Matrix controllable switch fault self-detection and life prediction system and working method

Abstract

The invention discloses a matrix controllable switch fault self-inspection and life prediction system and a working method, and belongs to the field of matrix switch non-intrusive testing, and the matrix controllable switch fault self-inspection and life prediction system comprises a test module composed of a matrix switch unit, a to-be-tested circuit assembly and a function test unit, a non-intrusive matrix self-inspection module and a life prediction and data recording module. Wherein the matrix switch unit comprises # imgabs0 # matrix switches, # imgabs1 # is the total number of test points of a circuit assembly to be tested, # imgabs2 # is the total number of test points accessed by the function test unit, and # imgabs3 # is the number of common column lines; the matrix switch unit is connected with the non-intrusive matrix self-checking module; and the non-intrusive matrix self-checking module is connected with the life prediction and data recording module. By adopting the matrix controllable switch fault self-inspection and life prediction system and the working method, high efficiency, safety and intelligence of circuit assembly testing are realized through non-intrusive self-inspection, matrix optimization, an intelligent prediction model and accurate fault positioning.

Description

Technical Field

[0001] The present invention relates to the technical field of non-invasive testing of matrix switches, and particularly to a matrix controllable switch fault self-check and life prediction system and working method. Background Art

[0002] In an automatic test system for circuit components, the fault detection of matrix controllable switches mainly relies on the technical means of real-time monitoring of the on-resistance value, and judges the health status of the switch through the change of the resistance value. However, the existing detection methods have the following significant limitations in complex systems: 1. Detection efficiency bottleneck: Traditional detection schemes (such as CN202410529278.1) adopt a serial detection mode of activating each path and measuring the resistance, and the single measurement takes up to several seconds, and manual intervention is required to switch the test channels. For a large-scale matrix system containing switch nodes, it takes several hours or even longer to complete the full detection, seriously restricting the production test efficiency.

[0003] 2. Operation safety hazard: Although CN202110588376.9 proposes a parallel detection architecture, it is still necessary to physically open the matrix switch cabinet and establish a hard-wired connection between the relay and the non-invasive matrix self-check module. This physical connection method not only significantly increases the operation complexity, but also introduces safety hazards in a high-voltage environment.

[0004] 3. Lack of prediction ability: The existing technology can only achieve post-fault detection, lacking the ability to predict the remaining life of the switch and the early warning mechanism for potential faults. Summary of the Invention

[0005] The purpose of the present invention is to provide a matrix controllable switch fault self-check and life prediction system and working method to solve the above technical problems.

[0006] To achieve the above purpose, the present invention provides a matrix controllable switch fault self-check and life prediction system, including a test module composed of a matrix switch unit, a circuit component to be tested, and a function test unit, a non-invasive matrix self-check module, and a life prediction and data recording module; wherein, the matrix switch unit includes matrix switches, is the total number of test points of the circuit component to be tested, is the total number of test points accessed by the function test unit, is the number of shared column lines; The matrix switch unit is connected to the non-invasive matrix self-check module for real-time measurement of the on-resistance of paired matrix switches; The non-invasive matrix self-check module is connected to the life prediction and data recording module for storing the time stamp and the corresponding on-resistance data, and outputting the fault warning and life prediction results.

[0007] Preferably, the matrix switch unit is a controllable switch network structure formed by root row lines and root common column lines cross-connected, and the common column lines are connected to the circuit component to be measured, and the row lines are adjacent to the function test unit; The non-invasive matrix self-test module includes two leads, and the two leads are respectively connected to the upper column line and the lower column line of the common column line; When the matrix switch unit includes an even number of common column lines, each of the two leads is connected to root common column lines; when the matrix switch unit includes an odd number of common column lines, each of the two leads is connected to and root common column lines.

[0008] Preferably, the life prediction and data recording module includes a time series database, a life prediction model, and a warning model; The time series database is used to store the on-resistance values, measurement timestamps, and corresponding switch numbers of each matrix switch; The life prediction model includes a life prediction model based on the resistance change rate and a state prediction model based on deep learning , which is used to predict the remaining life and usage status of the matrix switch unit; The warning model is used to trigger a dual warning signal according to the output result of the life prediction model.

[0009] Preferably, the state prediction model is a deep learning model.

[0010] A working method of a matrix controllable switch fault self-test and life prediction system includes the following steps: S1. Build a test module and connect the upper column line and the lower column line of the common column line of the matrix switch unit through two leads; S2. Measure the on-resistance of the matrix switch at a preset period and record the test timestamp; S3. Periodically trigger self-test: Determine whether the number of measurements of the on-resistance reaches the set number threshold. If not, return to step S2. Otherwise, trigger the self-test by monitoring the change rate of the on-resistance of the paired matrix switches; During the self-test, when the change rate of the on-resistance of the paired matrix switches is higher than the threshold, a fault mark is made; otherwise, the life prediction result and the usage status are output through the continuously recorded on-resistance and test timestamp; S4. Give a warning based on the life prediction result and the usage status.

[0011] Preferably, the set number threshold described in step S3 is 1000 times; Change rate of on-resistance The calculation formula is as follows: (1); In the formula, and are the on-resistance values when the number of test times is and respectively.

[0012] Preferably, in step S3, the life prediction model is used to output the life prediction result, and the expression of the life prediction model is as follows: (2); wherein, (3); In the formula, represents the prediction model of the on-resistance value of the paired matrix switch; , , , are all regression coefficients; , , and are all power exponents; is the initial resistance value; , , are the growth coefficients in the initial stage, the middle stage and the late stage respectively; , are the end time of the initial stage and the end time of the middle stage respectively; is the acceleration degree of the resistance increase.

[0013] Preferably, in step S3, the state prediction model is used to output the usage state, and the construction steps of the state prediction model are as follows: First step, data annotation: Annotate the usage state corresponding to the on-resistance value of each pair of matrix switches to obtain a training set, where the usage state includes normal and potential failure; Second step, input the training set into the deep learning model, and use the deep learning model to learn the relationship between the on-resistance value and the usage state until convergence to obtain the state prediction model .

[0014] Preferably, in step S4, when the state prediction model outputs potential failure, or the life prediction result output by the life prediction model is lower than the set life threshold, an early warning signal is output.

[0015] Therefore, the present invention adopts the above-mentioned matrix controllable switch fault self-checking and life prediction system and working method, and the beneficial effects are as follows: 1. Adopt a non-intrusive self-checking method, which can measure without opening the matrix switch cabinet, avoiding physical operations on the switch cabinet and ensuring the efficiency and safety of the test process; 2. By optimizing the test object and access method, two matrix switches can be accessed for self-checking each time, making the self-checking process of the matrix switch more efficient, reducing manual intervention and operations, and significantly improving the self-checking efficiency; 3. Based on historical data, establish a matrix switch life prediction model and a state prediction model , which can intelligently predict the remaining life and state changes of the switch. Through a dual warning mechanism, potential faults can be identified in advance, so as to timely detect potential faults, give a warning prompt for replacement, and avoid downtime or test errors caused by faults.

[0016] In summary, the present invention can not only realize the self-checking of matrix controllable switches, but also accurately predict the remaining service life of the switches, thereby significantly improving the self-checking efficiency and reducing the maintenance cost.

[0017] Next, through the drawings and embodiments, the technical solutions of the present invention will be further described in detail. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a principle block diagram of a matrix controllable switch fault self-checking and life prediction system according to the present invention; Figure 2 It is a graph showing the change of the on-resistance of the matrix switch over time in a matrix controllable switch fault self-checking and life prediction system according to the present invention; Figure 3 It is a flowchart of the working method of a matrix controllable switch fault self-checking and life prediction system according to the present invention; Figure 4 It is a schematic connection diagram of Test 1 in the first embodiment of the present invention; Figure 5 It is a schematic connection diagram of Test 2 in the first embodiment of the present invention; Figure 6 It is a schematic connection diagram of Test 3 in the first embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer and more understandable, the embodiments of the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention and are not used to limit the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout.

[0020] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0021] The embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0022] As Figure 1 and Figure 2 shown, a matrix controllable switch fault self-checking and life prediction system includes a test module composed of a matrix switch unit, a circuit component to be tested, and a functional test unit, a non-invasive matrix self-checking module, and a life prediction and data recording module; wherein, the matrix switch unit includes matrix switches, is the total number of test points of the circuit component to be tested, is the total number of test points accessed by the functional test unit, is the number of shared column lines. Since usually the value is small, so compared with the traditional matrix switches, a large number of matrix switches can be saved, reducing the hardware requirements and test time; the matrix switch unit is connected to the non-invasive matrix self-checking module, which is used to measure the on-resistance of paired matrix switches in real time and convert the resistance value signal into a voltage signal for output; the non-invasive matrix self-checking module is connected to the life prediction and data recording module, which is used to store the time stamp and the corresponding on-resistance data, and output a fault warning and a life prediction result.

[0023] Among them, the matrix switch unit is composed of row lines and A controllable switch network structure composed of cross-connected shared column lines, where the shared column lines are connected to the circuit components under test, and the row lines are adjacent to the functional test unit; the non-invasive matrix self-test module includes two leads, and the two leads (SC1, SC2) are respectively connected to the upper and lower column lines of the shared column lines; when the matrix switch unit includes an even number of shared column lines, each of the two leads is connected to shared column lines; when the matrix switch unit includes an odd number of shared column lines, each of the two leads is connected to and shared column lines. The matrix switch unit is divided into two parts by the two leads (SC1 and SC2). The non-invasive matrix self-test module can connect two matrix switches on different column lines for testing each time, avoiding interference and conflicts caused by the connection of shared column lines during the testing process, thereby improving the testing efficiency. And by optimizing the connection method of the matrix unit under test and only using two leads SC1 and SC2 for connection, it not only avoids the bottleneck of a single test channel but also effectively reduces repeated testing and circuit load, improving the overall testing ability and accuracy of the system.

[0024] The life prediction and data recording module includes a time series database, a life prediction model, and an early warning model; the time series database is used to store the on-resistance values, measurement timestamps, and corresponding switch numbers of each matrix switch; the life prediction model includes a life prediction model based on the resistance change rate and a state prediction model based on deep learning , which are used to predict the remaining life and usage status of the matrix switch unit; the early warning model is used to trigger a dual early warning signal according to the output result of the life prediction model.

[0025] The state prediction model is a deep learning model.

[0026] As Figure 3 shown, a working method of a matrix controllable switch fault self-test and life prediction system includes the following steps: S1. Build a test module and connect the upper and lower column lines of the shared column lines of the matrix switch unit through two leads; S2. Measure the on-resistance of the matrix switch at a preset period and record the test timestamp; S3. Periodically trigger self-test: Determine whether the number of measurements of the on-resistance reaches the set number threshold. If not, return to step S2; otherwise, trigger the self-test by monitoring the change rate of the on-resistance of the paired matrix switches; During the self-test, when the change rate of the on-resistance of the paired matrix switches is higher than the threshold, perform a fault mark and issue an alarm; otherwise, output the life prediction result and usage status through the continuously recorded on-resistance and test timestamp. The set number threshold described in step S3 is 1000 times; Conduction resistance change rate The calculation formula is as follows: (1); In the formula, and are the conduction resistance values when the test times are (representing the usage cycle of the matrix switch (unit: times)) and respectively.

[0027] In step S3, the life prediction model outputs the life prediction result (that is, a life prediction model is established based on the mutated conduction resistance signal ), and the expression of the life prediction model is as follows: (2); Among them, (3); In the formula, represents the prediction model of the conduction resistance value of the paired matrix switch (considering that the resistance value has different change patterns in different usage stages (initial, middle, and late) of the switch, so the change of the resistance value is divided into three stages, and a non-linear model is used to fit the resistance change law in each stage. In the initial usage stage of the switch: low-speed growth. In the initial stage when the matrix switch is put into use, the change of the resistance value is usually slow, showing a linear or slow growth trend. At this time, the change of the resistance is mainly affected by factors such as initial wear and oxidation of the contact surface, and the growth is relatively stable, so ; in the middle usage stage of the switch: relatively balanced. Entering the middle stage, the change of the resistance value is relatively stable, and the amplitude of the resistance change is even smaller than that in the initial stage. At this time, the resistance growth is no longer as significant as in the initial stage. In this stage, the resistance change tends to be stable, and the change speed begins to slow down, so there is ; in the late usage stage of the switch: accelerated growth. In the late stage of the switch, the growth of the resistance value intensifies, especially when there is severe wear of the contact surface, contact oxidation, ablation, or material aging, etc., the growth speed of the resistance significantly accelerates. The resistance change in this stage conforms to the law of accelerated growth, so there is ; and and and are all regression coefficients, which are obtained by fitting according to historical data; and and and They are all power exponents used to describe the non-linear relationship among the number of tests, the resistance value, and the remaining number of uses, which are determined by regression analysis or machine learning methods based on actual resistance data and switch types; is the initial resistance value; , , are the growth coefficients in the initial stage, the middle stage, and the late stage respectively, which are determined by regression analysis or machine learning methods based on actual resistance data and switch types; , are the end time of the initial stage and the end time of the middle stage respectively; is the acceleration degree of resistance increase.

[0028] In step S3, the state prediction model is used to output the usage state, and the construction steps of the state prediction model are as follows: The first step, data annotation: Annotate the usage state corresponding to each pair of matrix switch conduction resistance values to obtain a training set, where the usage state includes normal and potential faults, and the annotation data can be actual fault records, maintenance records, or remaining life evaluations; The second step, input the training set into the deep learning model, and use the deep learning model to learn the relationship between the conduction resistance value and the usage state until convergence to obtain the state prediction model ; Input the resistance value of the current matrix switch into the state prediction model to obtain its current usage state, and when the usage state is potential fault, an early warning signal is sent to prompt the maintenance personnel to perform maintenance or replacement in time to avoid downtime or test errors caused by faults.

[0029] S4. Perform early warning based on the life prediction result and the usage state.

[0030] In step S4, when the state prediction model outputs a potential fault, or the life prediction result output by the life prediction model is lower than the set life threshold, an early warning signal is output.

[0031] To further disclose the present invention, Supplementary Example 1 and Supplementary Example 2 are provided, and in the above two examples, is the current in the circuit; is the basic current source; 10 is the 10-fold basic current source; 20 is the 20-fold basic current source; Sel0, Sel1, and Sel2 are the control switches of the basic current source, the 10-fold basic current source, and the 20-fold basic current source respectively; SC1 is a lead from the non-invasive matrix self-test module to the matrix switch unit M; SC2 is another lead from the non-invasive matrix self-test module to the matrix switch unit M; is the total number of test points of the circuit components to be tested in the test module; is the total number of test points accessed by the function test unit; is the number of common column lines, and is also the maximum number of connection lines that need to be connected to the test module in the function test unit; C0~C (c-1) are the common column lines of the matrix switch; is the wire resistance of the constant current source passing through the SC2 lead to the matrix switch of the matrix switch unit M, and its value is a constant; The line resistance of the constant current source passing through the SC2 lead to the matrix switch of the matrix switch unit M, and its value is a constant; are all the resistance values of the matrix switches in the matrix switch unit M, and their values are constants; is the circuit resistance, and its value is a constant; If is an even number, then is the resistance value of the matrix switch connected to SC1 in the test module, is the resistance value of the matrix switch connected to SC2 in the test module, is the line resistance between the matrix switch of the matrix switch unit M connected to SC2 and the matrix switch of the test module, is the line resistance between the matrix switch of the matrix switch unit M connected to SC1 and the matrix switch of the test module, is the line resistance between the two matrix switches of the test module, and the above values are all constants; If is an odd number, then is the resistance value of the matrix switch connected to SC1 in the test module, is the resistance value of the matrix switch connected to SC2 in the test module, is the line resistance between the matrix switch of the matrix switch unit M connected to SC2 and the matrix switch of the test module, is the line resistance between the matrix switch of the matrix switch unit M connected to SC1 and the matrix switch of the test module, is the line resistance between the two matrix switches of the test module, and the above values are all constants; is the output voltage signal of the matrix non-invasive matrix self-test module; GND represents the ground wire.

[0032] Embodiment 1 Such as Figure 4As shown, (Test 1) this is the first self-check of the matrix switch, and one of the switches is damaged. At this time, the life prediction model has not been established yet. and the state prediction model ; and the total number of test points of the circuit component to be tested is 200, the total number of test points accessed for functional testing is 8, and the total number of shared column lines is , is an even number, then the total number of switches is , and at this time, SC1 and SC2 can be respectively connected to the upper and lower , column lines.

[0033] The matrix non-invasive matrix self-check module includes three constant current sources that can be configured by selecting through Sel0, Sel1, and Sel2 switches. During the test, a stable current signal is provided through configuration to prevent the final output voltage from being too large due to the damage of the matrix switch and the increase in the switch resistance.

[0034] Moreover, since the on-resistance of the matrix switch gradually increases under the influence of factors such as mechanical fatigue, material aging, contact oxidation, working environment temperature, excitation voltage, and current, the resistance value of the matrix switch can be converted into a voltage signal and output through the circuit. Because a single switch cannot be measured separately, the non-invasive matrix self-check module connects two matrix switches for self-check operation each time. The test output can be obtained according to Ohm's law through the constant current source and the resistance value in the circuit: ;

[0035] In the formula, , , , and are all constants, and in this embodiment, their sum is denoted as C. Let and adopt the same controllable switches as the matrix. It is set that every 1000 tests, the matrix non-invasive matrix self-check module performs 1 self-check. Obviously, the number of times the controllable switches in the matrix switch unit M are used is only one-thousandth of the controllable switches in the two matrices of the test module. Assuming that the on-resistance of these switches is constant during the full life cycle test of this self-check device is reasonable, then , and are all constants. Therefore, the output voltage Vout is as follows: ;

[0036] In the formula, , and All are constants, and their sum is denoted as D. These three resistance values are only related to the wiring of the matrix non-invasive matrix self-test module and will not change once the device is implemented. Thus, it can be known that ;

[0037] That is, is only determined by and .

[0038] Then it can be deduced that: ;

[0039] If one of the two switches being tested is damaged, that is, when the of the test output is too large, the matrix non-invasive matrix self-test module will immediately issue an alarm and mark the fault to avoid test errors caused by these problems.

[0040] Furthermore, according to its resistance value , using the multiple measurement method, as shown in Figure 5 , disconnect the matrix switch connected to SC1 and close the other matrix switch (Test 2); as shown in Figure 6 , disconnect the matrix switch connected to SC2 and close the other matrix switch (Test 3). The on-resistance of each controllable switch is included in the two tests. According to the results of the three self-test measurements, it can be obtained that: ; ; ;

[0041] In the results of the three measurements, and both appear twice. If only one of the two controllable switches mutates, assuming the damaged switch situation is is A and the normal situation is is B, taking the switch corresponding to being damaged as an example, ; ; ;

[0042] In the two tests involving , the switch damage situation appears in both tests, while only appears in the switch damage situation when jointly participating in the test with , thus realizing that only relying on the three self-test measurements can determine the damage of the switch corresponding to .

[0043] Embodiment 2 This embodiment is based on Embodiment 1 and adds a life prediction model and a state prediction model , and conducts state and life prediction of the matrix switch based on parameters such as the on-resistance value of the matrix switch, the number of test times , and the on-resistance change rate of the matrix switch

[0044] After the matrix self-check unit conducts a test, the matrix switch resistance value at this time is input into the state prediction model , and the state of the pair of matrix switches can be obtained as "normal" or "potential failure". If the state prediction model predicts that the switch state is "normal", the system will continue to monitor the change of the on-resistance value and regularly update the life prediction model to ensure that the switch operates within the normal working range. If the state prediction model predicts that the switch state is "potential failure", the system will immediately issue a warning signal to remind the maintenance personnel to check and consider replacing the pair of matrix switches. This early warning mechanism can effectively avoid system downtime or test errors caused by switch failures

[0045] And after the state prediction is completed, the system will input the measured on-resistance value and the current number of test times into the life prediction model , and use the life prediction model to output the remaining number of usage times (service life) of each matrix switch, and compare the output remaining number of usage times with the set number threshold to determine whether there is a potential failure. If so, an early warning signal is output

[0046] In this embodiment, after each test, the system will add the new on-resistance value , the number of test times and data such as the resistance value change rate to the historical data, and retrain and update the life prediction model and the state prediction model . By continuously updating the model, the system can adapt to the changes of the switch in different working environments and improve the accuracy and reliability of the prediction

[0047] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that: they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention

Claims

1. A matrix controllable switch fault self-detection and life prediction system, characterized by: The invention comprises a test module composed of a matrix switch unit, a circuit component to be tested and a functional test unit, a non-intrusive matrix self-test module and a life prediction and data recording module; wherein the matrix switch unit comprises Matrix switches, is the total number of test points of the circuit component to be tested, The total number of test points accessed by the functional test unit, is the number of shared column lines; The matrix switch unit is connected to a non-intrusive matrix self-test module for real-time measurement of the on-resistance of paired matrix switches; The non-intrusive matrix self-check module is connected to the life prediction and data recording module, and is used to store timestamps and corresponding on-resistance data, and output fault warnings and life prediction results.

2. A matrix controllable switch fault self-detection and life prediction system according to claim 1, characterized in that: The matrix switch unit is Root line and A controllable switch network structure composed of cross-connected common column lines, the common column lines are connected to the circuit components to be tested, and the row lines are adjacent to the functional test units; The non-intrusive matrix self-test module includes two leads, and the two leads are respectively connected to an upper column line and a lower column line of a common column line; When the matrix switch unit includes an even number of common column lines, two leads are connected to each When the matrix switch unit includes an odd number of shared column lines, two leads are connected to each as well as Root common column lines.

3. A matrix controllable switch fault self-detection and life prediction system according to claim 2, characterized in that: The life prediction and data recording module includes a time series database, a life prediction model, and an early warning model; The timing database is used to store the on-resistance value, measurement timestamp and corresponding switch number of each matrix switch; Life prediction models include life prediction models based on resistance change rate and state prediction models based on deep learning , used to predict the remaining life and usage status of matrix switch units; The early warning model is used to trigger a dual early warning signal according to the output of the life prediction model.

4. A matrix controllable switch fault self-detection and life prediction system according to claim 3, characterized in that: State prediction model For deep learning models.

5. A working method of a matrix controllable switch fault self-detection and life prediction system according to claim 4, characterized in that: The following steps are involved: S1. Build a test module and connect the upper column line and the lower column line of the common column line of the matrix switch unit through two leads; S2, measuring the on-resistance of the matrix switch according to a preset period and recording the test timestamp; S3, periodically trigger self-check: determine whether the number of on-resistance measurements reaches a set number threshold, if not, return to step S2, otherwise trigger monitoring of the on-resistance change rate of the paired matrix switches to perform self-check; During the self-test process, when the rate of change of the on-resistance of the paired matrix switches is higher than a threshold, a fault is marked; Otherwise, the lifetime prediction results and usage status are output through the continuously recorded on-resistance and test timestamp; S4. Provide early warning based on life prediction results and usage status.

6. The working method of the matrix controllable switch fault self-detection and life prediction system according to claim 5 is characterized by: The setting number threshold described in step S3 is 1000 times; On-resistance change rate The calculation formula is as follows: （1）； In the formula, , The number of tests is and The on-resistance value when .

7. The working method of the matrix controllable switch fault self-detection and life prediction system according to claim 6 is characterized by: In step S3, the life prediction model is used Output life prediction results and life prediction model The expression is as follows: （2）； in, （3）； In the formula, A prediction model for on-resistance values ​​of paired matrix switches; , , , All are regression coefficients; , , and All are power exponents; is the initial resistance value; , , are the growth coefficients for the initial stage, the middle stage, and the late stage, respectively; , They are the end time of the initial stage and the end time of the middle stage respectively; The acceleration degree of resistance increase.

8. The working method of the matrix controllable switch fault self-detection and life prediction system according to claim 7 is characterized by: In step S3, the state prediction model is used Output usage status and status prediction model The construction steps are as follows: Step 1: Data labeling: label the usage status corresponding to the on-resistance value of each pair of matrix switches to obtain a training set, where the usage status includes normal and potential faults; The second step is to input the training set into the deep learning model, and use the deep learning model to learn the relationship between the on-resistance value and the usage status until convergence to obtain the status prediction model .

9. The working method of the matrix controllable switch fault self-detection and life prediction system according to claim 8, characterized in that: In step S4, when the state prediction model Output potential failure or life prediction model When the output life prediction result is lower than the set life threshold, an early warning signal is output.

Images