A membrane switch performance detection method and system

Through the combination of clustering algorithm, four-probe detection and BP neural network, the problems of large errors and slow speeds of thin-film switch performance detection are solved, and high-accuracy performance prediction is achieved.

CN118777861BActive Publication Date: 2024-11-22GUANGZHOU JIAYI ELECTRONICS THIN FILM SWITCH
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411283159.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-13
Publication Date
2024-11-22
Estimated Expiration
2044-09-13

AI Technical Summary

Technical Problem

The existing film switch performance detection methods have large errors, slow speeds and have not used high-tech technologies, making it difficult to accurately detect film switch performance and make predictions.

Method used

The clustering algorithm is used to record the film switch stroke, calculate the deterioration state and residual service life; the resistance is measured using the four-probe detection method and the error is optimized by the cubic spline interpolation method; performance index score evaluation and prediction are combined with the BP neural network and the zebra optimization algorithm. .

Benefits of technology

The accuracy and speed of film switch performance detection are improved, and high accuracy prediction of film switch performance is achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118777861B_ABST
    Figure CN118777861B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of industrial detection, and discloses a membrane switch performance detection method and system. First, the stroke of the membrane switch is recorded after the membrane switch is pressed, and the stroke of the membrane switch is clustered using a clustering algorithm to obtain the degradation state of the membrane switch and calculate the remaining service life of the membrane switch; secondly, the voltage and current of the membrane switch are measured and the resistance value is calculated and fitted, and the fitting error is optimized using a cubic spline interpolation method to obtain the membrane switch resistance; then, according to the data detection point on the membrane switch, the voltage and voltage threshold of the data detection point are compared to obtain the membrane switch withstand voltage; the membrane switch performance is evaluated using the membrane switch performance index score, and finally the BP neural network is trained, and the BP neural network parameters are optimized using a zebra optimization algorithm to achieve the prediction of the membrane switch performance. The present invention reflects the performance of the membrane switch through the membrane switch performance index, and the detection result is objective and accurate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of industrial detection, and in particular to a membrane switch performance detection method and system. Background Art

[0002] Chinese patent CN117849607A discloses an automobile switch performance detection system and method based on artificial intelligence. The method specifically includes: after the automobile switch leaves the factory, a switch sample is selected from the current batch of switches as the first sample, the first sample is tested using the same pressing parameters, the quality coefficient of the current batch of switches is calculated, and the extraction ratio is determined; a second sampling is performed according to the extraction ratio to obtain a second sample, the second sample is connected to the control circuit, and the switch in the second sample is set with different pressing parameters, the switch is tested, and the response time is recorded, and the correlation function between the response time and the pressing parameter is established; in the pressing test, the current and voltage after each switch is pressed are recorded, and the influence of the number of presses on the circuit control performance is evaluated according to the changes in the current and voltage, and the probability of normal operation of the switch after each press is calculated; then, the actual control time of the switch and the action curve of the pressing parameter are calculated according to the normal operation probability of the switch, and the degradation coefficient of the switch is calculated according to the action curve, and the degradation coefficient and the threshold are compared to detect the switch performance. However, the measurement data of the invention is prone to errors, and the results are inaccurate.

[0003] Membrane switches are different from traditional switches, and their performance testing is more difficult than that of traditional switches. The traditional membrane switch performance testing method relies on a machine to press the switch and record the parameters of each membrane switch press, which is time-consuming and labor-intensive, and does not use high-tech technologies such as neural networks. The testing speed is slow and the performance of the membrane switch cannot be predicted. Summary of the invention

[0004] In view of the problems in the related art, the present invention provides a membrane switch performance detection method to overcome the above technical problems existing in the existing related art.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] The present invention is a membrane switch performance detection method, comprising the following steps:

[0007] S1. Select a membrane switch, record the travel of the membrane switch after pressing, cluster the travel of the membrane switch using a clustering algorithm, obtain a set of membrane switch degradation states, generate a membrane switch state transition probability matrix, calculate the remaining service life of the membrane switch, and use the remaining service life of the membrane switch as the first membrane switch performance indicator;

[0008] S2. After measuring the voltage and current of the membrane switch to be measured by a four-probe detection method, the resistance value is calculated and fitted, the fitting error is optimized by a cubic spline interpolation method, and the resistance value after fitting is judged to obtain the membrane switch resistance, and the membrane switch resistance is used as the second membrane switch performance indicator;

[0009] S3, sampling the data detection points on the membrane switch to generate a membrane switch detection matrix, passing different voltages into the membrane switch in sequence to obtain the membrane switch withstand voltage, and using the membrane switch withstand voltage as the third membrane switch performance indicator;

[0010] S4. Comprehensively consider the first membrane switch performance index, the second membrane switch performance index and the third membrane switch performance index, and use the membrane switch performance index score to evaluate the membrane switch performance; and train the BP neural network, and use the zebra optimization algorithm to optimize the BP neural network parameters to achieve the prediction of the membrane switch performance.

[0011] The invention first records the travel of the membrane switch by pressing the membrane switch, and clusters the travel of the membrane switch, records the membrane switch travel that does not meet the conditions as a degraded state of the membrane switch, calculates the degradation transfer probability of the membrane switch, and obtains the normal number of membrane switch travels and the degraded number of membrane switch travels, thereby predicting the remaining service life of the membrane switch, which is recorded as the first membrane switch performance index; secondly, the membrane switches are grouped to measure the current and voltage, first fit the experimental data of the first group, use the cubic spline interpolation method to optimize the fitting error for the second group of experimental data, and then use the third group of experimental data to verify the accuracy of the fitting results, and obtain the membrane switch resistance, which is recorded as the second membrane The switch performance index is verified by repeated fitting and the result error is small; the data monitoring points of the membrane switch are sampled again, and the voltage threshold at the data monitoring points of the membrane switch is judged by different voltages to obtain the membrane switch withstand voltage, which is recorded as the third membrane switch performance index; combined with the obtained membrane switch performance index, the membrane switch performance index score is calculated to evaluate the membrane switch performance; the membrane switch performance data is used to train the BP neural network, and the BP neural network parameters are optimized based on the zebra optimization algorithm. The algorithm can speed up the convergence speed of the neural network and has a strong optimization ability; finally, the membrane switch performance is predicted through the final BP neural network, and the prediction accuracy of the neural network is high.

[0012] Preferably, the S1 comprises the following steps:

[0013] S11. Select a membrane switch, denoted as membrane switch set A. 1 ={a 1 ,a 2 ,a 3 ,…,a i}, where a irepresents the i-th membrane switch selected, pressing the membrane switches in the membrane switch set, recording the travel of the membrane switches in the membrane switch set, setting that all membrane switches in the membrane switch set fail after j times of pressing, and using the travel values ​​of the membrane switches in the membrane switch set as the data elements of the membrane switch travel matrix to generate the membrane switch travel matrix B 1 as follows,

[0014] ,

[0015] Among them, a ij Indicates the j-th pressing stroke data of the i-th membrane switch;

[0016] A row in the membrane switch stroke matrix is ​​selected as a set of membrane switch strokes to be detected, and one data element is selected from the membrane switch stroke set to be detected as a cluster center, which is recorded as the first cluster center. The distance from the data element in the membrane switch stroke set to be detected to the first cluster center is calculated, and the data element corresponding to the maximum distance is used as the second cluster center. The distance from the data element in the membrane switch stroke set to be detected to the first cluster center and the second cluster center is calculated respectively, and the data element corresponding to the maximum distance is used as the third cluster center. The third cluster center satisfies that the distance from the third cluster center to the first cluster center and the distance from the second cluster center are greater than the distance from the first cluster center to the second cluster center. The data elements in the membrane switch stroke set to be detected are repeatedly selected to obtain k cluster centers to form an initial cluster center set.

[0017] For each initial cluster center in the initial cluster center set, one class is corresponding to each initial cluster center. The initial cluster center of the data element in the to-be-detected membrane switch stroke set closest to the initial cluster center set is found, and the corresponding data element in the to-be-detected membrane switch stroke set is assigned to the class corresponding to the nearest initial cluster center, and k assigned classes are obtained, which are recorded as the initial class set; the cluster center is then found from the initial class set, and S11 is repeated until the distance from the data element in the initial class set to the cluster center converges, and a final class set is obtained;

[0018] S12, clustering different membrane switch strokes according to the final class set, recording different membrane switch strokes as membrane switch resident states, and obtaining a membrane switch resident state set. ,in Indicates a membrane switch resident state; setting a degradation state threshold value as ω, extracting membrane switch resident states less than ω from the membrane switch resident state set to obtain a membrane switch degradation state set; the membrane switch degradation state in the membrane switch degradation state set Transformed into membrane switch deterioration state The probability of the membrane switch state transition probability is recorded as the membrane switch state transition probability, and the characteristic dimension of the membrane switch state transition probability is set to , generate the membrane switch state transition probability matrix B 2 as follows,

[0019] ,

[0020] in, Indicates membrane switch states, and the characteristic dimension is The membrane switch state transition probability;

[0021] A new membrane switch is obtained, and a new membrane switch stroke matrix is ​​generated after the new membrane switch is pressed. The new membrane switch stroke matrix is ​​transformed through the membrane switch state transition probability matrix, and the current state times and the degraded state times are calculated. The residence time of the new membrane switch in the current state is set to , the residence time of the new membrane switch in the degraded state is , taking the remaining service life of the membrane switch as the first membrane switch performance indicator, the remaining service life of the membrane switch is calculated as follows:

[0022] ,

[0023] Among them, C represents the remaining service life of the membrane switch, Indicates the current state i 1 The dwell time, Indicates the degradation state i 1 The dwell time, Indicates the number of degraded states.

[0024] The invention records the travel of the membrane switch by pressing the membrane switch, clusters the membrane switch travel using a clustering algorithm, divides the membrane switch travel into a membrane switch degradation state and a membrane switch current state according to the resident state, calculates the degradation transfer probability of the membrane switch, obtains the normal number of membrane switch travels and the degraded number of membrane switch travels, thereby predicting the remaining service life of the membrane switch, which is recorded as the first membrane switch performance indicator.

[0025] Preferably, S2 comprises the following steps:

[0026] S21, set the four probes to be α 1 , α 2 , α 3 and α 4, respectively denoted as a first voltage probe, a second voltage probe, a first current probe, and a second current probe; selecting a membrane switch to be measured, and dividing the membrane switch to be measured into a first test membrane switch group, a second test membrane switch group, and a third test membrane switch group;

[0027] In the first test membrane switch group, fix the power supply in the middle of the membrane switch, and select b from left to right on the outer side of the test membrane switch in the first test membrane switch group. 1 test probe points, respectively place the first voltage probe and the second voltage probe on the leftmost side of the test membrane switch in the first test membrane switch group, and continuously move the first voltage probe and the second voltage probe to the right to ensure that the first voltage probe and the second voltage probe are on the test probe points until the first voltage probe and the second voltage probe coincide with the power supply position, and record the voltage value, which is recorded as the voltage record set of the first test membrane switch group;

[0028] In the second test membrane switch group, fix the power supply at the leftmost position of the membrane switch, and select b from left to right on the inner side of the test membrane switch in the second test membrane switch group. 2 test probe points, respectively place the first voltage probe and the second voltage probe on the rightmost side of the test membrane switch in the second test membrane switch group, and continuously move the first voltage probe and the second voltage probe to the left to ensure that the first voltage probe and the second voltage probe are on the test probe points until the first voltage probe and the second voltage probe coincide with the power supply position, and record the voltage value, which is recorded as the voltage record set of the second test membrane switch group;

[0029] In the third test membrane switch group, the voltage is fixed at the leftmost position of the membrane switch, and the outer side of the test membrane switch in the third test membrane switch group is selected from left to right. 3 test probe points, respectively placing the first current probe and the second current probe on the rightmost side of the test membrane switch in the third test membrane switch group, and continuously moving the first current probe and the second current probe to the left to ensure that the first current probe and the second current probe are on the test probe points until the first current probe and the second current probe coincide with the power supply position, and recording the current value, which is recorded as the current record set of the third test membrane switch group;

[0030] S22, using the first test membrane switch group voltage record set as a test data set, using the second test membrane switch group voltage record set as a verification data set, calculating the resistance of different test probe points in the test data set to obtain a test data resistance set, fitting the test data resistance values ​​in the test data resistance set, and using a cubic spline interpolation method to optimize the fitting error to obtain a processed fitting graph, the specific steps are as follows:

[0031] S221. Calculate the average value of the test data resistance of the test data concentration , set b c represents the cth test data resistor in the test data resistor set, then the error function ; Perform quadratic polynomial fitting on the test data resistance value of the test data resistance set, calculate the fitting error, and set the fitting error threshold to , when the fitting error is less than When the test data is obtained, the quadratic polynomial fitting is the best fitting method; otherwise, the cubic polynomial fitting is used, and the cubic polynomial fitting is the best fitting method; the resistance of different test probe points in the verification data set is calculated to obtain a verification data resistance set, and the verification data resistance set is subjected to a cubic polynomial fitting to obtain a fitting graph;

[0032] S222, select nodes on the fitting graph, and record them as fitting graph node set A. 3 ={c 1 ,c 2 ,c 3 ,…,c d}, where c d represents the dth node on the fitting graph, and calculates the cubic spline function in the fitting graph that has a second-order derivative and does not contain polynomials greater than the third order, which is recorded as the cubic spline function f of the fitting graph node set 2 , substitute the nodes in the fitting graph node set into f 2 , obtaining the fitting graph node function value, connecting the fitting graph node function value and the fitting value in the fitting graph to generate a processed fitting graph;

[0033] S23, calculating the resistance of the third test membrane switch group current record set, recording it as a third resistance set, fitting the resistance values ​​in the third resistance set to obtain a comparative fitting graph, comparing the processed fitting graph with the comparative fitting graph, and setting the fitting error threshold to , when the error between the processed fitting graph and the comparison fitting graph is less than , the first resistance value in the third resistance set is used as the membrane switch resistance; otherwise, the resistance value corresponding to the highest point in the processed fitting graph is used as the membrane switch resistance, and the membrane switch resistance is used as the second membrane switch performance indicator.

[0034] The invention groups membrane switches, uses a four-probe detection method to measure current and voltage in different groups, first fits the experimental data of the first group, uses the second group of experimental data to optimize the fitting error based on the cubic spline interpolation method, and then uses the third group of experimental data to verify the accuracy of the fitting result, and obtains the membrane switch resistance, which is recorded as the second membrane switch performance indicator. The error can be effectively reduced through repeated fitting.

[0035] Preferably, S3 comprises the following steps:

[0036] S31, select m×n data detection points on the membrane switch for sampling, perform quantization processing on the data detection points, and generate a membrane switch detection matrix B 3 as follows,

[0037] ,

[0038] in, Indicates the mth horizontal and nth vertical detection point on the membrane switch;

[0039] Set voltages of different values, and combine the voltages of different values ​​into a voltage set in ascending order ,in Indicates voltage values, use the voltages in the voltage set in turn to perform a withstand voltage test on the membrane switch, and record the voltage values ​​of the detection points in the membrane switch detection matrix; set the detection threshold to , when the voltage set The voltage of the detection points in the membrane switch detection matrix is ​​greater than , and the voltage set A voltage of a voltage value is passed into the membrane switch, and the average voltage value of the detection point in the membrane switch detection matrix is ​​less than When The voltage of the voltage value is the possible withstand voltage of the membrane switch;

[0040] S32. After the withstand voltage test is completed, the smallest possible withstand voltage of all membrane switches is selected as the membrane switch withstand voltage, and the membrane switch withstand voltage is used as the third membrane switch performance indicator.

[0041] Preferably, S4 comprises the following steps:

[0042] S41: comprehensively evaluating the first membrane switch performance index, the second membrane switch performance index, and the third membrane switch performance index, and assigning a first weight β to the remaining service life of the membrane switch 1 , assigning a second weight β to the thin film switch resistance 2 , a third weight β is assigned to the withstand voltage of the membrane switch 3 , set e 1 Indicates the remaining service life of the membrane switch, e 2 represents the resistance of the membrane switch, e 3 Indicates the withstand voltage of the membrane switch. The calculation formula for the membrane switch performance index score D is as follows:

[0043] ;

[0044] The score threshold is set to σ. When the score of the membrane switch performance index is greater than σ, the membrane switch performance is qualified, otherwise the membrane switch performance is unqualified.

[0045] S42, obtaining a batch of membrane switches, calculating the remaining service life of new membrane switches, new membrane switch resistance and new membrane switch withstand voltage, and forming a new membrane switch performance parameter set ,in Indicates The remaining service life of a new membrane switch is Indicates A new thin film switching resistor, Indicates A new membrane switch withstand voltage; the new membrane switch performance parameter set is divided into a membrane switch performance training set and a membrane switch performance test set, the number of input nodes of the BP (backward propagation) neural network is set to 3, the number of output nodes is set to 1, the learning rate is δ, the node weights and node thresholds are set to and , input the membrane switch performance training set into the BP neural network, set the current iteration number to 1, the maximum iteration number to L, and when the current iteration number is equal to the maximum iteration number, stop the iteration to obtain a trained BP neural network; input the membrane switch performance test set into the trained BP neural network, use the zebra optimization algorithm to optimize the node weights and node thresholds, and obtain the final BP neural network. The specific steps are as follows:

[0046] S421, the accuracy of the trained BP neural network is used as the fitness function, the number of zebra populations in the optimization space is set to E, and the optimization upper bound of the pth zebra in the zebra population is , the optimal lower bound of the pth zebra in the zebra population is , l 1 represents a random number and , then the initial position of the pth zebra in the zebra population is The calculation formula is as follows,

[0047] ;

[0048] In the foraging stage, there are elite zebras in the zebra population. The elite zebra position corresponds to the optimal fitness function value. The elite zebra position is set to , l 2 represents a random number and , l 3 represents a random number and , represents the iterative position of the pth zebra in the zebra population, represents the fitness function value of the initial position of the pth zebra in the zebra population, represents the fitness function value of the iterative position of the pth zebra in the zebra population. The calculation formula for the iterative position of the pth zebra in the zebra population is as follows:

[0049] , ;

[0050] S422, in the defense phase, the zebra population continuously iterates the fitness function value to find the best fitness function value. At this time, the zebra chooses the escape strategy or the attack strategy. The probability of the zebra choosing the escape strategy or the attack strategy is set to ε, and the attack strategy zebra position is ,l 4 represents a random number and , represents the defense parameter and , It means the pth zebra in the zebra population iterates again. It represents the fitness function value of the iterative position of the pth zebra in the zebra population. The current number of iterations is q, and the maximum number of iterations is Q. The calculation formula for the iterative position of the pth zebra in the zebra population is as follows:

[0051] , ;

[0052] The pth zebra in the zebra population continuously changes its position to iterate the fitness function value during the foraging stage and the defense stage. When the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain the final zebra position. The rectangular coordinate values ​​of the final zebra position correspond to the node weight and the node threshold respectively.

[0053] S43, the remaining service life of the membrane switch, the membrane switch resistance and the membrane switch withstand voltage are combined into a membrane switch performance parameter set, and the membrane switch performance parameter set is input into the final BP neural network, and the final BP neural network outputs the predicted value of the remaining service life of the membrane switch, the predicted value of the membrane switch resistance and the predicted value of the membrane switch withstand voltage, thereby completing the membrane switch performance prediction.

[0054] The invention evaluates the performance of the membrane switch by combining the obtained membrane switch performance indicators and using the membrane switch performance indicator score; and uses the membrane switch performance data to train the BP neural network, and optimizes the BP neural network parameters based on the zebra optimization algorithm, so that the BP neural network converges quickly. The algorithm has a fast convergence speed and strong optimization ability; and finally, the membrane switch performance is predicted through the final BP neural network, and the neural network has a high prediction accuracy.

[0055] The present invention also discloses a system for a membrane switch performance detection method, which specifically includes: a membrane switch remaining service life module, a membrane switch resistance module, a membrane switch withstand voltage module, and a membrane switch performance evaluation and prediction module;

[0056] The membrane switch remaining service life module is used to cluster the travel of the membrane switch and calculate the remaining service life of the membrane switch using the membrane switch state transition probability matrix;

[0057] The thin film switch resistance module is used to fit the resistance value and obtain the thin film switch resistance after optimizing the fitting error;

[0058] The membrane switch withstand voltage module is used to sequentially pass different voltages into the membrane switch to obtain the membrane switch withstand voltage;

[0059] The membrane switch state transition probability matrix is ​​used to evaluate the membrane switch performance with comprehensive performance indicators, and the BP neural network is used to predict the membrane switch performance.

[0060] The present invention has the following beneficial effects:

[0061] 1. The invention first records the travel of the membrane switch by pressing the membrane switch, and clusters the travel of the membrane switch. The membrane switch travel that does not meet the conditions is recorded as a degraded state of the membrane switch, and the normal number of membrane switch travels and the degraded number of membrane switch travels are obtained, thereby predicting the remaining service life of the membrane switch.

[0062] 2. The invention measures the current and voltage of the membrane switch by grouping. The experimental data of the first group is first fitted, and the fitting error of the second group of experimental data is optimized by using the cubic spline interpolation method. The accuracy of the fitting result is verified by the third group of experimental data to obtain the resistance of the membrane switch. The result error is small through repeated fitting verification. The data monitoring points of the membrane switch are then sampled, and the voltage threshold at the data monitoring points of the membrane switch is judged by different voltages to obtain the withstand voltage of the membrane switch.

[0063] 3. The invention evaluates the performance of the membrane switch by combining the obtained membrane switch performance indicators and calculating the membrane switch performance indicator score; uses the membrane switch performance data to train the BP neural network, and optimizes the BP neural network parameters based on the zebra optimization algorithm, which can accelerate the convergence speed of the neural network and has a strong optimization ability; finally, the membrane switch performance is predicted through the final BP neural network, and the prediction accuracy of the neural network is high.

[0064] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0066] Figure 1 A schematic diagram of a flow chart of membrane switch performance detection by a membrane switch performance detection system provided by the present invention. DETAILED DESCRIPTION

[0067] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0069] This embodiment discloses a membrane switch performance detection method, which specifically includes the following contents:

[0070] S1. Select a membrane switch, record the travel of the membrane switch after pressing, cluster the travel of the membrane switch using a clustering algorithm, obtain a set of membrane switch degradation states, generate a membrane switch state transition probability matrix, calculate the remaining service life of the membrane switch, and use the remaining service life of the membrane switch as the first membrane switch performance indicator;

[0071] The S1 comprises the following steps:

[0072] S11. Select a membrane switch, denoted as membrane switch set A. 1 ={a 1 ,a 2 ,a 3 ,…,a i}, where a irepresents the i-th membrane switch selected, pressing the membrane switches in the membrane switch set, recording the travel of the membrane switches in the membrane switch set, setting that all membrane switches in the membrane switch set fail after j times of pressing, and using the travel values ​​of the membrane switches in the membrane switch set as the data elements of the membrane switch travel matrix to generate the membrane switch travel matrix B 1 as follows,

[0073] ,

[0074] Among them, a ij Indicates the j-th pressing stroke data of the i-th membrane switch;

[0075] A row in the membrane switch stroke matrix is ​​selected as a set of membrane switch strokes to be detected, and one data element is selected from the membrane switch stroke set to be detected as a cluster center, which is recorded as the first cluster center. The distance from the data element in the membrane switch stroke set to be detected to the first cluster center is calculated, and the data element corresponding to the maximum distance is used as the second cluster center. The distance from the data element in the membrane switch stroke set to be detected to the first cluster center and the second cluster center is calculated respectively, and the data element corresponding to the maximum distance is used as the third cluster center. The third cluster center satisfies that the distance from the third cluster center to the first cluster center and the distance from the second cluster center are greater than the distance from the first cluster center to the second cluster center. The data elements in the membrane switch stroke set to be detected are repeatedly selected to obtain k cluster centers to form an initial cluster center set.

[0076] For each initial cluster center in the initial cluster center set, one class is corresponding to each initial cluster center. The initial cluster center of the data element in the to-be-detected membrane switch stroke set closest to the initial cluster center set is found, and the corresponding data element in the to-be-detected membrane switch stroke set is assigned to the class corresponding to the nearest initial cluster center, and k assigned classes are obtained, which are recorded as the initial class set; the cluster center is then found from the initial class set, and S11 is repeated until the distance from the data element in the initial class set to the cluster center converges, and a final class set is obtained;

[0077] S12, clustering different membrane switch strokes according to the final class set, recording different membrane switch strokes as membrane switch resident states, and obtaining a membrane switch resident state set. ,in Indicates a membrane switch resident state; setting a degradation state threshold value as ω, extracting membrane switch resident states less than ω from the membrane switch resident state set to obtain a membrane switch degradation state set; the membrane switch degradation state in the membrane switch degradation state set Transformed into membrane switch deterioration state The probability of the membrane switch state transition probability is recorded as the membrane switch state transition probability, and the characteristic dimension of the membrane switch state transition probability is set to , generate the membrane switch state transition probability matrix B 2 as follows,

[0078] ,

[0079] in, Indicates membrane switch states, and the characteristic dimension is The membrane switch state transition probability;

[0080] A new membrane switch is obtained, and a new membrane switch stroke matrix is ​​generated after the new membrane switch is pressed. The new membrane switch stroke matrix is ​​transformed through the membrane switch state transition probability matrix, and the current state times and the degraded state times are calculated. The residence time of the new membrane switch in the current state is set to , the residence time of the new membrane switch in the degraded state is , taking the remaining service life of the membrane switch as the first membrane switch performance indicator, the remaining service life of the membrane switch is calculated as follows:

[0081] ,

[0082] Among them, C represents the remaining service life of the membrane switch, Indicates the current state i 1 The dwell time, Indicates the degradation state i 1 The dwell time, Indicates the number of degraded states.

[0083] S2. After measuring the voltage and current of the membrane switch to be measured by a four-probe detection method, the resistance value is calculated and fitted, the fitting error is optimized by a cubic spline interpolation method, and the resistance value after fitting is judged to obtain the membrane switch resistance, and the membrane switch resistance is used as the second membrane switch performance indicator;

[0084] The S2 comprises the following steps:

[0085] S21, set the four probes to be α 1 , α 2 , α 3 and α 4 , respectively denoted as a first voltage probe, a second voltage probe, a first current probe, and a second current probe; selecting a membrane switch to be measured, and dividing the membrane switch to be measured into a first test membrane switch group, a second test membrane switch group, and a third test membrane switch group;

[0086] In the first test membrane switch group, fix the power supply in the middle of the membrane switch, and select b from left to right on the outer side of the test membrane switch in the first test membrane switch group. 1 test probe points, respectively place the first voltage probe and the second voltage probe on the leftmost side of the test membrane switch in the first test membrane switch group, and continuously move the first voltage probe and the second voltage probe to the right to ensure that the first voltage probe and the second voltage probe are on the test probe points until the first voltage probe and the second voltage probe coincide with the power supply position, and record the voltage value, which is recorded as the voltage record set of the first test membrane switch group;

[0087] In the second test membrane switch group, fix the power supply at the leftmost position of the membrane switch, and select b from left to right on the inner side of the test membrane switch in the second test membrane switch group. 2 test probe points, respectively place the first voltage probe and the second voltage probe on the rightmost side of the test membrane switch in the second test membrane switch group, and continuously move the first voltage probe and the second voltage probe to the left to ensure that the first voltage probe and the second voltage probe are on the test probe points until the first voltage probe and the second voltage probe coincide with the power supply position, and record the voltage value, which is recorded as the voltage record set of the second test membrane switch group;

[0088] In the third test membrane switch group, the voltage is fixed at the leftmost position of the membrane switch, and the outer side of the test membrane switch in the third test membrane switch group is selected from left to right. 3 test probe points, respectively placing the first current probe and the second current probe on the rightmost side of the test membrane switch in the third test membrane switch group, and continuously moving the first current probe and the second current probe to the left to ensure that the first current probe and the second current probe are on the test probe points until the first current probe and the second current probe coincide with the power supply position, and recording the current value, which is recorded as the current record set of the third test membrane switch group;

[0089] S22, using the first test membrane switch group voltage record set as a test data set, using the second test membrane switch group voltage record set as a verification data set, calculating the resistance of different test probe points in the test data set to obtain a test data resistance set, fitting the test data resistance values ​​in the test data resistance set, and using a cubic spline interpolation method to optimize the fitting error to obtain a processed fitting graph, the specific steps are as follows:

[0090] S221. Calculate the average value of the test data resistance of the test data concentration , set b c represents the cth test data resistor in the test data resistor set, then the error function ; Perform quadratic polynomial fitting on the test data resistance value of the test data resistance set, calculate the fitting error, and set the fitting error threshold to , when the fitting error is less than When the test data is obtained, the quadratic polynomial fitting is the best fitting method; otherwise, the cubic polynomial fitting is used, and the cubic polynomial fitting is the best fitting method; the resistance of different test probe points in the verification data set is calculated to obtain a verification data resistance set, and the verification data resistance set is subjected to a cubic polynomial fitting to obtain a fitting graph;

[0091] S222, select nodes on the fitting graph, and record them as fitting graph node set A. 3 ={c 1 ,c 2 ,c 3 ,…,c d}, where c d represents the dth node on the fitting graph, and calculates the cubic spline function in the fitting graph that has a second-order derivative and does not contain polynomials greater than the third order, which is recorded as the cubic spline function f of the fitting graph node set 2 , substitute the nodes in the fitting graph node set into f 2 , obtaining the fitting graph node function value, connecting the fitting graph node function value and the fitting value in the fitting graph to generate a processed fitting graph;

[0092] S23, calculating the resistance of the third test membrane switch group current record set, recording it as a third resistance set, fitting the resistance values ​​in the third resistance set to obtain a comparative fitting graph, comparing the processed fitting graph with the comparative fitting graph, and setting the fitting error threshold to , when the error between the processed fitting graph and the comparison fitting graph is less than , the first resistance value in the third resistance set is used as the membrane switch resistance; otherwise, the resistance value corresponding to the highest point in the processed fitting graph is used as the membrane switch resistance, and the membrane switch resistance is used as the second membrane switch performance indicator;

[0093] S3, sampling the data detection points on the membrane switch to generate a membrane switch detection matrix, passing different voltages into the membrane switch in sequence to obtain the membrane switch withstand voltage, and using the membrane switch withstand voltage as the third membrane switch performance indicator;

[0094] The S3 comprises the following steps:

[0095] S31, select m×n data detection points on the membrane switch for sampling, perform quantization processing on the data detection points, and generate a membrane switch detection matrix B 3 as follows,

[0096] ,

[0097] in, Indicates the mth horizontal and nth vertical detection point on the membrane switch;

[0098] Set voltages of different values, and combine the voltages of different values ​​into a voltage set in ascending order ,in Indicates voltage values, use the voltages in the voltage set in turn to perform a withstand voltage test on the membrane switch, and record the voltage values ​​of the detection points in the membrane switch detection matrix; set the detection threshold to , when the voltage set The voltage of the detection points in the membrane switch detection matrix is ​​greater than , and the voltage set A voltage of a voltage value is passed into the membrane switch, and the average voltage value of the detection point in the membrane switch detection matrix is ​​less than When The voltage of the voltage value is the possible withstand voltage of the membrane switch;

[0099] S32, after the withstand voltage test is completed, the smallest possible withstand voltage of all membrane switches is selected as the membrane switch withstand voltage, and the membrane switch withstand voltage is used as the third membrane switch performance indicator;

[0100] S4, comprehensively analyzing the first membrane switch performance index, the second membrane switch performance index and the third membrane switch performance index, and using the membrane switch performance index score to evaluate the membrane switch performance; and training the BP neural network, and using the zebra optimization algorithm to optimize the BP neural network parameters to achieve the prediction of the membrane switch performance;

[0101] The S4 comprises the following steps:

[0102] S41: comprehensively evaluating the first membrane switch performance index, the second membrane switch performance index, and the third membrane switch performance index, and assigning a first weight β to the remaining service life of the membrane switch 1 , assigning a second weight β to the thin film switch resistance 2 , a third weight β is assigned to the withstand voltage of the membrane switch 3 , set e 1 Indicates the remaining service life of the membrane switch, e 2 represents the resistance of the membrane switch, e 3 Indicates the withstand voltage of the membrane switch. The calculation formula for the membrane switch performance index score D is as follows:

[0103] ;

[0104] The score threshold is set to σ. When the score of the membrane switch performance index is greater than σ, the membrane switch performance is qualified, otherwise the membrane switch performance is unqualified.

[0105] S42, obtaining a batch of membrane switches, calculating the remaining service life of new membrane switches, new membrane switch resistance and new membrane switch withstand voltage, and forming a new membrane switch performance parameter set ,in Indicates The remaining service life of a new membrane switch is Indicates A new thin film switching resistor, Indicates The new membrane switch withstand voltage is divided into a membrane switch performance parameter set and a membrane switch performance test set. The number of input nodes of the BP neural network is set to 3, the number of output nodes is set to 1, the learning rate is δ, and the node weight and node threshold are set to and , input the membrane switch performance training set into the BP neural network, set the current iteration number to 1, the maximum iteration number to L, and when the current iteration number is equal to the maximum iteration number, stop the iteration to obtain a trained BP neural network; input the membrane switch performance test set into the trained BP neural network, use the zebra optimization algorithm to optimize the node weights and node thresholds, and obtain the final BP neural network. The specific steps are as follows:

[0106] S421, the accuracy of the trained BP neural network is used as the fitness function, the number of zebra populations in the optimization space is set to E, and the optimization upper bound of the pth zebra in the zebra population is , the optimal lower bound of the pth zebra in the zebra population is , l 1 represents a random number and , then the initial position of the pth zebra in the zebra population is The calculation formula is as follows,

[0107] ;

[0108] In the foraging stage, there are elite zebras in the zebra population. The elite zebra position corresponds to the optimal fitness function value. The elite zebra position is set to , l 2 represents a random number and , l 3 represents a random number and , represents the iterative position of the pth zebra in the zebra population, represents the fitness function value of the initial position of the pth zebra in the zebra population, represents the fitness function value of the iterative position of the pth zebra in the zebra population. The calculation formula for the iterative position of the pth zebra in the zebra population is as follows:

[0109] , ;

[0110] S422, in the defense phase, the zebra population continuously iterates the fitness function value to find the best fitness function value. At this time, the zebra chooses the escape strategy or the attack strategy. The probability of the zebra choosing the escape strategy or the attack strategy is set to ε, and the attack strategy zebra position is ,l 4 represents a random number and , represents the defense parameter and , It means the pth zebra in the zebra population iterates again. It represents the fitness function value of the iterative position of the pth zebra in the zebra population. The current number of iterations is q, and the maximum number of iterations is Q. The calculation formula for the iterative position of the pth zebra in the zebra population is as follows:

[0111] , ;

[0112] The pth zebra in the zebra population continuously changes its position to iterate the fitness function value during the foraging stage and the defense stage. When the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain the final zebra position. The rectangular coordinate values ​​of the final zebra position correspond to the node weight and the node threshold respectively.

[0113] S43, the remaining service life of the membrane switch, the membrane switch resistance and the membrane switch withstand voltage are combined into a membrane switch performance parameter set, and the membrane switch performance parameter set is input into the final BP neural network, and the final BP neural network outputs the predicted value of the remaining service life of the membrane switch, the predicted value of the membrane switch resistance and the predicted value of the membrane switch withstand voltage, thereby completing the membrane switch performance prediction.

[0114] The present invention also discloses a system for a membrane switch performance detection method, which specifically includes: a membrane switch remaining service life module, a membrane switch resistance module, a membrane switch withstand voltage module, and a membrane switch performance evaluation and prediction module;

[0115] The membrane switch remaining service life module is used to cluster the travel of the membrane switch and calculate the remaining service life of the membrane switch using the membrane switch state transition probability matrix;

[0116] The thin film switch resistance module is used to fit the resistance value and obtain the thin film switch resistance after optimizing the fitting error;

[0117] The membrane switch withstand voltage module is used to sequentially pass different voltages into the membrane switch to obtain the membrane switch withstand voltage;

[0118] The membrane switch state transition probability matrix is ​​used to evaluate the membrane switch performance with comprehensive performance indicators, and the BP neural network is used to predict the membrane switch performance.

[0119] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0120] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A membrane switch performance detection method, characterized in that: The steps include: S1. Select a membrane switch, record the travel of the membrane switch after pressing, cluster the travel of the membrane switch using a clustering algorithm, obtain a set of membrane switch degradation states, generate a membrane switch state transition probability matrix, calculate the remaining service life of the membrane switch, and use the remaining service life of the membrane switch as the first membrane switch performance indicator; S2. After measuring the voltage and current of the membrane switch to be measured by a four-probe detection method, the resistance value is calculated and fitted, the fitting error is optimized by a cubic spline interpolation method, and the resistance value after fitting is judged to obtain the membrane switch resistance, and the membrane switch resistance is used as the second membrane switch performance indicator; S3, sampling the data detection points on the membrane switch to generate a membrane switch detection matrix, passing different voltages into the membrane switch in sequence to obtain the membrane switch withstand voltage, and using the membrane switch withstand voltage as the third membrane switch performance indicator; S4, comprehensively evaluating the performance of the membrane switch using the membrane switch performance index score by combining the first membrane switch performance index, the second membrane switch performance index and the third membrane switch performance index; And train the BP neural network, optimize the BP neural network parameters, and realize the prediction of membrane switch performance; S41, comprehensively analyzing the first membrane switch performance index, the second membrane switch performance index, and the third membrane switch performance index, assigning weights and calculating a membrane switch performance index score; S42, obtaining a batch of membrane switches, calculating the remaining service life of new membrane switches, new membrane switch resistance and new membrane switch withstand voltage, and forming a new membrane switch performance parameter set ,in Indicates The remaining service life of a new membrane switch is Indicates A new thin film switching resistor, Indicates The new membrane switch withstand voltage is divided into a membrane switch performance parameter set and a membrane switch performance test set. The number of input nodes of the BP neural network is set to 3, the number of output nodes is set to 1, the learning rate is δ, and the node weight and node threshold are set to and , input the membrane switch performance training set into the BP neural network, set the current iteration number to 1, the maximum iteration number to L, and when the current iteration number is equal to the maximum iteration number, stop the iteration to obtain a trained BP neural network; S43, inputting the membrane switch performance test set into the trained BP neural network, optimizing the node weights and node thresholds using the zebra optimization algorithm, and obtaining the final BP neural network; S44, the remaining service life of the membrane switch, the resistance of the membrane switch and the withstand voltage of the membrane switch are combined into a membrane switch performance parameter set, and the membrane switch performance parameter set is input into the final BP neural network, and the final BP neural network outputs a predicted value of the remaining service life of the membrane switch, a predicted value of the resistance of the membrane switch and a predicted value of the withstand voltage of the membrane switch, thereby completing the performance prediction of the membrane switch.

2. A membrane switch performance detection method according to claim 1, characterized in that: The S1 comprises the following steps: S11, selecting a membrane switch, pressing the membrane switch, and recording the travel of the membrane switch in the membrane switch set; clustering the membrane switch travel using a clustering algorithm to obtain a membrane switch resident state set; S12, generating a membrane switch state transition probability matrix according to the membrane switch resident state set; S13, obtaining a new membrane switch stroke, converting the new membrane switch stroke through a membrane switch state transition probability matrix, calculating the current state times and the degraded state times, calculating the remaining service life of the membrane switch, and using the remaining service life of the membrane switch as a first membrane switch performance indicator; A new membrane switch is obtained, and a new membrane switch stroke matrix is ​​generated after the new membrane switch is pressed. The new membrane switch stroke matrix is ​​transformed through the membrane switch state transition probability matrix, and the current state times and the degraded state times are calculated. The residence time of the new membrane switch in the current state is set to , the residence time of the new membrane switch in the degraded state is , taking the remaining service life of the membrane switch as the first membrane switch performance indicator, the remaining service life of the membrane switch is calculated as follows: , Among them, C represents the remaining service life of the membrane switch, Indicates the residence time in the current state i1, represents the residence time in the degraded state i1, Indicates the number of degraded states.

3. A membrane switch performance detection method according to claim 2, characterized in that: The S2 comprises the following steps: S21, using a four-probe detection method to measure the voltage and current of the membrane switch to be measured respectively; S22, obtaining a resistance value of the membrane switch according to the voltage and current of the membrane switch to be measured, and fitting the resistance value of the membrane switch; S23, using cubic spline interpolation method to optimize fitting error; S24, obtaining the membrane switch resistance by comparing the fitting error threshold, and using the membrane switch resistance as the second membrane switch performance indicator.

4. A membrane switch performance detection method according to claim 3, characterized in that: The S23 comprises the following steps: S231, performing cubic polynomial fitting on the resistance value of the thin film switch to obtain a fitting graph; S232, finding a cubic spline function with a second-order derivative and not including a polynomial greater than the third order in the fitting graph, obtaining a node function value of the fitting graph, connecting the node function value of the fitting graph and the fitting value in the fitting graph, and completing the optimization of the fitting error.

5. A membrane switch performance detection method according to claim 4, characterized in that: The S3 comprises the following steps: S31, selecting detection points on the membrane switch for sampling, and generating a membrane switch detection matrix; S32, performing a withstand voltage test on the membrane switch, and recording voltage values ​​of detection points in the membrane switch detection matrix, finding the withstand voltage of the membrane switch, and using the withstand voltage of the membrane switch as the third membrane switch performance indicator.

6. A membrane switch performance detection method according to claim 5, characterized in that: The score threshold is set to σ. When the score of the membrane switch performance index is greater than σ, the membrane switch performance is qualified, otherwise the membrane switch performance is unqualified.

7. A membrane switch performance detection method according to claim 6, characterized in that: The S43 comprises the following steps: S431, the accuracy of the trained BP neural network is used as the fitness function, the number of zebra populations in the optimization space is set to E, and the optimization upper bound of the pth zebra in the zebra population is , the optimal lower bound of the pth zebra in the zebra population is , l1 represents a random number and , then the initial position of the pth zebra in the zebra population is The calculation formula is as follows, ; In the foraging stage, there are elite zebras in the zebra population. The elite zebra position corresponds to the optimal fitness function value. The elite zebra position is set to , l2 represents a random number and , l3 represents a random number and , represents the iterative position of the pth zebra in the zebra population, represents the fitness function value of the initial position of the pth zebra in the zebra population, represents the fitness function value of the iterative position of the pth zebra in the zebra population. The calculation formula for the iterative position of the pth zebra in the zebra population is as follows: , ; S432, in the defense phase, the zebra population continuously iterates the fitness function value to find the best fitness function value. At this time, the zebra chooses the escape strategy or the attack strategy. The probability of the zebra choosing the escape strategy or the attack strategy is set to ε, and the attack strategy zebra position is ,l4 represents a random number and , represents the defense parameter and , It means the pth zebra in the zebra population iterates again. It represents the fitness function value of the iterative position of the pth zebra in the zebra population. The current number of iterations is q, and the maximum number of iterations is Q. The calculation formula for the iterative position of the pth zebra in the zebra population is as follows: , ; The pth zebra in the zebra population continuously changes its position to iterate the fitness function value during the foraging stage and the defense stage. When the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain the final zebra position, and the rectangular coordinate values ​​of the final zebra position correspond to the node weight and the node threshold respectively.

8. A system for implementing the membrane switch performance detection method according to any one of claims 1 to 7, characterized in that: Specifically include: Membrane switch remaining service life module, membrane switch resistance module, membrane switch withstand voltage module and membrane switch performance evaluation and prediction module; The membrane switch remaining service life module is used to cluster the travel of the membrane switch and calculate the remaining service life of the membrane switch using the membrane switch state transition probability matrix; The thin film switch resistance module is used to fit the resistance value and obtain the thin film switch resistance after optimizing the fitting error; The membrane switch withstand voltage module is used to sequentially pass different voltages into the membrane switch to obtain the membrane switch withstand voltage; The membrane switch state transition probability matrix is ​​used to evaluate the membrane switch performance with comprehensive performance indicators, and the BP neural network is used to predict the membrane switch performance.

Citation Information

Patent Citations

  • Automobile switch performance detection system and method based on artificial intelligence

    CN117849607A

  • High-voltage circuit breaker auxiliary switch internal characteristic parameter fault inversion method and system

    CN115078984A

  • Membrane switch service life detection device

    CN210513680U

  • Membrane switch circuit testing system

    US20150331048A1

  • Device and method for detecting wear in an electromechanical switching apparatus

    WO2020115313A1