Wet clutch friction element damage failure judgment method, system and storage medium

Abstract

The present invention relates to a wet clutch friction element damage failure discrimination method, system and storage medium, which comprises: obtaining the measured characteristic parameters of the friction pair surface, dividing them into a training set and a test set; the characteristic parameters include working condition parameters and state parameters, the state parameters are the friction pair surface roughness Ra change rate, Fe-Cu element concentration change rate and critical circumferential temperature gradient, and the working condition parameters are input speed and engagement oil pressure; the characteristic parameters are normalized and then dimensionally reduced, and the particle swarm optimization method is used to determine the optimal weight of the BP neural network; the optimal weight is input into the BP neural network, the BP neural network is trained using the training set, and the test set is input into the BP neural network to complete the discrimination of the friction element damage failure based on the friction pair surface roughness Ra change rate, Fe-Cu element concentration change rate and critical circumferential temperature gradient, and output the result. The present invention can realize the quantitative analysis of multiple influencing factors that cause friction element damage failure.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical component damage determination, and in particular to a method, system and storage medium for determining damage and failure of a wet clutch friction component based on multi-data fusion. Background Art

[0002] There are many factors that induce wet clutch failure, including ablation, damage, slippage, wear, warping and deformation of the friction pair surface. The basis for determining the damage and failure of friction elements also includes factors such as temperature, wear, surface roughness, and oil spectral analysis. The mutual influence relationship between the various factors is complex and not clear enough. This puts forward an urgent requirement for studying the comprehensive damage and failure mechanism of wet clutches. Since the damage and failure of wet clutches are induced by multi-parameter factors, if only spectral measurement data is considered, it will inevitably cause inaccurate prediction of failures and damage. Domestic and foreign scholars' research on the engagement characteristics of wet clutch friction pairs mainly focuses on simulation calculation efficiency and hot spot phenomena in the engagement process, the temperature rise of the disc surface in all directions, and the friction and wear conditions under oil spectral analysis. It also includes conditional parameters in the engagement process, such as the study of friction and wear and disc temperature field by considering factors such as materials, engagement pressure, and maximum speed difference.

[0003] At present, multi-source data fusion methods are mainly used in two aspects: one is to extract the characteristic values ​​of the collected multi-source data, determine the fuzziness, and then perform entity modeling based on the influencing factors; the other is to use neural networks or deep learning to train the collected multi-source data for algorithm research. Among the research results of multi-source data fusion, most of them only rely on the influence of a certain data characteristic parameter to predict equipment performance, or just train the algorithm, and do not give a unified judgment on the final training results.

[0004] In order to explore the damage and failure of the friction pair of the wet clutch under complex working conditions, combined with the current research status and extended to the damage and failure of the clutch, when studying the damage and failure of the friction pair, most scholars only studied the working conditions under a single limit law. However, the failure of the wet clutch is induced by multiple factors, and the mutual influence relationship between the various factors is complex and not clear enough. The influence of the fusion of multiple parameters on the damage and failure of the friction pair is not considered. Damage prediction of the wet clutch under the influence of multiple factors is of great value and significance to avoid the wet clutch friction pair damage working condition and maintain the safe operation of the clutch. Summary of the invention

[0005] In view of the above problems, the purpose of the present invention is to provide a method, system and storage medium for distinguishing damage and failure of friction elements of a wet clutch based on multi-data fusion, which integrates multi-category data such as temperature field, oil spectrum, surface morphology, etc., constructs a neural network threshold model of multi-source data fusion, and gives a unified judgment on the damage and failure of multiple influencing factors of the wet clutch, thereby realizing quantitative analysis of the multiple influencing factors that lead to damage and failure of the friction element.

[0006] To achieve the above-mentioned purpose, the present invention adopts the following technical scheme: a method for distinguishing damage and failure of friction elements of a wet clutch, comprising: obtaining characteristic parameters of the measured friction pair surface, and dividing them into a training set and a test set; the characteristic parameters include operating parameters and state parameters, the state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the operating parameters are the input speed and the engagement oil pressure; the characteristic parameters are subjected to data normalization and dimension reduction, and the particle swarm optimization method is used to determine the optimal weight of the BP neural network; the optimal weight is input into the BP neural network, the BP neural network is trained using the training set, and the test set is input into the BP neural network to complete the judgment of damage and failure of the friction element based on the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the result is output.

[0007] Furthermore, the method for obtaining the state parameter includes:

[0008] Change rate of the friction pair surface roughness Ra: Use white light interferometer to extract the surface morphology features of the friction pair before and after wear, specify the extraction point and extraction range, set sampling points on the friction plate surface, and obtain the change rate of the friction pair surface roughness Ra based on the surface roughness Ra difference of the sampling points;

[0009] The Fe-Cu element concentration change rate is obtained by using oil spectral data, performing spectral observation on oil samples in the sliding friction test, extracting the percentage concentration of iron and copper elements in the oil, determining the damage characteristics of the steel sheet by the iron content, and determining the damage characteristics of the friction plate by the copper content;

[0010] The critical circumferential temperature gradient: the steel sheet and friction plate of the friction pair are modeled, the paths with the highest average temperatures on different circumferences are selected, the node temperature values ​​on the paths are extracted and the temperature differences between adjacent nodes are calculated, and the maximum temperature difference between adjacent nodes on the circumferential path is used as the temperature gradient feature.

[0011] Further, the dimension reduction after data normalization processing on the characteristic parameters includes:

[0012] The min-max linear normalization method is used to perform linear transformation on the characteristic parameters, scaling the data values ​​to a preset range and being in the same order of magnitude, thereby improving the speed of the optimal solution and the accuracy of data analysis.

[0013] Further, the particle swarm optimization method is used to determine the optimal weight of the BP neural network, including:

[0014] Set the number of layers and neurons of the neural network, initialize the parameter values, and encode the initial parameters;

[0015] Calculate the fitness, and update the particle position and velocity according to the fitness, and determine whether the updated position and velocity meet the set termination condition. If so, output the optimal weight. Otherwise, update the acceleration factor and continue to update.

[0016] Furthermore, the termination condition is a preset maximum number of iterations or a preset minimum error from an expected result.

[0017] Further, updating the particle position and velocity according to the fitness includes:

[0018] Position update: If the calculated fitness value is better than the current individual extreme value, the individual optimal position is updated to the calculated particle position; if there is an optimal value among all the currently provided individual extreme values ​​of the particles, if it is better than the current global extreme value, the global extreme value is updated to the optimal value, and the global optimal position is updated to the individual optimal position of the particle;

[0019] Speed ​​update: If V i <V min Then V i Update to V min ,V i >V max Then V i Update to V max Otherwise, update the acceleration factor and continue to search until the condition is met; among them, V i is the current particle velocity, V min 、V max The minimum and maximum values ​​for the particle velocity search interval.

[0020] Further, the step of inputting the optimal weight into the BP neural network and training the BP neural network using the training set includes:

[0021] Calculate the joint state energy function and hidden layer node activation probability according to the optimal weights;

[0022] The input array is fused through the hidden layer nodes;

[0023] Calculate the error, update the weights and thresholds;

[0024] Determine whether the pre-set conditions are met. If they are met, output the prediction result. Otherwise, recalculate the error.

[0025] A wet clutch friction element damage failure discrimination system comprises: a parameter acquisition module, which acquires the characteristic parameters of the measured friction pair surface and divides them into a training set and a test set; the characteristic parameters include operating parameters and state parameters, the state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the operating parameters are the input speed and the engagement oil pressure; a particle swarm optimization module, which performs data normalization processing on the characteristic parameters and then reduces the dimension, and uses the particle swarm optimization method to determine the optimal weight of the BP neural network; a discrimination output module, which inputs the optimal weight into the BP neural network, trains the BP neural network with the training set, and then inputs the test set into the BP neural network, completes the discrimination of the friction element damage failure based on the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and outputs the result.

[0026] A computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any of the above methods.

[0027] A computing device comprises: one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the above methods.

[0028] The present invention adopts the above technical solution, which has the following advantages:

[0029] 1. The friction pair of the present invention is mainly composed of a friction plate with an oil groove and a dual steel plate. The finite element simulation of the temperature field of the wet clutch friction pair is carried out using ABAQUS simulation software. The simulation is carried out from a single-pair multi-field coupled sliding friction simulation. By analyzing the temperature in the model, the critical damage temperature of the friction pair is estimated, and a temperature threshold model is constructed.

[0030] 2. Observing the changes in surface morphology before and after sliding is also the basis for determining damage. The present invention uses white light interferometer to extract surface morphology features of the friction pair before and after wear. The damaged disk surface can be repaired by software and the wear amount can be obtained by comparing the initial disk surface morphology features, and then the surface roughness information can be extracted in combination with the surface concave and convex conditions. Roughness information also provides a data basis for multi-source data fusion.

[0031] 3. The present invention uses oil spectral analysis technology to perform spectral observation during the entire sliding stage of the friction pair, and records the concentration of iron and other metals in the damaged state. The content of Fe element is mainly used to determine the failure of the steel sheet, while the content of Cu element is mainly used to determine the failure of the friction plate. The percentage of each element concentration obtained provides spectral data support for multi-source data fusion.

[0032] 4. The present invention integrates all the collected data according to the data classification form and data processing method, and then uniformly processes the data to obtain a unified evaluation. The neural network algorithm is used to determine the data proportion based on the fusion of multi-source data (surface roughness, temperature, spectrum) and construct a feature matrix based on the eigenvalue, and finally a conclusion of fusion of multiple data is obtained, and a multi-parameter judgment basis is given for the damage and failure of multiple influencing factors of the wet clutch. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the overall flow of the discrimination method in one embodiment of the present invention;

[0034] Figure 2 It is a test implementation scheme in one embodiment of the present invention;

[0035] Figure 3 is a surface topography extraction result and a roughness calculation area in one embodiment of the present invention;

[0036] Figure 4 is a diagram of the temperature field analysis result of a finite element unit in one embodiment of the present invention;

[0037] Figure 5 is a PSO-BP neural network data processing flow chart in one embodiment of the present invention;

[0038] Figure 6 It is a schematic diagram of the structure of a computing device in one embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0040] The wet clutch friction element damage and failure judgment method based on multi-data fusion of the present invention mainly adopts the wet clutch friction element damage and failure judgment method based on multi-data fusion as the main body, and judges the damage and failure of the wet clutch friction element affected by multiple parameters through three major aspects: theoretical research, experimental simulation and data fusion; gives a unified judgment on the damage and failure of the wet clutch caused by multiple influencing factors, and realizes the quantitative analysis of the multiple influencing factors that cause the damage and failure of the friction element.

[0041] In one embodiment of the present invention, a method for distinguishing damage and failure of a wet clutch friction element based on multi-data fusion is provided. Since there are many situations in which a wet clutch may experience damage failure, the form of damage can be characterized by different characteristic parameters. The distinguishing method provided in this embodiment determines that the damage to the friction element includes multiple criteria such as temperature and wear. In order to study the damage characteristics of the friction element, characteristic quantities are extracted to characterize the damage to the friction pair. The temperature damage characteristics are characterized by the circumferential temperature gradient; the wear damage characteristics are characterized by the maximum change rate of the Fe and Cu element concentrations and the maximum change rate of the surface roughness Ra. In this embodiment, Figure 1 As shown, the method comprises the following steps:

[0042] 1) Obtain the characteristic parameters of the measured friction pair surface and divide them into a training set and a test set; the characteristic parameters include working condition parameters and state parameters, the state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the working condition parameters are the input speed and the engagement oil pressure;

[0043] 2) After normalizing the characteristic parameters, the dimension is reduced and the optimal weight of the BP neural network is determined by particle swarm optimization method;

[0044] 3) Input the optimal weight into the BP neural network, use the training set to train the BP neural network, and then input the test set into the BP neural network to complete the judgment of friction element damage and failure based on the change rate of friction pair surface roughness Ra, the change rate of Fe-Cu element concentration and the critical circumferential temperature gradient, and output the results.

[0045] In this embodiment, based on the friction pair surface roughness and Fe-Cu element concentration data obtained by building a wet clutch friction and wear comprehensive transmission test bench and the friction pair circumferential temperature obtained by finite element simulation, the surface roughness Ra change rate, the Fe-Cu element concentration change rate and the critical circumferential temperature are extracted as characteristic parameters. The min-max linear normalization method is used to linearly transform the characteristic data, and the data values ​​are scaled to the specified range and in the same order of magnitude.

[0046] The construction method of the wet clutch friction and wear comprehensive test bench is as follows: build a 1:1 wet clutch friction and wear comprehensive test bench, which includes a motor, a speed torque sensor, an inertia, a clutch package, a speed torque sensor and a brake device. The test scheme is short-term multi-condition sliding and long-term single-condition sliding, such as Figure 2 The operating information covers the oil pressure of 0.3Mpa to 1.0Mpa, the speed of 100r / min to 1800r / min, the pause and cooling at intervals of 10min, the collection of Fe-Cu element concentration and the re-circulation test, and the accelerated life until damage failure.

[0047] In the above step 1), the method for obtaining the state parameter includes:

[0048] Change rate of the friction pair surface roughness Ra: Use white light interferometer to extract the surface morphology features of the friction pair before and after wear, specify the extraction point and extraction range, set sampling points on the friction plate surface, and obtain the change rate of the friction pair surface roughness Ra based on the surface roughness Ra difference of the sampling points;

[0049] For example, Figure 3 As shown in the figure, a sampling point is set every 45 degrees on the friction plate surface. The difference in color represents the different heights of the disc surface. The reason for the alternating strips is the difference in surface roughness Ra. The depression in the middle indicates that this area is subjected to greater contact force and torque during the engagement process during sliding and wear, resulting in severe wear.

[0050] Fe-Cu element concentration change rate: The oil spectral data is obtained by spectrally observing the oil samples in the sliding test, extracting the percentage concentration of iron and copper elements in the oil, judging the damage characteristics of the steel sheet by the iron content, and judging the damage characteristics of the friction plate by the copper content;

[0051] For example, a sliding friction test was conducted with the help of a wet clutch friction and wear comprehensive test bench. During the sliding friction test, the oil samples were sampled at 10-minute intervals for spectral observation to extract the percentage concentration of iron and copper elements in the oil. The damage characteristics of the steel sheet were mainly determined by the iron content; the damage characteristics of the friction sheet were determined by the copper content.

[0052] Critical circumferential temperature gradient: Model the steel sheet and friction plate of the friction pair, select the path with the highest average temperature on different circumferences, extract the node temperature values ​​on the path and calculate the temperature difference between adjacent nodes, and use the maximum temperature difference between adjacent nodes on the circumferential path as the temperature gradient feature;

[0053] For example, the steel plate and friction plate are modeled according to the 1:1 ratio of the actual friction pair, and the material property parameter settings are shown in Table 1. After assembly, meshing, material property setting, setting analysis steps, load application, and boundary constraint processes, the sliding process of the friction pair is simulated and analyzed, and the temperature field results are shown in Table 1. Figure 4 As shown. Formulate the critical circumferential temperature gradient and select the path with the highest average temperature on different circumferences. Figure 4 As shown in the middle circle in . The 64 node temperatures with equal gradient on the circumferential path of the friction plate are taken as the research objects, the node temperature values ​​on the path are extracted and the temperature difference values ​​of adjacent nodes are calculated, and the maximum temperature difference value of adjacent nodes on the circumferential path is taken as the temperature gradient feature.

[0054] Table 1 Friction pair geometric dimensions and material property parameters

[0055]

[0056] In the above step 2), the feature parameters are normalized and then dimensionally reduced. Specifically, the feature parameters are linearly transformed using the min-max linear normalization method, and the data values ​​are scaled to a preset range and to the same order of magnitude, thereby improving the speed of the optimal solution and the accuracy of data analysis.

[0057] In the above step 2), if Figure 5 As shown in the figure, the particle swarm optimization method is used to determine the optimal weight of the BP neural network. Assuming that the search space is an N-dimensional space, the population consists of n particles. This population X=(x1,x2,…,x n ), where the position of the kth particle is represented by X k =(x k1 ,x k2 ,…,x kn ) and the speed is expressed by V k =(v k1 ,v k2 ,…,v kn ) represents the individual extreme value of the particle, and P k =(p k1 ,p k2 ,…,p kn ) represents the overall global extreme value, and P g =(p g1 ,p g2 ,…,p gn ). Specifically, the following steps are included:

[0058] 2.1) Set the number of layers and neurons of the neural network, initialize the parameter values, and encode the initial parameters;

[0059] Specifically: select various parameters involved in the PSO algorithm, the maximum number of iterations Tmax , learning training factor c1=c2=2, particle speed search interval [V min ,V max ],V min 、V max The minimum and maximum values ​​of the particle velocity search interval are initialized randomly, the position and velocity of the search point are set in advance, the initial position of each particle is set, the global extreme value is found from the individual extreme value, and the position is recorded.

[0060] Among them, c1 and c2 represent pushing each particle to P i and P g The weight of the statistical acceleration term for the position. Low values ​​allow particles to wander outside the target area before being pulled back, while high values ​​cause particles to suddenly rush towards or over the target area.

[0061] 2.2) Calculate the fitness, and update the particle position and velocity according to the fitness, and determine whether the updated position and velocity meet the set termination conditions. If they meet, output the optimal weight. Otherwise, update the acceleration factor and continue updating.

[0062] In this embodiment, the termination condition is a preset maximum number of iterations T. max Or pre-set the minimum error from the expected result.

[0063] In this embodiment, the particle position and velocity are updated according to the fitness, including:

[0064] (1) Position update: If the calculated fitness value is better than the current individual extreme value, the individual optimal position is updated to the calculated particle position; if there is an optimal value among all the currently provided individual extreme values ​​of the particles, and if it is better than the current global extreme value, the global extreme value is updated to the optimal value, and the global optimal position is updated to the individual optimal position of the particle;

[0065] (2) Speed ​​update: If V i <V min Then V i Update to V min ,V i >V max Then V i Update to V max Otherwise, update the acceleration factor and continue to search until the condition is met; among them, V i is the current particle speed.

[0066] In the above step 3), the optimal weight is input into the BP neural network, and the BP neural network is trained using the training set, including:

[0067] 3.1) Calculate the joint state energy function and hidden layer node activation probability according to the optimal weight;

[0068] 3.2) Perform data fusion on the input array through hidden layer nodes;

[0069] 3.3) Calculate the error, update the weights and thresholds;

[0070] 3.4) Determine whether the pre-set conditions are met. If so, output the prediction result. Otherwise, recalculate the error.

[0071] In this embodiment, the weights and thresholds of the training network are initialized, and random values ​​in the interval [-0.999, 0.999] are selected to connect the weights V, W and the threshold. The output after the input parameters are calculated, the output vector of the input layer in the network is consistent with the input pattern vector, and the input and output of each neuron in the hidden layer are calculated according to formula (1) and formula (2):

[0072]

[0073]

[0074] Where j = 1, 2, ..., n, and p represents the number of neurons in the input layer.

[0075] Adjust the error of each neuron in the output layer, according to the given expected output, feedforward loop training, check the error value, and reach the given error range:

[0076]

[0077] Where t = 1, 2, ..., m, and k represents the number of neurons in the output layer.

[0078] The connection weight W between the input layer and the hidden layer, the connection weight V between the hidden layer and the output layer, and the threshold θ of the hidden layer neurons are corrected by checking the error.

[0079]

[0080]

[0081]

[0082] Where α and β represent learning rates, 0<α<1, j=1,2,…,p, t=1,2,…,m; 0<β<1, i=1,2,…,n, j=1,2,…,p.

[0083] Determine whether the number of global network training times reaches the set training requirements and satisfies the error setting value ε at the same time, that is, E≤ε. If the conditions are met at the same time, it means that the training results converge and the learning ends. If the number of network learning times is less than the set number, or there is still a situation of E≤ε, continue the feedforward loop training until convergence.

[0084] In one embodiment of the present invention, a wet clutch friction element damage failure judgment system based on multi-data fusion is provided, which comprises:

[0085] The parameter acquisition module acquires the characteristic parameters of the measured friction pair surface and divides them into a training set and a test set; the characteristic parameters include operating condition parameters and state parameters. The state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the operating condition parameters are the input speed and the engagement oil pressure;

[0086] The particle swarm optimization module normalizes the characteristic parameters and then reduces the dimension, and uses the particle swarm optimization method to determine the optimal weight of the BP neural network;

[0087] The discrimination output module inputs the optimal weight into the BP neural network, trains the BP neural network with the training set, and then inputs the test set into the BP neural network to complete the discrimination of friction element damage and failure based on the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and output the results.

[0088] The system provided in this embodiment is used to execute the above-mentioned method embodiments. Please refer to the above-mentioned embodiments for specific processes and detailed contents, which will not be repeated here.

[0089] like Figure 6As shown, it is a schematic diagram of the structure of a computing device provided in an embodiment of the present invention. The computing device can be a terminal, which can include: a processor, a communication interface, a memory, a display screen and an input device. Among them, the processor, the communication interface and the memory complete mutual communication through a communication bus. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. When the computer program is executed by the processor, a discrimination method is implemented; the internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be implemented through WIFI, a management network, NFC (near field communication) or other technologies. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the housing of the computing device, or an external keyboard, touchpad or mouse. The processor can call the logical instructions in the memory to execute the following method:

[0090] The characteristic parameters of the measured friction pair surface are obtained and divided into a training set and a test set; the characteristic parameters include operating condition parameters and state parameters, the state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the operating condition parameters are the input speed and the engagement oil pressure; the characteristic parameters are normalized and then dimensionally reduced, and the particle swarm optimization method is used to determine the optimal weight of the BP neural network; the optimal weight is input into the BP neural network, the BP neural network is trained using the training set, and then the test set is input into the BP neural network to complete the judgment of the damage and failure of the friction element based on the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the result is output.

[0091] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0092] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the computing device to which the scheme of the present application is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0093] In one embodiment of the present invention, a computer program product is provided, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, when the program instructions are executed by a computer, the computer can execute the methods provided by the above-mentioned method embodiments, for example comprising: obtaining the characteristic parameters of the friction pair surface measured, and dividing them into a training set and a test set; the characteristic parameters include operating condition parameters and state parameters, the state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the operating condition parameters are the input speed and the engagement oil pressure; performing data normalization processing on the characteristic parameters and then reducing the dimension, and using the particle swarm optimization method to determine the optimal weight of the BP neural network; inputting the optimal weight into the BP neural network, training the BP neural network with the training set, and then inputting the test set into the BP neural network, completing the judgment of the damage and failure of the friction element based on the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and outputting the result.

[0094] In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided, which stores server instructions, and the computer instructions enable a computer to execute the methods provided in the above embodiments, for example, including: obtaining the characteristic parameters of the measured friction pair surface, dividing them into a training set and a test set; the characteristic parameters include operating parameters and state parameters, the state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the operating parameters are the input speed and the engagement oil pressure; the characteristic parameters are normalized and then reduced in dimension, and the particle swarm optimization method is used to determine the optimal weight of the BP neural network; the optimal weight is input into the BP neural network, the BP neural network is trained using the training set, and the test set is input into the BP neural network to complete the judgment of the damage and failure of the friction element based on the change rate of the friction pair surface roughness Ra, the change rate of the Fe-Cu element concentration and the critical circumferential temperature gradient, and the result is output.

[0095] The above embodiment provides a computer-readable storage medium, whose implementation principle and technical effect are similar to those of the above method embodiment, and will not be repeated here.

[0096] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0097] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0098] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining damage and failure of a wet clutch friction element, characterized in that: include: Obtain the measured characteristic parameters of the friction pair surface and divide them into a training set and a test set; The characteristic parameters include operating parameters and state parameters, the state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe element concentration, the change rate of the Cu element concentration and the critical circumferential temperature gradient, and the operating parameters are the input speed and the engagement oil pressure; The characteristic parameters are subjected to data normalization and dimension reduction, and the optimal weight of the BP neural network is determined by using a particle swarm optimization method; Inputting the optimal weight into the BP neural network, training the BP neural network with the training set, and then inputting the test set into the BP neural network to complete the judgment of friction element damage failure based on the change rate of friction pair surface roughness Ra, the change rate of Fe element concentration, the change rate of Cu element concentration and the critical circumferential temperature gradient, and output the result; The method for obtaining the state parameter includes: Change rate of the friction pair surface roughness Ra: Use white light interferometer to extract the surface morphology features of the friction pair before and after wear, specify the extraction point and extraction range, set sampling points on the friction plate surface, and obtain the change rate of the friction pair surface roughness Ra based on the surface roughness Ra difference of the sampling points; The Fe element concentration change rate and the Cu element concentration change rate are obtained by using oil spectral data, performing spectral observation on oil samples in the sliding friction test, extracting the percentage concentration of iron and copper elements in the oil, determining the damage characteristics of the steel sheet by the iron content, and determining the damage characteristics of the friction plate by the copper content; The critical circumferential temperature gradient: the steel sheet and friction plate of the friction pair are modeled, the paths with the highest average temperatures on different circumferences are selected, the node temperature values ​​on the paths are extracted, and the temperature difference values ​​of adjacent nodes are calculated, and the maximum temperature difference value of adjacent nodes on the circumferential path is used as the temperature gradient feature; The step of performing data normalization processing on the characteristic parameters and then reducing the dimension includes: The characteristic parameters are linearly transformed using the min-max linear normalization method, and the data values ​​are scaled to a predetermined range and to be in the same order of magnitude, thereby improving the speed of the optimal solution and the accuracy of data analysis; The method of using particle swarm optimization to determine the optimal weight of the BP neural network includes: Set the number of layers and neurons of the neural network, initialize the parameter values, and encode the initial parameters; Calculate the fitness, and update the particle position and velocity according to the fitness, and determine whether the updated position and velocity meet the set termination condition. If so, output the optimal weight, otherwise, update the acceleration factor and continue to update; The step of inputting the optimal weight into the BP neural network and training the BP neural network using the training set comprises: Calculate the joint state energy function and hidden layer node activation probability according to the optimal weights; The input array is fused through the hidden layer nodes; Calculate the error, update the weights and thresholds; Determine whether the pre-set conditions are met. If they are met, output the prediction result. Otherwise, recalculate the error.

2. The identification method according to claim 1, characterized in that: The termination condition is a preset maximum number of iterations or a preset minimum error from an expected result.

3. The identification method according to claim 1, characterized in that: The updating of the particle position and velocity according to the fitness comprises: Position update: If the calculated fitness value is better than the current individual extreme value, the individual optimal position is updated to the calculated particle position; if there is an optimal value among all the currently provided individual extreme values ​​of the particles, if it is better than the current global extreme value, the global extreme value is updated to the optimal value, and the global optimal position is updated to the individual optimal position of the particle; Speed ​​update: If V i <V min Then V i Update to V min ,V i >V max Then V i Update to V max Otherwise, update the acceleration factor and continue to search until the condition is met; among them, V i is the current particle velocity, V min 、V max The minimum and maximum values ​​for the particle velocity search interval.

4. A wet clutch friction element damage failure judgment system, characterized in that: include: A parameter acquisition module is used to acquire the characteristic parameters of the friction pair surface measured and divided into a training set and a test set; the characteristic parameters include operating condition parameters and state parameters, the state parameters are the change rate of the friction pair surface roughness Ra, the change rate of the Fe element concentration, the change rate of the Cu element concentration and the critical circumferential temperature gradient, and the operating condition parameters are the input speed and the engagement oil pressure; A particle swarm optimization module performs data normalization on the characteristic parameters and then reduces the dimension, and uses a particle swarm optimization method to determine the optimal weight of the BP neural network; a discrimination output module, inputting the optimal weight into the BP neural network, training the BP neural network with the training set, and then inputting the test set into the BP neural network, completing the discrimination of friction element damage failure based on the change rate of the friction pair surface roughness Ra, the change rate of the Fe element concentration, the change rate of the Cu element concentration and the critical circumferential temperature gradient, and outputting the result; The method for obtaining the state parameter includes: Change rate of the friction pair surface roughness Ra: Use white light interferometer to extract the surface morphology features of the friction pair before and after wear, specify the extraction point and extraction range, set sampling points on the friction plate surface, and obtain the change rate of the friction pair surface roughness Ra based on the surface roughness Ra difference of the sampling points; The Fe element concentration change rate and the Cu element concentration change rate are obtained by using oil spectral data, performing spectral observation on oil samples in the sliding friction test, extracting the percentage concentration of iron and copper elements in the oil, determining the damage characteristics of the steel sheet by the iron content, and determining the damage characteristics of the friction plate by the copper content; The critical circumferential temperature gradient: the steel sheet and friction plate of the friction pair are modeled, the paths with the highest average temperatures on different circumferences are selected, the node temperature values ​​on the paths are extracted, and the temperature difference values ​​of adjacent nodes are calculated, and the maximum temperature difference value of adjacent nodes on the circumferential path is used as the temperature gradient feature; The step of performing data normalization processing on the characteristic parameters and then reducing the dimension includes: The characteristic parameters are linearly transformed using the min-max linear normalization method, and the data values ​​are scaled to a predetermined range and to be in the same order of magnitude, thereby improving the speed of the optimal solution and the accuracy of data analysis; The method of using particle swarm optimization to determine the optimal weight of the BP neural network includes: Set the number of layers and neurons of the neural network, initialize the parameter values, and encode the initial parameters; Calculate the fitness, and update the particle position and velocity according to the fitness, and determine whether the updated position and velocity meet the set termination condition. If so, output the optimal weight, otherwise, update the acceleration factor and continue to update; The step of inputting the optimal weight into the BP neural network and training the BP neural network using the training set comprises: Calculate the joint state energy function and hidden layer node activation probability according to the optimal weights; The input array is fused through the hidden layer nodes; Calculate the error, update the weights and thresholds; Determine whether the pre-set conditions are met. If they are met, output the prediction result. Otherwise, recalculate the error.

5. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any one of the methods of claims 1 to 3.

6. A computing device, characterized in that include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any one of the methods described in claims 1 to 3.

Images