control continuity is poor. Among them, the calculation of the matching degree depends on the characteristics of the control parameter and the analysis requirements. For example, the deviation between the parameter value sequence and the ideal parameter value sequence is obtained to obtain the deviation value sequence, and then the mean of the deviation value is obtained. The inverse of the deviation mean is used as the matching degree. The smaller the mean, the higher the matching degree.

[0038] By performing a horizontal analysis on the instruction execution matrix and calculating the matching degree of each control parameter between adjacent control nodes, M parameter self-matching analysis results are obtained, which reflect the continuous regulation capability of the electric drive controller for different control parameters. This is used to evaluate the control stability of the controller and provide a basis for online testing of the performance of the electric drive controller.

[0039] S6: Integrate the N different-parameter matching analysis results and the M parameter self-matching analysis results to obtain online the control performance result of the electric drive controller executing the target control instruction.

[0040] Specifically, after completing the vertical and horizontal analysis of the instruction execution matrix, N heterogeneous parameter matching analysis results and M parameter self-matching analysis results are obtained. These two types of analysis results evaluate the dynamic control performance of the electric drive controller from two perspectives: the matching relationship between different control parameters and the continuous matching of a single control parameter in the time dimension.

[0041] In order to obtain the control performance results of the target control instructions, the two types of analysis results are integrated to comprehensively reflect the dynamic response capability, parameter coordination control capability, and control continuity of the electric drive controller in the process of executing the target control instructions. For example, the arithmetic mean of the N different parameter matching analysis results and the M parameter self-matching analysis results are taken respectively, and then the two averages are multiplied to obtain the control performance result; or the minimum value of the N different parameter matching analysis results and the M parameter self-matching analysis results is taken as the control performance result to reflect the minimum performance level of the electric drive controller in each evaluation dimension.

[0042] By integrating the results of heterogeneous parameter matching analysis and parameter self-matching analysis, comprehensive and objective control performance results are obtained, which not only takes into account the coordination between different control parameters, but also takes into account the continuity of a single control parameter in the time dimension, accurately reflects the comprehensive performance of the electric drive controller in the process of executing target control instructions, and realizes online testing of the electric drive controller performance.

[0043] By obtaining the control performance results online in real time during the actual execution of the control task by the electric drive controller, compared with offline evaluation, it is able to dynamically track the performance changes of the electric drive controller, timely discover and locate potential problems, and provide a basis for vehicle control optimization.

[0044] Furthermore, the embodiment of the present application also includes:

[0045] The instruction execution matrix is ​​divided into columns to obtain N matrix column parameter values, wherein each matrix column parameter value corresponds to a control node;

[0046] Build a heterogeneous parameter matching analyzer;

[0047] Traversing N matrix column parameter values, obtaining a first matrix column parameter value, combining the target control instruction with the input into a different parameter matching analyzer, and obtaining a first different parameter matching analysis result;

[0048] The first different-parameter matching analysis result is added to the N different-parameter matching analysis results.

[0049] In a feasible implementation, the obtained instruction execution matrix is ​​segmented along the column direction to obtain N independent matrix column parameter values ​​to provide support for heterogeneous parameter matching analysis. Wherein, each matrix column parameter value corresponds to a column in the instruction execution matrix, representing the values ​​of all control parameters at a specific control node. Then, a heterogeneous parameter matching analyzer is pre-constructed, which is an analysis module for evaluating the matching relationship between different control parameters at the same control node. The heterogeneous parameter matching analyzer can be based on a specific matching degree calculation method, such as a correlation coefficient method, an Euclidean distance method, etc., to quantify the matching degree between different parameter values, or it can be constructed using a machine learning method. By constructing a heterogeneous parameter matching analyzer, a unified computing framework and interface are provided for subsequent matching analysis to ensure the consistency and repeatability of the analysis process.

[0050] Then, traverse the N matrix column parameter values, select a matrix column parameter value from the N matrix column parameter values ​​as the first matrix column parameter value, corresponding to a column of the instruction execution matrix, indicating the value of all control parameters at a specific control node. Subsequently, the first matrix column parameter value is input into the heterogeneous parameter matching analyzer together with the obtained target control instruction. The heterogeneous parameter matching analyzer evaluates the degree of matching between different control parameter values ​​at the control node to obtain the first heterogeneous parameter matching analysis result. After obtaining the first heterogeneous parameter matching analysis result, it is added to a set for storing all heterogeneous parameter matching analysis results, which will eventually contain N heterogeneous parameter matching analysis results, corresponding to the N columns of the instruction execution matrix, that is, the execution effect of the electric drive controller at the N control nodes.

[0051] By performing a vertical analysis on the instruction execution matrix, based on operations such as splitting the matrix by columns, building a heterogeneous parameter matching analyzer, and traversing the matrix column parameter values, the matching degree of the control parameters of the electric drive controller at each control node is evaluated one by one, and a set of heterogeneous parameter matching analysis results reflecting the execution effect of the electric drive controller at different control nodes are obtained, providing reliable data support for accurate performance testing.

[0052] Furthermore, the embodiment of the present application also includes:

[0053] Retrieving historical instruction execution records based on the electric drive controller, clustering the historical instruction execution records based on the control instruction type to obtain multiple instruction execution record clusters, and performing instruction type identification on each instruction execution record cluster to obtain multiple instruction type identifications;

[0054] According to the plurality of instruction execution record clusters, a plurality of sample matrix column parameter value sets are obtained, and matching degree marking is performed on the sample matrix column parameter values ​​to obtain a plurality of sample different parameter matching degree sets;

[0055] Using a plurality of sample matrix column parameter value sets and the plurality of sample different parameter matching degree sets, a plurality of different parameter matching analysis channels are constructed, wherein each different parameter matching analysis channel has an instruction type identifier;

[0056] Multiple instruction type identifiers and multiple heterogeneous parameter matching analysis channels are fused in parallel to obtain a heterogeneous parameter matching analyzer.

[0057] In a preferred embodiment, in the process of constructing the heterogeneous parameter matching analyzer, machine learning technology is introduced to improve the intelligence level and analysis efficiency of the heterogeneous parameter matching analysis.

[0058] First, based on the electric drive controller, the historical instruction execution records of similar electric drive controllers are retrieved, covering the situations where the electric drive controller executes different types of control instructions under various working conditions. As a training data set for machine learning, the historical instruction execution records include key information such as the type of control instruction, execution time, and control parameter value. Then, clustering algorithms such as K-means, hierarchical clustering, and DBSCAN are used to cluster the training data set, and similar instruction execution records are classified into one category according to the type of control instruction to form multiple instruction execution record clusters. At the same time, the instruction type is identified for each instruction execution record cluster to indicate the control instruction type corresponding to the cluster, and multiple instruction type identifiers are obtained, such as acceleration instructions, deceleration instructions, and constant speed cruise instructions.

[0059] Then, the control parameter values ​​are extracted from each instruction execution record cluster to form a corresponding sample matrix column parameter value set as the input feature of the machine learning model. For each instruction execution record in the cluster, the numerical values ​​related to the control parameters, such as motor speed, current, voltage, etc., are extracted to form a sample matrix column parameter value. The matrix column parameter values ​​of all records in the cluster are combined to obtain a complete sample matrix column parameter value set, which reflects the actual value of the control parameters during the execution of a certain type of control instruction. At the same time, for each sample matrix column parameter value set, the matching degree is marked to obtain the corresponding sample different parameter matching degree set. Matching degree marking refers to calculating and marking the corresponding matching degree value according to the value of different parameters in the sample matrix column parameter value by means of correlation coefficient, Euclidean distance, etc.

[0060] Subsequently, for each instruction type identifier, a corresponding sample matrix column parameter value set and a sample heteroparameter matching degree set are used to train an independent machine learning model as a heteroparameter matching analysis channel. For example, a neural network, a support vector machine, a decision tree, etc. are used. Each heteroparameter matching analysis channel is specifically used to process a specific type of control instruction. By learning the relationship between the characteristics and matching degree in the sample data, a mapping function between the matrix column parameter value and the matching degree is established. Afterwards, the trained multiple heteroparameter matching analysis channels are connected in parallel according to their corresponding instruction type identifiers to form a complete heteroparameter matching analyzer. When the target control instruction and the first matrix column parameter value are input, the heteroparameter matching analyzer allocates the first matrix column parameter value to the corresponding heteroparameter matching analysis channel according to the instruction type identifier of the target control instruction for matching degree prediction, and obtains the corresponding heteroparameter matching analysis result.

[0061] Through the above steps, the construction of the heterogeneous parameter matching analyzer is realized. Make full use of the historical instruction execution records of the electric drive controller, and through clustering processing, matching degree annotation, parallel fusion and other operations, build multiple heterogeneous parameter matching analysis channels for different instruction types, and integrate them into a complete heterogeneous parameter matching analyzer, so that the heterogeneous parameter matching analyzer can more accurately and efficiently evaluate the heterogeneous parameter matching of the electric drive controller when executing different types of control instructions, providing reliable tool support for performance testing.

[0062] Furthermore, the embodiment of the present application also includes:

[0063] The instruction execution matrix is ​​divided into rows to obtain M matrix row parameter values, wherein each matrix row parameter value corresponds to a control parameter;

[0064] Construct parameter self-matching analyzer;

[0065] Traverse the M matrix row parameter values, obtain the first matrix row parameter value, copy and obtain three first matrix row parameter values, sort the three first matrix row parameter values ​​in a staggered manner according to one control node, and construct a first parameter input matrix;

[0066] Combining the target control instruction with the first parameter input matrix, inputting the matrix into a parameter self-matching analyzer, and outputting a first parameter self-matching analysis result;

[0067] The first parameter self-matching analysis result is added to the M parameter self-matching analysis results.

[0068] In a feasible implementation, the obtained instruction execution matrix is ​​segmented along the row direction to obtain M independent matrix row parameter values ​​to provide support for subsequent parameter self-matching analysis. Each matrix row parameter value corresponds to a row in the instruction execution matrix, indicating the value of a specific control parameter at all control nodes. Then, a parameter self-matching analyzer is pre-constructed, and the constructed parameter self-matching analyzer is an analysis module for evaluating the value matching relationship of a single control parameter between continuous control nodes. The parameter self-matching analyzer can be based on a specific matching degree calculation method, such as adjacent value difference method, slope change method, etc., to quantify the matching degree of control parameters between adjacent control nodes, or it can be constructed using a machine learning method. By constructing a parameter self-matching analyzer, a unified calculation framework and interface are provided for subsequent matching analysis to ensure the consistency and repeatability of the analysis process.

[0069] Subsequently, the M matrix row parameter values ​​are traversed, and a matrix row parameter value is selected from the M matrix row parameter values ​​as the first matrix row parameter value, corresponding to a row of the instruction execution matrix, indicating the value of the specific control parameter at all control nodes. Then, the three first matrix row parameter values ​​are copied and obtained, and they are dislocated and sorted according to the interval of a control node. Among them, the dislocation sorting refers to cyclically shifting the elements of the three first matrix row parameter values ​​to the left or right by one position, so that the elements of each row are staggered by one control node with the elements of its adjacent row. Then, the three first matrix row parameter values ​​after the dislocation sorting are combined into a first parameter input matrix, each column of which contains the values ​​of the control parameter at three consecutive control nodes, which is used for subsequent parameter self-matching analysis. Thereafter, the obtained target control instruction and the constructed first parameter input matrix are input into the parameter self-matching analyzer, and the parameter self-matching analyzer evaluates the matching of the control parameters in the first parameter input matrix between the consecutive control nodes, and obtains the analysis result reflecting the parameter self-matching situation, that is, the first parameter self-matching analysis result, which quantitatively describes the continuity and stability of the specific control parameter in the entire control process. After obtaining the first parameter self-matching analysis result, it is added to a set for storing all parameter self-matching analysis results. The set will contain M parameter self-matching analysis results, corresponding to the M rows of the instruction execution matrix, that is, the M control parameters of the electric drive controller.

[0070] Through horizontal analysis of the instruction execution matrix, according to operations such as splitting the matrix by rows, constructing a parameter self-matching analyzer, and constructing a parameter input matrix by staggered sorting, the matching degree of each control parameter between continuous control nodes is evaluated one by one, and a set of parameter self-matching analysis results reflecting the continuity and stability of each control parameter in the electric drive controller are obtained. The structural characteristics of the instruction execution matrix are fully utilized, and the changing rules of the control parameters in the time dimension are effectively captured through staggered sorting and continuity analysis, providing an important basis for the performance test of the electric drive controller.

[0071] Furthermore, the embodiment of the present application also includes:

[0072] The first matrix row parameter value is copied to obtain one first matrix row parameter value, two first matrix row parameter values ​​and three first matrix row parameter values ​​respectively;

[0073] Based on the two first matrix row parameter values, align the second control node of one first matrix row parameter value with the first control node of the two first matrix row parameter values, and align the first control node of the three first matrix row parameter values ​​with the second control node of the two first matrix row parameter values, to construct a first parameter initial input matrix;

[0074] The first column and the last column of the first parameter initial input matrix are deleted to obtain a first parameter input matrix.

[0075] In a feasible implementation, first, the obtained first matrix row parameter value is copied to obtain three copies of the matrix row parameter value with the same content, which are used as one first matrix row parameter value, two first matrix row parameter values ​​and three first matrix row parameter values ​​respectively. Then, as Figure 2 As shown, based on two copies of the first matrix row parameter values, one copy of the first matrix row parameter value and three copies of the first matrix row parameter value are staggered and sorted. Specifically, one copy of the first matrix row parameter value is cyclically moved rightward by one control node so that its second control node is aligned with the first control node of the two copies of the first matrix row parameter value; at the same time, the three copies of the first matrix row parameter value are cyclically moved leftward by one control node so that its first control node is aligned with the second control node of the two copies of the first matrix row parameter value. Through staggered sorting, a matrix consisting of three copies of the first matrix row parameter value is obtained, that is, the first parameter initial input matrix, and each column of the matrix contains the values ​​of the control parameter at three consecutive control nodes.

[0076] After obtaining the first parameter initial input matrix, such as Figure 2As shown, the first column and the last column of the first parameter initial input matrix are deleted to obtain a matrix of 3 rows and N-2 columns as the first parameter input matrix. By deleting the first and last columns, the data with only a single control parameter value due to cyclic shift is removed, so that each column in the first parameter input matrix contains the values ​​of the control parameter at least at two consecutive control nodes. In the process of constructing the first parameter input matrix, one copy of the first matrix row parameter value and three copies of the first matrix row parameter value are cyclically shifted, resulting in missing values ​​in the upper left corner and lower right corner of the first parameter input matrix. In order to ensure the processability of the first parameter input matrix, a special value, such as 0 or -1, is filled in the upper left corner and lower right corner of the first parameter initial input matrix, indicating that there is no valid control parameter value at this position. In the parameter self-matching analysis, the missing position is specially processed according to this special value, such as skipping the analysis or assigning a lower weight.

[0077] By staggering the row parameter values ​​of the first matrix and inputting the first parameter into the matrix, utilizing the structural characteristics of the matrix and through the reorganization and improvement of local data, a new matrix containing parameter continuity information is obtained, providing complete and effective data support for parameter self-matching analysis.

[0078] Furthermore, the embodiment of the present application also includes:

[0079] Retrieving historical instruction execution records based on the electric drive controller, clustering the historical instruction execution records based on the control instruction type to obtain multiple instruction execution record clusters, and performing instruction type identification on each instruction execution record cluster to obtain multiple instruction type identifications;

[0080] According to the plurality of instruction execution record clusters, a plurality of sample parameter input matrix sets are obtained, and matching degree annotations are performed on the sample parameter input matrices to obtain a plurality of sample parameter self-matching degree sets;

[0081] Constructing a plurality of parameter self-matching analysis channels using a plurality of sample parameter input matrix sets and the plurality of sample parameter self-matching degree sets, wherein each parameter self-matching analysis channel has an instruction type identifier;

[0082] Multiple instruction type identifiers and multiple parameter self-matching analysis channels are fused in parallel to obtain a parameter self-matching analyzer.

[0083] First, based on the electric drive controller, the historical instruction execution records of similar electric drive controllers are retrieved, covering the situations where the electric drive controller executes different types of control instructions under various working conditions. As a training data set for machine learning, the historical instruction execution records include key information such as the type of control instruction, execution time, and control parameter value. Then, clustering algorithms such as K-means, hierarchical clustering, and DBSCAN are used to cluster the training data set, and similar instruction execution records are classified into one category according to the type of control instruction to form multiple instruction execution record clusters. At the same time, the instruction type is identified for each instruction execution record cluster to indicate the control instruction type corresponding to the cluster, and multiple instruction type identifiers are obtained, such as acceleration instructions, deceleration instructions, and constant speed cruise instructions.

[0084] Then, for each instruction execution record cluster, the control parameter value sequence is extracted, and a sample parameter input matrix is ​​constructed by copying, dislocation sorting, and head-to-tail deletion of the parameter value sequence. This process is repeated until all instruction execution record clusters are processed, and multiple sample parameter input matrix sets equal to the number of clusters are obtained. Each sample parameter input matrix set corresponds to a control instruction type, reflecting the continuity characteristics of the control parameters of this type of instruction during execution. At the same time, the matrix in each sample parameter input matrix set is marked with matching degree. According to the similarity between the elements of each column in the matrix and the rational control, the corresponding matching degree value is calculated and attached, indicating the matching degree between the parameter changes between the continuous control nodes represented by the column and the ideal situation. After matching degree marking, multiple sample parameter self-matching degree sets equal to the number of sample parameter input matrix sets are obtained. Each sample parameter self-matching degree set contains the parameter matching degree values ​​of a type of control instruction between continuous control nodes.

[0085] Afterwards, for each instruction type, the corresponding sample parameter input matrix set is used as the input feature of the training data, and the corresponding sample parameter self-matching degree set is used as the label of the training data. The parameter self-matching analysis channel for the instruction type is trained, such as support vector machine, decision tree, neural network, etc., which can be used to construct the parameter self-matching analysis channel. During the training process, the machine learning model establishes a mapping function by learning the relationship between the parameter input matrix and the self-matching degree in the sample data, so as to predict the corresponding self-matching degree according to the new parameter input matrix. Through supervised learning, each parameter self-matching analysis channel can master the parameter self-matching law under a specific instruction type and realize the intelligent evaluation of the control parameters during the execution of this type of instruction. At the same time, each channel is given the same instruction type identifier as its corresponding instruction type, which is used to identify the control instruction type targeted by the channel. Subsequently, multiple parameter self-matching analysis channels are arranged in parallel according to their instruction type identifiers to form a complete parameter self-matching analyzer. In the parameter self-matching analyzer, each parameter self-matching analysis channel is independent of each other, but can work together. When multiple control instructions are involved, they are assigned to the corresponding parameter self-matching analysis channel for processing according to the instruction type identification of each control instruction. Each channel predicts the self-matching degree of the input parameter matrix and gives the evaluation result of the parameter control during the execution of the instruction.

[0086] Furthermore, the embodiment of the present application also includes:

[0087] Based on multiple cluster centers of multiple instruction execution record clusters, heterogeneous parameter matching constraints and parameter self-matching constraints of multiple control instruction types are respectively constructed;

[0088] Parse the target control instruction to obtain a first control instruction type, and obtain corresponding N different parameter matching constraints and M parameter self-matching constraints from different parameter matching constraints and parameter self-matching constraints based on the first control instruction type;

[0089] The N different parameter matching constraints and the M parameter self-matching constraints are used to judge the N different parameter matching analysis results and the M parameter self-matching analysis results, so as to obtain the control performance result of the electric drive controller executing the target control instruction.

[0090] In a feasible implementation, first, the cluster center of each cluster is obtained by using the obtained multiple instruction execution record clusters to reflect the typical characteristics of the control instruction execution process in the cluster. Then, based on the cluster center of each instruction execution record cluster, with the help of expert knowledge and ideal control conditions, the heterogeneous parameter matching constraints and parameter self-matching constraints of the corresponding control instruction type are constructed. Among them, the heterogeneous parameter matching constraints refer to the constraints and threshold ranges of the matching degree between different control parameters for each control instruction type based on the characteristics of its cluster center; the parameter self-matching constraints refer to the constraints and threshold ranges of the matching degree between a single control parameter between continuous control nodes for each control instruction type based on the characteristics of its cluster center. By analyzing the characteristics of the cluster center, the typical matching relationship between the control parameters and the control parameters between continuous nodes in the execution process of this type of instruction is extracted, and the corresponding constraints are set accordingly. Repeat the above process until all instruction execution record clusters are processed, and multiple groups of heterogeneous parameter matching constraints and parameter self-matching constraints equal to the number of clusters are obtained, and each group of constraints corresponds to a control instruction type.

[0091] Subsequently, the acquired target control instruction is parsed, and the control instruction type information therein is extracted to obtain the first control instruction type. The first control instruction type refers to the instruction type to which the target control instruction belongs, such as an acceleration instruction, a braking instruction, etc. Then, based on the first control instruction type, the corresponding constraint group is searched in the constructed multiple groups of heterogeneous parameter matching constraints and parameter self-matching constraints to obtain the constraint group corresponding to the first control instruction type. From the located constraint group, N heterogeneous parameter matching constraints and M parameter self-matching constraints are obtained, where N and M are the number of heterogeneous parameter matching analysis results and the number of parameter self-matching analysis results, respectively.

[0092] Then, for each heterogeneous parameter matching analysis result, it is compared with the corresponding heterogeneous parameter matching constraint to determine whether the result satisfies the constraint condition and falls within the constraint threshold range. If the constraint condition is satisfied, it indicates that at the control node, the coordinated control capability of the electric drive controller for different control parameters meets expectations; otherwise, it indicates that there is a deviation or abnormality in the heterogeneous parameter matching performance of the electric drive controller. At the same time, for each parameter self-matching analysis result, it is compared with the corresponding parameter self-matching constraint to determine whether the result satisfies the constraint condition and falls within the constraint threshold range. If the constraint condition is satisfied, it indicates that the electric drive controller has good continuity in regulating the control parameter between continuous control nodes; otherwise, it indicates that there is a deviation or abnormality in the parameter self-matching performance of the electric drive controller. By performing constraint judgment on all heterogeneous parameter matching analysis results and parameter self-matching analysis results, a set of Boolean values ​​or scoring results are obtained to reflect the control performance of the electric drive controller in the process of executing the target control instruction. Then, these judgment results are comprehensively evaluated, such as calculating the proportion of Boolean values, the average value of the scoring results, etc., to obtain quantitative control performance results, which characterize the overall effect of the electric drive controller in executing the target control instruction.

[0093] By making full use of historical instruction execution records for constraint construction, we obtain heterogeneous parameter matching constraints and parameter self-matching constraints for different control instruction types. These constraints are used to judge the real-time analysis results, and the online test and evaluation of the control performance of the electric drive controller in executing the target control instructions is realized. This effectively reveals the real performance of the electric drive controller in the actual execution of control instructions, thereby promptly discovering potential performance problems.

[0094] Furthermore, the embodiment of the present application also includes:

[0095] An evaluation cycle is set, and when the evaluation cycle is triggered, multiple control instructions and corresponding multiple control performance results within the collection cycle are collected;

[0096] Generating a periodic performance test report of the electric drive controller based on the multiple control instructions and the corresponding multiple control performance results;

[0097] The electric drive controller is optimized and managed according to the periodic performance test report.

[0098] In a preferred embodiment, first, a fixed time interval is set as an evaluation period, such as every hour, every day or every week. The evaluation period is determined according to the actual working characteristics and evaluation requirements of the electric drive controller. When the system time reaches the preset evaluation period trigger condition, all control instructions executed by the electric drive controller within the period and the corresponding control performance results are collected. The control instructions refer to various control requests input to the electric drive controller within the evaluation period; the control performance results refer to the performance results obtained by evaluating the execution effect of each control instruction within the period. By collecting the control instructions and performance result data within the evaluation period, a comprehensive and continuous execution record of the electric drive controller in actual work is obtained, providing data support for the generation of performance test reports.

[0099] Then, using the collected control instructions and performance result data within an evaluation cycle, a performance test report reflecting the overall working performance of the electric drive controller within the cycle is comprehensively generated. Among them, the cycle performance test report includes statistics on the execution of control instructions, such as the number, proportion, and time distribution of different types of control instructions; statistics on the distribution of control performance results, such as the number, proportion, and time trend of results of different performance levels; detailed analysis of the control instruction execution process, such as parameter matching, abnormal conditions, etc.; calculation of comprehensive performance indicators, such as average performance score, pass rate, and excellent rate; performance change trend analysis, such as comparison with historical cycles, analysis of reasons for performance improvement or decline, etc. By generating a cycle performance test report, the working status and execution effect of the electric drive controller within an evaluation cycle can be intuitively and comprehensively displayed, so as to gain an in-depth understanding of the actual performance of the electric drive controller and identify potential problems and optimization space.

[0100] Next, based on the generated periodic performance test report, the working status and execution performance of the electric drive controller are analyzed and evaluated, the existing problems and deficiencies are identified, and targeted optimization management measures are formulated and implemented accordingly. For example, parameter adjustment and optimization, such as resetting or tuning the key parameters of the electric drive controller according to the performance test results to improve the control accuracy and stability; algorithm improvement and upgrade, such as improving and optimizing the control algorithm of the electric drive controller for the problems found in the test, such as introducing new control strategies, optimizing algorithm processes, etc.; hardware maintenance and upgrade, such as troubleshooting and repairing the hardware bottlenecks or hidden dangers exposed in the test, and replacing or upgrading related hardware equipment when necessary. By optimizing the management of the electric drive controller according to the periodic performance test report, the control performance of the electric drive controller can be effectively improved, the accuracy, stability and efficiency of its execution of control instructions can be improved, and ultimately the working quality and reliability of the entire electric drive system can be improved.

[0101] By introducing evaluation cycles, periodic performance test reports and optimization management, a complete performance testing and optimization closed-loop mechanism has been established. Based on periodic performance tests, continuous monitoring and improvement of the performance of the electric drive controller is achieved, thereby improving the reliability of the vehicle's electric drive system control.

[0102] In summary, the performance testing method for an electric drive controller provided in the embodiments of the present application has the following technical effects:

[0103] Receive the target control instruction, and determine the control execution window based on the target control instruction, wherein the control execution window includes N control nodes, N is a positive integer, which specifies the time range and sampling nodes for subsequent data collection and analysis, and provides a benchmark for online testing of the performance of the electric drive controller. The target control instruction is executed by the electric drive controller, and the time series data of M control parameters in the control execution window are collected, wherein M is a positive integer, and the original data reflecting the dynamic execution effect of the electric drive controller is obtained, laying a data foundation for building an analysis matrix. The instruction execution matrix is ​​constructed based on the time series data and the control execution window, wherein in the instruction execution matrix, the rows correspond to the control parameters, and the columns correspond to the control parameter values. The original data is integrated from the two dimensions of parameters and time to form a data structure that is convenient for matching degree analysis. According to the instruction execution matrix, the matrix is ​​analyzed vertically, and the matching degrees of multiple control parameter values ​​under N control nodes are obtained respectively, and N heterogeneous parameter matching analysis results are obtained. The matching degrees of different control parameters in the same control node are evaluated, and the effect of the electric drive controller executing the control instruction at each time node is quantitatively characterized, reflecting the dynamic coordination ability of the electric drive controller for each control parameter. According to the instruction execution matrix, the matrix is ​​analyzed horizontally to obtain the matching degree of each control parameter between adjacent control parameter values ​​at each control node, and M parameter self-matching analysis results are obtained to evaluate the matching degree of each control parameter between adjacent control nodes, quantitatively characterize the relationship between the execution of control instructions by the electric drive controller in a continuous time period, and reflect the dynamic stability of the electric drive controller. Integrate the N heterogeneous parameter matching analysis results with the M parameter self-matching analysis results to obtain the control performance results of the electric drive controller executing the target control instruction online. By integrating the heterogeneous parameter matching degree obtained by the longitudinal analysis and the parameter self-matching degree obtained by the horizontal analysis, the control performance results of the electric drive controller are obtained, and the online test of the dynamic execution effect of the electric drive controller is realized.

[0104] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0105] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A performance test method for an electric drive controller, characterized in that: The method comprises: receiving a target control instruction, and determining a control execution window based on the target control instruction, wherein the control execution window includes N control nodes, where N is a positive integer; Executing the target control instruction through the electric drive controller, collecting the time series data of M control parameters in the control execution window, where M is a positive integer; Building an instruction execution matrix based on the timing data and the control execution window, wherein in the instruction execution matrix, rows correspond to control parameters and columns correspond to control parameter values; According to the instruction execution matrix, a matrix vertical analysis is performed to obtain the matching degree of multiple control parameter values ​​under N control nodes respectively, and N different parameter matching analysis results are obtained; According to the instruction execution matrix, a matrix horizontal analysis is performed to obtain the matching degree of each control parameter between adjacent control parameter values ​​under each control node, and M parameter self-matching analysis results are obtained; Integrate the N different parameter matching analysis results and the M parameter self-matching analysis results to obtain online the control performance result of the electric drive controller executing the target control instruction; According to the instruction execution matrix, a matrix vertical analysis is performed to obtain the matching degrees of multiple control parameter values ​​under N control nodes respectively, and N different parameter matching analysis results are obtained, including: The instruction execution matrix is ​​divided into columns to obtain N matrix column parameter values, wherein each matrix column parameter value corresponds to a control node; Build a heterogeneous parameter matching analyzer; Traversing N matrix column parameter values, obtaining a first matrix column parameter value, combining the target control instruction with the input into a different parameter matching analyzer, and obtaining a first different parameter matching analysis result; Adding the first different-parameter matching analysis result to the N different-parameter matching analysis results; Build a different parameter matching analyzer, including: Retrieving historical instruction execution records based on the electric drive controller, clustering the historical instruction execution records based on the control instruction type to obtain multiple instruction execution record clusters, and performing instruction type identification on each instruction execution record cluster to obtain multiple instruction type identifications; According to the plurality of instruction execution record clusters, a plurality of sample matrix column parameter value sets are obtained, and matching degree marking is performed on the sample matrix column parameter values ​​to obtain a plurality of sample different parameter matching degree sets; Using a plurality of sample matrix column parameter value sets and the plurality of sample different parameter matching degree sets, a plurality of different parameter matching analysis channels are constructed, wherein each different parameter matching analysis channel has an instruction type identifier; A plurality of instruction type identifiers and a plurality of heterogeneous parameter matching analysis channels are integrated in parallel to obtain a heterogeneous parameter matching analyzer; According to the instruction execution matrix, a matrix horizontal analysis is performed to obtain the matching degree of each control parameter between adjacent control parameter values ​​under each control node, and M parameter self-matching analysis results are obtained, including: The instruction execution matrix is ​​divided into rows to obtain M matrix row parameter values, wherein each matrix row parameter value corresponds to a control parameter; Construct parameter self-matching analyzer; Traverse the M matrix row parameter values, obtain the first matrix row parameter value, copy and obtain three first matrix row parameter values, sort the three first matrix row parameter values ​​in a staggered manner according to one control node, and construct a first parameter input matrix; Combining the target control instruction with the first parameter input matrix, inputting the matrix into a parameter self-matching analyzer, and outputting a first parameter self-matching analysis result; Adding the first parameter self-matching analysis result to the M parameter self-matching analysis results; Build a parameter self-matching analyzer, including: Retrieving historical instruction execution records based on the electric drive controller, clustering the historical instruction execution records based on the control instruction type to obtain multiple instruction execution record clusters, and performing instruction type identification on each instruction execution record cluster to obtain multiple instruction type identifications; According to the plurality of instruction execution record clusters, a plurality of sample parameter input matrix sets are obtained, and matching degree annotations are performed on the sample parameter input matrices to obtain a plurality of sample parameter self-matching degree sets; Constructing a plurality of parameter self-matching analysis channels using a plurality of sample parameter input matrix sets and the plurality of sample parameter self-matching degree sets, wherein each parameter self-matching analysis channel has an instruction type identifier; A plurality of instruction type identifiers and a plurality of parameter self-matching analysis channels are integrated in parallel to obtain a parameter self-matching analyzer; Integrating the N different parameter matching analysis results and the M parameter self-matching analysis results to obtain online the control performance results of the electric drive controller executing the target control instruction, including: Based on multiple cluster centers of multiple instruction execution record clusters, heterogeneous parameter matching constraints and parameter self-matching constraints of multiple control instruction types are respectively constructed; Parse the target control instruction to obtain a first control instruction type, and obtain corresponding N different parameter matching constraints and M parameter self-matching constraints from different parameter matching constraints and parameter self-matching constraints based on the first control instruction type; The N different parameter matching constraints and the M parameter self-matching constraints are used to judge the N different parameter matching analysis results and the M parameter self-matching analysis results, so as to obtain the control performance result of the electric drive controller executing the target control instruction.

2. The method according to claim 1, characterized in that Copy and obtain three first matrix row parameter values, sort the three first matrix row parameter values ​​in a staggered manner according to one control node, and construct a first parameter input matrix, including: The first matrix row parameter value is copied to obtain one first matrix row parameter value, two first matrix row parameter values ​​and three first matrix row parameter values ​​respectively; Based on the two first matrix row parameter values, align the second control node of one first matrix row parameter value with the first control node of the two first matrix row parameter values, and align the first control node of the three first matrix row parameter values ​​with the second control node of the two first matrix row parameter values, to construct a first parameter initial input matrix; The first column and the last column of the first parameter initial input matrix are deleted to obtain a first parameter input matrix.

3. The method according to claim 1, characterized in that The method further comprises: An evaluation cycle is set, and when the evaluation cycle is triggered, multiple control instructions and corresponding multiple control performance results within the collection cycle are collected; Generating a periodic performance test report of the electric drive controller based on the multiple control instructions and the corresponding multiple control performance results; The electric drive controller is optimized and managed according to the periodic performance test report.

Images