Automobile suspension strength testing method, device, equipment, medium and product

Abstract

The invention discloses an automotive suspension strength testing method, device and equipment, a medium and a product, and relates to the technical field of vehicle testing. The method comprises the following steps: firstly, acquiring multi-dimensional time sequence sensing data acquired by a plurality of sensors in the process of carrying out multiple automobile suspension strength test experiments on a target automobile suspension by adopting a drop method, and then carrying out normalization processing on the multi-dimensional time sequence sensing data to obtain two-dimensional normalized matrix data; and finally, de-noising the two-dimensional normalized matrix data by adopting an MC-WNNM algorithm and / or a WSNM algorithm to obtain new two-dimensional normalized matrix data, importing the new two-dimensional normalized matrix data into an automotive suspension strength estimation model which completes pre-training based on a long short-term memory neural network, and outputting to obtain a strength estimation value of a target automotive suspension. Therefore, the purpose of ensuring the accuracy of the strength test result of the automotive suspension by performing non-elimination preprocessing on the obtained sensing data can be achieved, and the method is convenient for practical application and popularization.

Description

Technical Field

[0001] This invention belongs to the field of vehicle testing technology, specifically relating to a method, apparatus, equipment, medium, and product for testing the strength of automobile suspension. Background Technology

[0002] The automotive suspension is a force-transmitting component that connects the vehicle frame (or body) to the axles (or wheels), and it is also a crucial part of ensuring vehicle driving safety. The materials used in the automotive suspension structure have a significant impact on the vehicle's performance and safety; therefore, it is necessary to conduct specialized performance evaluations during the vehicle design and manufacturing process.

[0003] Currently, the commonly used method for evaluating automotive suspension performance is the drop test. This method simulates a vehicle's fall under different road conditions to assess the strength and durability of the suspension structure, providing important reference data for automakers. Specifically, during the automotive suspension strength test, testers drop the vehicle freely from a certain height, observing and recording the deformation and degree of damage to the suspension structure to evaluate its performance.

[0004] During automotive suspension strength testing, if the sensors used for parameter measurement lack accuracy, are improperly installed, or the data acquisition system malfunctions, the collected data may be inaccurate, affecting the judgment of the automotive suspension structure's response characteristics. Therefore, preprocessing of the obtained sensor data is necessary, such as removing outliers. However, when using the existing LOF algorithm (Local Outlier Factor, a density-based unsupervised anomaly detection algorithm that identifies outliers by comparing the local density differences between data points and their neighborhoods) for anomaly detection, the data parameters of the automotive suspension change as the test progresses. This means that the anomaly detection algorithm cannot effectively obtain anomaly detection results, and the remaining outlier data can negatively impact the automotive suspension strength test.

[0005] To address the aforementioned issues, existing patent CN118332473A provides an artificial intelligence-based method for testing automotive suspension strength. This method first obtains indicators such as outliers, influence differences, and correlation anomaly indices based on the correlation between sensor parameters. Then, it removes outlier data based on these indicators. Finally, the anomaly-free sensor data is input into a neural network to complete the suspension strength test. However, since the quantity and location of outlier data are uncertain, removing too much outlier data can affect the completeness of the input items to the suspension strength prediction model and the accuracy of the automotive suspension strength test results.

[0006] Therefore, how to provide a new solution for automobile suspension strength testing that can ensure the accuracy of automobile suspension strength test results by performing non-elimination preprocessing on the obtained sensor data is a topic that urgently needs to be studied by those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for testing the strength of automobile suspension, in order to solve the problem that existing automobile suspension strength testing schemes may affect the accuracy of automobile suspension strength test results due to the removal of too much abnormal data.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] Firstly, a method for testing the strength of automotive suspension is provided, including:

[0010] During multiple vehicle suspension strength tests conducted using the drop method on the target vehicle suspension, multidimensional time-series sensing data collected by multiple sensors is obtained. The multidimensional time-series sensing data corresponds one-to-one with the multiple sensors, and each time-series sensing data package contains multiple sensing values ​​that correspond one-to-one with the multiple vehicle suspension strength tests and are arranged in the order of the test time.

[0011] The multidimensional time-series sensing data is normalized to obtain two-dimensional normalized matrix data, wherein the two-dimensional normalized matrix data contains R×Q elements, where R represents the total number of the multiple sensors and Q represents the total number of the multiple vehicle suspension strength test experiments.

[0012] The two-dimensional normalized matrix data is denoised using the MC-WNNM algorithm and / or the WSNM algorithm to obtain new two-dimensional normalized matrix data.

[0013] The new two-dimensional normalized matrix data is imported into a pre-trained automotive suspension strength estimation model based on a long short-term memory neural network, and the strength estimate of the target automotive suspension is output.

[0014] Based on the above-mentioned invention, a novel scheme is provided for preprocessing the obtained sensor data using image denoising principles without removing data, and then estimating the vehicle suspension strength based on the processing results. Specifically, multidimensional time-series sensor data collected by multiple sensors during multiple vehicle suspension strength tests using the drop method on the target vehicle suspension is first acquired. Then, the multidimensional time-series sensor data is normalized to obtain a two-dimensional normalized matrix. Finally, the MC-WNNM algorithm and / or WSNM algorithm are used to denoise the two-dimensional normalized matrix data, resulting in new two-dimensional normalized matrix data, which is then imported into a long-term and short-term... The memory neural network completes the pre-trained car suspension strength estimation model and outputs the strength estimate of the target car suspension. In this way, abnormal sensor data is corrected and restored to true sensor data through image denoising. This can eliminate the adverse effects of abnormal data on the car suspension strength test results to a certain extent. Since there is no need to remove the obtained sensor data, it can also avoid the situation where removing too much abnormal data affects the accuracy of the car suspension strength test results. Thus, it can achieve the goal of ensuring the accuracy of car suspension strength test results even by performing non-removal preprocessing on the obtained sensor data, which is convenient for practical application and promotion.

[0015] In one possible design, the plurality of sensors include a pressure sensor mounted in the shock absorber for measuring the pressure of the fluid during shock absorption, a force sensor mounted at the suspension connection point for measuring the force acting on the suspension, a displacement sensor mounted at the spring for measuring the degree of spring compression, and / or a steering angle sensor mounted at the steering wheel for measuring the angle change of the steering wheel.

[0016] In one possible design, the MC-WNNM algorithm and the WSNM algorithm are used to denoise the two-dimensional normalized matrix data to obtain new two-dimensional normalized matrix data, including the following steps S31 to S38:

[0017] S31. Select the k-th two-dimensional matrix sub-block in the time dimension from the two-dimensional normalized matrix data using a sliding window method, and then execute step S32, where k represents a positive integer with an initial value of 1, and the k-th two-dimensional matrix sub-block contains... One element, Represents a positive integer less than Q;

[0018] S32. Search for M similar two-dimensional matrix sub-blocks that are most similar to the k-th two-dimensional matrix sub-block within the predefined neighborhood of the k-th two-dimensional matrix sub-block from the two-dimensional normalized matrix data, and then execute step S33, where M represents a positive integer;

[0019] S33. The k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks are respectively vectorized and straightened to obtain M+1 one-dimensional vectors. These M+1 one-dimensional vectors are then superimposed to obtain a vector of size [missing value]. The original matrix Y is obtained, and then step S34 is executed;

[0020] S34. Assume the original matrix Y = X + N + S, and use the ADMM algorithm to solve the minimization objective function to obtain the denoised matrix X. Then, execute step S36, wherein the minimization objective function is expressed as follows:

[0021]

[0022] In the formula, N represents Gaussian noise, S represents sparse noise, ||S||1 represents the L1 norm used to constrain the sparse noise S, and ||N|| F This represents the F-norm used to constrain Gaussian noise N. λS represents the Schattenp-norm used to constrain the denoising matrix X, λ1, λ2, and λ3 represent preset regularization coefficients, W represents a multi-channel weighting matrix in diagonal form, r represents a positive integer less than or equal to R, and δ r Let I represent the noise standard deviation of the r-th channel, I represent the identity matrix, min() represent the minimum function, i represent a positive integer, and w represent the minimum value. i Indicates assigning σ i The weights, σ i represents the i-th singular value in the denoising matrix X, p represents the parameter value used to determine the norm type and takes a value in the interval (0,1], and w represents the weighting of the nuclear norm;

[0023] S35. Based on the denoising matrix X, reconstruct a new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block and M new two-dimensional matrix similar sub-blocks corresponding one-to-one with the M two-dimensional matrix similar sub-blocks, and then execute step S36;

[0024] S36. In the two-dimensional normalized matrix data, the k-th two-dimensional matrix sub-block is replaced with the new two-dimensional matrix sub-block, and the M two-dimensional matrix similar sub-blocks are replaced one-to-one with the M new two-dimensional matrix similar sub-blocks to obtain new two-dimensional normalized matrix data, and then step S37 is executed.

[0025] S37. Determine whether the new two-dimensional matrix sub-block is the last two-dimensional matrix sub-block in the time dimension in the new two-dimensional normalized matrix data. If so, end the denoising process; otherwise, proceed to step S38.

[0026] S38. Increment k by 1, and then continue to use the sliding window method to select the kth two-dimensional matrix sub-block in the time dimension from the new two-dimensional normalized matrix data, and return to execute step S32.

[0027] In one possible design, M similar two-dimensional matrix sub-blocks are searched from the two-dimensional normalized matrix data, located within a predefined neighborhood of the k-th two-dimensional matrix sub-block and most similar to the k-th two-dimensional matrix sub-block, including:

[0028] For each other two-dimensional matrix sub-block that is located in the predefined neighborhood of the k-th two-dimensional matrix sub-block and has the same size as the k-th two-dimensional matrix sub-block in the two-dimensional normalized matrix data, the Eulerian distance between the k-th two-dimensional matrix sub-block and the corresponding sub-block is calculated.

[0029] Arrange the other two-dimensional matrix sub-blocks in order of Eulerian distance from nearest to farthest to obtain a sequence of two-dimensional matrix sub-blocks;

[0030] The first M two-dimensional matrix sub-blocks are selected from the sequence of two-dimensional matrix sub-blocks as the M two-dimensional matrix similar sub-blocks most similar to the k-th two-dimensional matrix sub-block, where M represents a positive integer.

[0031] In one possible design, the k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks are vectorized and straightened to obtain M+1 one-dimensional vectors. These M+1 one-dimensional vectors are then superimposed to obtain a result of size [missing value]. The original matrix Y includes:

[0032] For each sub-block in the k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks, the element located in the r′-th row and q′-th column of the corresponding sub-block is taken as the h′-th element in the corresponding one-dimensional vector, resulting in the corresponding one-dimensional vector, where r′ represents a positive integer less than or equal to R, and q′ represents a positive integer less than or equal to R. positive integers,

[0033] The m-th one-dimensional vector among the M+1 one-dimensional vectors that correspond one-to-one with the k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks will be used as the vector of size m. The original matrix Y is obtained by taking the element in the m-th row of the original matrix Y, where m represents a positive integer less than or equal to M+1.

[0034] In one possible design, based on the denoising matrix X, a new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block and M new two-dimensional matrix similar sub-blocks corresponding one-to-one with the M two-dimensional matrix similar sub-blocks are reconstructed, including:

[0035] If the element in the m-th row of the original matrix Y is a one-dimensional vector of the k-th two-dimensional matrix sub-block, then the element located in the m-th row and h′-th column of the noise-reducing matrix X will be used as the element located in the r′-th row and q′-th column of the new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block.

[0036] If the element in the m-th row of the original matrix Y is a one-dimensional vector of the m′-th two-dimensional matrix similarity sub-block among the M two-dimensional matrix similarity sub-blocks, then the element located in the m-th row and h′-th column of the denoising matrix X is taken as the element located in the r′-th row and q′-th column of the new two-dimensional matrix similarity sub-block corresponding to the m′-th two-dimensional matrix similarity sub-block, where m′ represents a positive integer less than or equal to M.

[0037] In a second aspect, an automotive suspension strength testing device is provided, comprising a sensor data acquisition unit, a normalization processing unit, a data denoising processing unit, and a suspension strength estimation unit that are sequentially connected in communication.

[0038] The sensing data acquisition unit is used to acquire multi-dimensional time-series sensing data collected by multiple sensors during multiple vehicle suspension strength test experiments using the drop method on the target vehicle suspension. The multi-dimensional time-series sensing data corresponds one-to-one with the multiple sensors, and each time-series sensing data package contains multiple sensing values ​​that correspond one-to-one with the multiple vehicle suspension strength test experiments and are arranged in the order of the test time.

[0039] The normalization processing unit is used to normalize the multidimensional time-series sensing data to obtain two-dimensional normalized matrix data, wherein the two-dimensional normalized matrix data contains R×Q elements, where R represents the total number of the multiple sensors and Q represents the total number of the multiple vehicle suspension strength test experiments.

[0040] The data denoising processing unit is used to denoise the two-dimensional normalized matrix data using the MC-WNNM algorithm and / or the WSNM algorithm to obtain new two-dimensional normalized matrix data.

[0041] The suspension strength estimation unit is used to import the new two-dimensional normalized matrix data into a pre-trained automotive suspension strength estimation model based on a long short-term memory neural network, and output the strength estimate of the target automotive suspension.

[0042] Thirdly, the present invention provides a computer device comprising a storage module, a processing module, and a transceiver module connected in sequence for communication, wherein the storage module is used to store a computer program, the transceiver module is used to send and receive messages, and the processing module is used to read the computer program and execute the automobile suspension strength testing method as described in the first aspect or any possible design in the first aspect.

[0043] Fourthly, the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, perform the automobile suspension strength testing method as described in the first aspect or any possible design of the first aspect.

[0044] Fifthly, the present invention provides a computer program product, including a computer program or instructions, which, when executed by a computer, implement the automobile suspension strength testing method as described in the first aspect or any possible design in the first aspect.

[0045] The beneficial effects of the above scheme are:

[0046] (1) This invention creatively provides a novel scheme for preprocessing the obtained sensor data using the principle of image denoising without removing data, and then estimating the strength of the vehicle suspension based on the processing results. Specifically, it first acquires multidimensional time-series sensor data collected by multiple sensors during multiple vehicle suspension strength tests conducted using the drop method on the target vehicle suspension. Then, it normalizes the multidimensional time-series sensor data to obtain two-dimensional normalized matrix data. Finally, it uses the MC-WNNM algorithm and / or WSNM algorithm to denoise the two-dimensional normalized matrix data, obtaining new two-dimensional normalized matrix data, which is then imported into a long-term and short-term model. The memory neural network completes the pre-trained car suspension strength estimation model and outputs the strength estimate of the target car suspension. In this way, abnormal sensor data is corrected and restored to true sensor data through image denoising. This can eliminate the adverse effects of abnormal data on the car suspension strength test results to a certain extent. Since there is no need to remove the obtained sensor data, it can also avoid the situation where removing too much abnormal data affects the accuracy of the car suspension strength test results. Thus, it can achieve the goal of ensuring the accuracy of car suspension strength test results even by performing non-removal preprocessing on the obtained sensor data, which is convenient for practical application and promotion. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a flowchart illustrating the automobile suspension strength testing method provided in an embodiment of this application.

[0049] Figure 2 This is a schematic diagram of the structure of the automobile suspension strength testing device provided in the embodiments of this application.

[0050] Figure 3 A schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in conjunction with the accompanying drawings and descriptions of the embodiments or the prior art. Obviously, the following description of the structure of the accompanying drawings is only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these embodiments without creative effort. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0052] It should be understood that although the terms "first" and "second", etc., may be used herein to describe various objects, these objects should not be limited by these terms. These terms are only used to distinguish one object from another. For example, the first object may be referred to as the second object, and similarly, the second object may be referred to as the first object, without departing from the scope of the exemplary embodiments of the invention.

[0053] It should be understood that the term "and / or" that may appear in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, or A and B exist simultaneously. Another example is A, B and / or C, which can mean that any one of A, B, and C or any combination thereof exists. The term " / and" that may appear in this document describes another relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone or A and B exist simultaneously. In addition, the character " / " that may appear in this document generally indicates that the related objects before and after it are in an "or" relationship.

[0054] Example

[0055] like Figure 1As shown, the vehicle suspension strength testing method provided in the first aspect of this embodiment can be executed, but is not limited to, by a computer device with certain computing resources and a communication connection to sensors used for measuring vehicle suspension strength testing parameters. For example, it can be executed by an electronic device such as a host computer, cloud server, edge computer, personal computer (PC, referring to a multi-purpose computer of size, price, and performance suitable for personal use; desktop computers, laptops, mini-laptops, tablets, and ultrabooks are all considered personal computers), smartphone, personal digital assistant (PDA), or wearable device. Figure 1 As shown, the automobile suspension strength test method includes, but is not limited to, the following steps S1 to S4.

[0056] S1. Acquire multi-dimensional time-series sensing data collected by multiple sensors during multiple vehicle suspension strength tests using the drop method on the target vehicle suspension. The multi-dimensional time-series sensing data corresponds one-to-one with the multiple sensors, and each dimension of the time-series sensing data includes, but is not limited to, multiple sensing values ​​that correspond one-to-one with the multiple vehicle suspension strength tests and are arranged sequentially according to the test time sequence.

[0057] In step S1, since the drop test is an existing method for evaluating automotive suspension performance, the specific process of the automotive suspension strength test can be derived conventionally based on existing technology, and will not be elaborated here. Specifically, the multiple sensors include, but are not limited to, pressure sensors installed in the shock absorber to measure the pressure of the fluid during shock absorption, force sensors installed at the suspension connection points to measure the force acting on the suspension, displacement sensors installed at the spring to measure the degree of spring compression, and / or angle sensors installed at the steering wheel to measure the angle change of the steering wheel, etc.; the aforementioned pressure sensors, force sensors, displacement sensors, and angle sensors can all be implemented using existing related sensor products; the multidimensional time-series sensing data specifically includes pressure time-series data collected by the pressure sensor, force time-series data collected by the force sensor, spring compression time-series data collected by the displacement sensor, and / or angle time-series data collected by the angle sensor, etc. The target vehicle suspension is the test object. Since the drop test typically requires hundreds of trials, such as 500, the pressure time-series data will contain 500 pressure values ​​corresponding to each of the 500 vehicle suspension strength tests, arranged chronologically. Similarly, the load-bearing capacity time-series data will contain 500 load-bearing capacity values, spring compression time-series data, and angle time-series data. Furthermore, this multi-dimensional time-series sensing data can be transmitted to a local device via conventional wireless communication technology to complete data acquisition.

[0058] S2. Normalize the multidimensional time-series sensing data to obtain two-dimensional normalized matrix data, wherein the two-dimensional normalized matrix data contains R×Q elements, where R represents the total number of the multiple sensors and Q represents the total number of the multiple vehicle suspension strength test experiments.

[0059] In step S2, the normalization process is performed using existing technologies, such as Min-Max normalization, Z-Score normalization, or nonlinear normalization. Specifically, when the plurality of sensors includes the pressure sensor, the force sensor, the displacement sensor, and the rotation angle sensor, R is set to 4, and Q can be, for example, 500. Thus, the two-dimensional normalized matrix data A can be represented as follows:

[0060]

[0061] In the formula, q represents a positive integer less than or equal to Q, and A 1,q A represents the normalized pressure value corresponding to the q-th automobile suspension strength test experiment. 2,q A represents the normalized value of the bearing force corresponding to the q-th automobile suspension strength test experiment. 3,q A represents the normalized value of spring compression corresponding to the q-th automobile suspension strength test experiment. 4,q This represents the angle normalization value corresponding to the q-th automobile suspension strength test experiment.

[0062] S3. The two-dimensional normalized matrix data is denoised using the MC-WNNM algorithm and / or the WSNM algorithm to obtain new two-dimensional normalized matrix data.

[0063] In step S3, the MC-WNNM (Multi-Channel Weighted Nuclear Norm Minimization) algorithm is an efficient existing image denoising method, particularly suitable for color and energy spectrum images. This algorithm performs sparse representation and recovery based on the nuclear norm of the matrix, optimizing the nuclear norm of different channels through weighting factors to achieve the best denoising effect. The WSNM (Weighted Singular Value Minimization) algorithm is another existing method for image denoising. Based on Singular Value Decomposition (SVD) technology, it can effectively remove noise from images while preserving important image features such as edges and textures. Since the two-dimensional normalized matrix data is a two-dimensional matrix containing R×Q elements, it can be regarded as a two-dimensional image containing R×Q pixels. Furthermore, since abnormal sensor data caused by factors such as insufficient accuracy of the multiple sensors, improper installation positions, or malfunctions in the data acquisition system can be considered as raw measurement data with added noise, the MC-WNNM algorithm and / or the WSNM algorithm can be applied to denoise the two-dimensional normalized matrix data to obtain new two-dimensional normalized matrix data that corrects and recovers the abnormal sensor data (i.e., by denoising the original measurement data, it is restored to true measurement data), enabling accurate estimation of the vehicle suspension strength. The specific process of the denoising process can be conventionally derived based on the MC-WNNM algorithm and / or the WSNM algorithm, and will not be elaborated here.

[0064] In step S3, in order to achieve better denoising effect by combining the MC-WNNM algorithm and the WSNM algorithm (i.e., using both the weighted moment W in the MC-WNNM algorithm and the Schatten p-norm in the WSNM algorithm to obtain the new two-dimensional normalized matrix data that is closer to the real measurement data), preferably, the MC-WNNM algorithm and the WSNM algorithm are used to denoise the two-dimensional normalized matrix data to obtain the new two-dimensional normalized matrix data, including but not limited to the following steps S31 to S38.

[0065] S31. Select the k-th two-dimensional matrix sub-block in the time dimension from the two-dimensional normalized matrix data using a sliding window method, and then execute step S32, where k represents a positive integer with an initial value of 1, and the k-th two-dimensional matrix sub-block contains... One element, It represents a positive integer less than Q.

[0066] In step S31, the sliding window method is an existing method for dividing time series data into multiple segments. For example, when the overlap rate is 0, the two-dimensional normalized matrix data A can be divided into multiple data segments A1, A2, ..., A j ,…:

[0067]

[0068] In the formula, j represents a positive integer, and A j This represents the j-th data segment in the time-series dimension obtained by dividing the two-dimensional normalized matrix data. Therefore, when k is initially 1, the k-th two-dimensional matrix sub-block is data segment A1. The value is 4.

[0069] S32. Search for M similar two-dimensional matrix sub-blocks in the two-dimensional normalized matrix data that are located in the predefined neighborhood of the k-th two-dimensional matrix sub-block and are most similar to the k-th two-dimensional matrix sub-block, and then execute step S33, where M represents a positive integer.

[0070] In step S32, the k-th two-dimensional matrix sub-block is a two-dimensional matrix, and the two-dimensional matrix similar sub-block is another two-dimensional matrix of the same size as the k-th two-dimensional matrix sub-block. For example, if A2 is the k-th two-dimensional matrix sub-block, then the M two-dimensional matrix similar sub-blocks may include, but are not limited to, A1, etc. The size of the predefined neighborhood can be specified by the tester or determined conventionally based on the results of multiple limited experiments. Specifically, searching from the two-dimensional normalized matrix data for the M two-dimensional matrix similar sub-blocks that are most similar to the k-th two-dimensional matrix sub-block within the predefined neighborhood of the k-th two-dimensional matrix sub-block includes, but is not limited to, the following steps S321 to S323.

[0071] S321. For each other two-dimensional matrix sub-block that is located in the predefined neighborhood of the k-th two-dimensional matrix sub-block and has the same size as the k-th two-dimensional matrix sub-block in the two-dimensional normalized matrix data, calculate the Eulerian distance between the k-th two-dimensional matrix sub-block and the corresponding sub-block.

[0072] In step S321, since the k-th two-dimensional matrix sub-block and the other two-dimensional matrix sub-blocks can all be regarded as two-dimensional vectors (i.e., all contain...), Since there are 10 elements, the Eulerian distance between the two can be calculated based on the existing Eulerian distance formula.

[0073] S322. Arrange the other two-dimensional matrix sub-blocks in order of Eulerian distance from nearest to farthest to obtain a sequence of two-dimensional matrix sub-blocks.

[0074] S323. Select the first M two-dimensional matrix sub-blocks from the sequence of two-dimensional matrix sub-blocks as the M two-dimensional matrix similar sub-blocks most similar to the k-th two-dimensional matrix sub-block, where M represents a positive integer.

[0075] S33. The k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks are respectively vectorized and straightened to obtain M+1 one-dimensional vectors. These M+1 one-dimensional vectors are then superimposed to obtain a vector of size [missing value]. The original matrix Y is obtained, and then step S34 is executed.

[0076] In step S34, the vectorization straightening process refers to converting the current two-dimensional vector of the corresponding sub-block into a one-dimensional vector. Specifically, the vectorization straightening process is performed on the k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks respectively, resulting in M+1 one-dimensional vectors. These M+1 one-dimensional vectors are then superimposed to obtain a vector of size [missing value]. The original matrix Y includes, but is not limited to, the following steps S331 to S332.

[0077] S331. For each sub-block in the k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks, the element located in the r′-th row and q′-th column of the corresponding sub-block is taken as the h′-th element in the corresponding one-dimensional vector to obtain the corresponding one-dimensional vector, where r′ represents a positive integer less than or equal to R, and q′ represents a positive integer less than or equal to R. positive integers,

[0078] S332. Take the m-th one-dimensional vector from the M+1 one-dimensional vectors that correspond one-to-one with the k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks, and use it as a vector of size... The original matrix Y is obtained by taking the element in the m-th row of the original matrix Y, where m represents a positive integer less than or equal to M+1.

[0079] S34. Assume the original matrix Y = X + N + S, and use the ADMM algorithm to solve the minimization objective function to obtain the denoised matrix X. Then, execute step S36, wherein the minimization objective function is expressed as follows:

[0080]

[0081] In the formula, N represents Gaussian noise, S represents sparse noise, ||S||1 represents the L1 norm used to constrain the sparse noise S, and ||N|| F This represents the F-norm used to constrain the Gaussian noise N. λS represents the Schattenp-norm used to constrain the denoising matrix X, λ1, λ2, and λ3 represent preset regularization coefficients, W represents a multi-channel weighting matrix in diagonal form, r represents a positive integer less than or equal to R, and δ r Let I represent the noise standard deviation of the r-th channel, I represent the identity matrix, min() represent the minimum function, i represent a positive integer, and w represent the minimum value. i Indicates assigning σ i The weights, σ i Let represent the i-th singular value in the denoising matrix X, p represent the parameter value used to determine the norm type and take values ​​in the interval (0,1], and w represent the weighted nuclear norm.

[0082] In step S34, the denoised matrix X represents the potential true measurement data (i.e., clean, noise-free matrix data that meets the requirements). The L1 norm, F norm, and Schatten p-norm are all existing norms. Furthermore, the ADMM algorithm (Alternating Direction Method of Multipliers) is a core existing algorithm in the field of mathematical optimization for handling separable block convex optimization problems. It can decompose large-scale problems into subproblems through a decomposition strategy and solve them alternately. Therefore, the specific solution process for minimizing the objective function can be derived conventionally based on the existing ADMM algorithm and will not be elaborated here.

[0083] S35. Based on the denoising matrix X, reconstruct a new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block and M new two-dimensional matrix similar sub-blocks corresponding one-to-one with the M two-dimensional matrix similar sub-blocks, and then execute step S36.

[0084] In step S35, the reconstruction process of the new two-dimensional matrix sub-block and the M new two-dimensional matrix similar sub-blocks is the inverse process of the aforementioned step S33. Specifically, based on the denoising matrix X, a new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block and M new two-dimensional matrix similar sub-blocks corresponding one-to-one with the M two-dimensional matrix similar sub-blocks are reconstructed, including but not limited to: if the element in the m-th row of the original matrix Y is a one-dimensional vector of the k-th two-dimensional matrix sub-block, then the element located in the m-th row of the denoising matrix X and The element in column h′ is taken as the element in row r′ and column q′ of the new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block; if the element in row m of the original matrix Y is a one-dimensional vector of the m′-th two-dimensional matrix similar sub-block among the M two-dimensional matrix similar sub-blocks, then the element in row m and column h′ of the denoising matrix X is taken as the element in row r′ and column q′ of the new two-dimensional matrix similar sub-block corresponding to the m′-th two-dimensional matrix similar sub-block, where m′ represents a positive integer less than or equal to M.

[0085] S36. In the two-dimensional normalized matrix data, the k-th two-dimensional matrix sub-block is replaced with the new two-dimensional matrix sub-block, and the M two-dimensional matrix similar sub-blocks are replaced one-to-one with the M new two-dimensional matrix similar sub-blocks to obtain new two-dimensional normalized matrix data, and then step S37 is executed.

[0086] S37. Determine whether the new two-dimensional matrix sub-block is the last two-dimensional matrix sub-block in the time dimension of the new two-dimensional normalized matrix data. If so, end the denoising process; otherwise, proceed to step S38.

[0087] S38. Increment k by 1, and then continue to use the sliding window method to select the kth two-dimensional matrix sub-block in the time dimension from the new two-dimensional normalized matrix data, and return to execute step S32.

[0088] Steps S31 to S36 described above constitute the process of locally denoising and updating the two-dimensional normalized matrix data. Therefore, through subsequent steps S37 to S38, the two-dimensional normalized matrix data can be iteratively denoised and updated once in the time dimension, achieving the purpose of globally denoising and updating the entire two-dimensional normalized matrix data. Furthermore, after step S38, the existing LOF algorithm can be used to detect whether there is abnormal data in the new two-dimensional normalized matrix data. If so, steps S31 to S38 described above can be used again to iteratively denoise and update the new two-dimensional normalized matrix data until no abnormal data exists.

[0089] S4. Import the new two-dimensional normalized matrix data into the pre-trained automobile suspension strength estimation model based on a long short-term memory neural network, and output the strength estimate of the target automobile suspension.

[0090] In step S4, the Long Short-Term Memory (LSTM) neural network is a type of recurrent neural network specifically designed to address the long-term dependency problem inherent in general recurrent neural networks. Therefore, it can be trained using the same amount of sample data and existing techniques (e.g., patent CN118332473A - A Method for Testing the Strength of Automobile Suspension Based on Artificial Intelligence) to obtain the automobile suspension strength estimation model. Furthermore, the strength estimate can specifically be an estimate of the automobile suspension strength factor, with a value ranging from (0,1), and a value closer to 1 indicating better quality of the target automobile suspension.

[0091] Therefore, based on the vehicle suspension strength testing method described in steps S1 to S4 above, a new scheme is provided for preprocessing the obtained sensor data using image denoising principles without removing data, and then estimating the vehicle suspension strength based on the processing results. Specifically, first, multi-dimensional time-series sensor data collected by multiple sensors during multiple vehicle suspension strength tests using the drop method on the target vehicle suspension is acquired. Then, the multi-dimensional time-series sensor data is normalized to obtain two-dimensional normalized matrix data. Finally, the MC-WNNM algorithm and / or WSNM algorithm are used to denoise the two-dimensional normalized matrix data to obtain new two-dimensional normalized matrix data. The model is pre-trained using a long short-term memory neural network to estimate the strength of the target vehicle suspension. By importing this model, abnormal sensor data is corrected and restored to true sensor data through image denoising. This can eliminate the adverse effects of abnormal data on the vehicle suspension strength test results to a certain extent. Since there is no need to remove the obtained sensor data, it also avoids the situation where removing too much abnormal data affects the accuracy of the vehicle suspension strength test results. Thus, it is possible to ensure the accuracy of the vehicle suspension strength test results even by performing non-removal preprocessing on the obtained sensor data, which is convenient for practical application and promotion.

[0092] like Figure 2 As shown, the second aspect of this embodiment provides a virtual device for implementing the automobile suspension strength testing method described in the first aspect, including a sensor data acquisition unit, a normalization processing unit, a data denoising processing unit, and a suspension strength estimation unit that are sequentially connected in communication.

[0093] The sensing data acquisition unit is used to acquire multi-dimensional time-series sensing data collected by multiple sensors during multiple vehicle suspension strength test experiments using the drop method on the target vehicle suspension. The multi-dimensional time-series sensing data corresponds one-to-one with the multiple sensors, and each time-series sensing data package contains multiple sensing values ​​that correspond one-to-one with the multiple vehicle suspension strength test experiments and are arranged in the order of the test time.

[0094] The normalization processing unit is used to normalize the multidimensional time-series sensing data to obtain two-dimensional normalized matrix data, wherein the two-dimensional normalized matrix data contains R×Q elements, where R represents the total number of the multiple sensors and Q represents the total number of the multiple vehicle suspension strength test experiments.

[0095] The data denoising processing unit is used to denoise the two-dimensional normalized matrix data using the MC-WNNM algorithm and / or the WSNM algorithm to obtain new two-dimensional normalized matrix data.

[0096] The suspension strength estimation unit is used to import the new two-dimensional normalized matrix data into a pre-trained automotive suspension strength estimation model based on a long short-term memory neural network, and output the strength estimate of the target automotive suspension.

[0097] The working process, working details and technical effects of the aforementioned device provided in the second aspect of this embodiment can be found in the automobile suspension strength test method described in the first aspect, and will not be repeated here.

[0098] like Figure 3 As shown, the third aspect of this embodiment provides a computer device for executing the vehicle suspension strength testing method as described in the first aspect. The device includes a storage module, a processing module, and a transceiver module connected in sequence. The storage module stores a computer program, the transceiver module sends and receives messages, and the processing module reads the computer program and executes the vehicle suspension strength testing method as described in the first aspect. Specifically, the storage module may include, but is not limited to, random-access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out (FIFO) memory, and / or first-in-last-out (FILO) memory, etc.; the processing module may, but is not limited to, use a microprocessor of the STM32F105 series. Furthermore, the computer device may also include, but is not limited to, a power supply module, a display screen, and other necessary components.

[0099] The working process, working details and technical effects of the aforementioned computer equipment provided in the third aspect of this embodiment can be found in the automobile suspension strength test method described in the first aspect, and will not be repeated here.

[0100] This fourth aspect of the embodiment provides a computer-readable storage medium storing instructions comprising the automobile suspension strength testing method as described in the first aspect. Specifically, the computer-readable storage medium stores instructions that, when executed on a computer, perform the automobile suspension strength testing method as described in the first aspect. The computer-readable storage medium refers to a data storage medium, and may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or Memory Sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0101] The working process, working details and technical effects of the aforementioned computer-readable storage medium provided in the fourth aspect of this embodiment can be found in the automobile suspension strength test method described in the first aspect, and will not be repeated here.

[0102] This fifth aspect of the embodiment provides a computer program product, including a computer program or instructions, which, when executed by a computer, implement the vehicle suspension strength testing method as described in the first aspect. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.

[0103] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for testing the strength of an automobile suspension, characterized in that, include: During multiple vehicle suspension strength tests conducted using the drop method on the target vehicle suspension, multidimensional time-series sensing data collected by multiple sensors is obtained. The multidimensional time-series sensing data corresponds one-to-one with the multiple sensors, and each time-series sensing data package contains multiple sensing values ​​that correspond one-to-one with the multiple vehicle suspension strength tests and are arranged in the order of the test time. The multidimensional time-series sensing data is normalized to obtain two-dimensional normalized matrix data, wherein the two-dimensional normalized matrix data contains R×Q elements, where R represents the total number of the multiple sensors and Q represents the total number of the multiple vehicle suspension strength test experiments. The two-dimensional normalized matrix data is denoised using the MC-WNNM algorithm and / or the WSNM algorithm to obtain new two-dimensional normalized matrix data. The new two-dimensional normalized matrix data is imported into a pre-trained automotive suspension strength estimation model based on a long short-term memory neural network, and the strength estimate of the target automotive suspension is output.

2. The method for testing the strength of an automobile suspension according to claim 1, characterized in that, The plurality of sensors include a pressure sensor installed in the shock absorber to measure the pressure of the fluid during shock absorption, a force sensor installed at the suspension connection point to measure the force acting on the suspension, a displacement sensor installed at the spring to measure the degree of spring compression, and / or a steering angle sensor installed at the steering wheel to measure the angle change of the steering wheel.

3. The method for testing the strength of an automobile suspension according to claim 1, characterized in that, The two-dimensional normalized matrix data is denoised using the MC-WNNM and WSNM algorithms to obtain new two-dimensional normalized matrix data, including the following steps S31 to S38: S31. Select the k-th two-dimensional matrix sub-block in the time dimension from the two-dimensional normalized matrix data using a sliding window method, and then execute step S32, where k represents a positive integer with an initial value of 1, and the k-th two-dimensional matrix sub-block contains... One element, Represents a positive integer less than Q; S32. Search for M similar two-dimensional matrix sub-blocks that are most similar to the k-th two-dimensional matrix sub-block within the predefined neighborhood of the k-th two-dimensional matrix sub-block from the two-dimensional normalized matrix data, and then execute step S33, where M represents a positive integer; S33. The k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks are respectively vectorized and straightened to obtain M+1 one-dimensional vectors. These M+1 one-dimensional vectors are then superimposed to obtain a vector of size [missing value]. The original matrix Y is obtained, and then step S34 is executed; S34. Assume the original matrix Y = X + N + S, and use the ADMM algorithm to solve the minimization objective function to obtain the denoised matrix X. Then, execute step S36, wherein the minimization objective function is expressed as follows: In the formula, N represents Gaussian noise, S represents sparse noise, ||S||1 represents the L1 norm used to constrain the sparse noise S, and ||N|| F This represents the F-norm used to constrain Gaussian noise N. λ1, λ2, and λ3 represent the preset regularization coefficients, W represents the multi-channel weighting matrix in diagonal form, r represents a positive integer less than or equal to R, and δ represents the Schatten p-norm used to constrain the denoising matrix X. r Let I represent the noise standard deviation of the r-th channel, I represent the identity matrix, min() represent the minimum function, i represent a positive integer, and w represent the minimum value. i Indicates assigning σ i The weights, σ i represents the i-th singular value in the denoising matrix X, p represents the parameter value used to determine the norm type and takes a value in the interval (0,1], and w represents the weighting of the nuclear norm; S35. Based on the denoising matrix X, reconstruct a new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block and M new two-dimensional matrix similar sub-blocks corresponding one-to-one with the M two-dimensional matrix similar sub-blocks, and then execute step S36; S36. In the two-dimensional normalized matrix data, the k-th two-dimensional matrix sub-block is replaced with the new two-dimensional matrix sub-block, and the M two-dimensional matrix similar sub-blocks are replaced one-to-one with the M new two-dimensional matrix similar sub-blocks to obtain new two-dimensional normalized matrix data, and then step S37 is executed. S37. Determine whether the new two-dimensional matrix sub-block is the last two-dimensional matrix sub-block in the time dimension in the new two-dimensional normalized matrix data. If so, end the denoising process; otherwise, proceed to step S38. S38. Increment k by 1, and then continue to use the sliding window method to select the kth two-dimensional matrix sub-block in the time dimension from the new two-dimensional normalized matrix data, and return to execute step S32.

4. The method for testing the strength of an automobile suspension according to claim 3, characterized in that, From the two-dimensional normalized matrix data, search for M two-dimensional matrix similar sub-blocks that are most similar to the k-th two-dimensional matrix sub-block within a predefined neighborhood of the k-th two-dimensional matrix sub-block, including: For each other two-dimensional matrix sub-block that is located in the predefined neighborhood of the k-th two-dimensional matrix sub-block and has the same size as the k-th two-dimensional matrix sub-block in the two-dimensional normalized matrix data, the Eulerian distance between the k-th two-dimensional matrix sub-block and the corresponding sub-block is calculated. Arrange the other two-dimensional matrix sub-blocks in order of Eulerian distance from nearest to farthest to obtain a sequence of two-dimensional matrix sub-blocks; The first M two-dimensional matrix sub-blocks are selected from the sequence of two-dimensional matrix sub-blocks as the M two-dimensional matrix similar sub-blocks most similar to the k-th two-dimensional matrix sub-block, where M represents a positive integer.

5. The method for testing the strength of an automobile suspension according to claim 3, characterized in that, The k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks are respectively vectorized and straightened to obtain M+1 one-dimensional vectors. These M+1 one-dimensional vectors are then superimposed to obtain a result of size [missing value]. The original matrix Y includes: For each sub-block in the k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks, the element located in the r′-th row and q′-th column of the corresponding sub-block is taken as the h′-th element in the corresponding one-dimensional vector, resulting in the corresponding one-dimensional vector, where r′ represents a positive integer less than or equal to R, and q′ represents a positive integer less than or equal to R. positive integers, The m-th one-dimensional vector among the M+1 one-dimensional vectors that correspond one-to-one with the k-th two-dimensional matrix sub-block and the M similar two-dimensional matrix sub-blocks will be used as the vector of size m. The original matrix Y is obtained by taking the element in the m-th row of the original matrix Y, where m represents a positive integer less than or equal to M+1.

6. The method for testing the strength of an automobile suspension according to claim 5, characterized in that, Based on the denoising matrix X, a new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block and M new two-dimensional matrix similar sub-blocks corresponding one-to-one with the M two-dimensional matrix similar sub-blocks are reconstructed, including: If the element in the m-th row of the original matrix Y is a one-dimensional vector of the k-th two-dimensional matrix sub-block, then the element located in the m-th row and h′-th column of the noise-reducing matrix X will be used as the element located in the r′-th row and q′-th column of the new two-dimensional matrix sub-block corresponding to the k-th two-dimensional matrix sub-block. If the element in the m-th row of the original matrix Y is a one-dimensional vector of the m′-th two-dimensional matrix similarity sub-block among the M two-dimensional matrix similarity sub-blocks, then the element located in the m-th row and h′-th column of the denoising matrix X is taken as the element located in the r′-th row and q′-th column of the new two-dimensional matrix similarity sub-block corresponding to the m′-th two-dimensional matrix similarity sub-block, where m′ represents a positive integer less than or equal to M.

7. A vehicle suspension strength testing device, characterized in that, It includes a sensor data acquisition unit, a normalization processing unit, a data denoising processing unit, and a suspension strength estimation unit, which are connected in sequence. The sensing data acquisition unit is used to acquire multi-dimensional time-series sensing data collected by multiple sensors during multiple vehicle suspension strength test experiments using the drop method on the target vehicle suspension. The multi-dimensional time-series sensing data corresponds one-to-one with the multiple sensors, and each time-series sensing data package contains multiple sensing values ​​that correspond one-to-one with the multiple vehicle suspension strength test experiments and are arranged in the order of the test time. The normalization processing unit is used to normalize the multidimensional time-series sensing data to obtain two-dimensional normalized matrix data, wherein the two-dimensional normalized matrix data contains R×Q elements, where R represents the total number of the multiple sensors and Q represents the total number of the multiple vehicle suspension strength test experiments. The data denoising processing unit is used to denoise the two-dimensional normalized matrix data using the MC-WNNM algorithm and / or the WSNM algorithm to obtain new two-dimensional normalized matrix data. The suspension strength estimation unit is used to import the new two-dimensional normalized matrix data into a pre-trained automotive suspension strength estimation model based on a long short-term memory neural network, and output the strength estimate of the target automotive suspension.

8. A computer device, characterized in that, The device includes a storage module, a processing module, and a transceiver module that are sequentially connected in communication. The storage module is used to store a computer program, the transceiver module is used to send and receive messages, and the processing module is used to read the computer program and execute the vehicle suspension strength test method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that... The computer-readable storage medium stores instructions that, when executed on a computer, perform the automobile suspension strength test method as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or the instructions are executed by the computer, they implement the automobile suspension strength test method as described in any one of claims 1 to 6.

Images