Man-machine collaborative high-real-time function test cross-domain optimization method and device and medium

Abstract

The invention discloses a man-machine collaborative high-real-time function test cross-domain optimization method and device and a medium, and relates to the technical field of electronic information, and the method comprises the steps: expanding a discrete optimization model into a continuous optimization model, the probability that multiple intermediate events in the fault tree fail at the same time is set to be the minimum value of the test proportion corresponding to all the intermediate events contained in the intersection item, and a continuous optimization model based on intersection item consideration is obtained; the method comprises the following steps: performing theoretical analysis on an inherent reliability calculation formula of a system, and cutting off the inherent reliability calculation formula of the system in combination with expert experience to obtain a continuous optimization model based on real-time consideration; a continuous optimization model based on additional constraints is obtained by additionally establishing a relation between decision variables and inherent reliability of the system; solving the continuous optimization model based on the additional constraint to obtain an optimal test strategy; according to the function test cross-domain optimization method and device and the medium, the real-time problem of model solving is solved, and the cross-domain overall optimization capability of the model is improved.

Description

Technical Field

[0001] The present invention relates to the field of electronic information technology, and particularly to a cross-domain optimization method, device and medium for high real-time function testing of human-machine collaboration. Background Art

[0002] Under the research and application background of flexible manufacturing, the existing test item selection method traces the root cause of product quality by integrating the expert experience of multiple process domains, improves the calculation accuracy of system reliability, and designs a main board function test strategy with cross-domain overall optimization ability. However, most of the existing methods establish discrete optimization models, that is, they often only select test items and do not perform selective testing on samples. In actual use, the test strategy obtained by using such methods either tests all samples of each test item or does not test any of them. When the attribute values of test items have polarized differences, such methods can effectively reduce the test cost. However, in reality, the attribute values of test items are not distributed in a polarized manner, and the values of many attributes are between large and small values. If the discrete optimization method is still used, if the obtained test strategy tests all samples of such attributes, it will increase the test time significantly; if such attribute samples are not tested at all, it is easy to lead to a high missed detection rate. Thus, it can be seen that the discrete optimization model is difficult to further reduce the test cost.

[0003] To solve the above problems, a discrete optimization model is extended to a continuous optimization model. The extended model realizes high-precision calculation of system reliability and further improves the cross-domain overall optimization ability of the model. However, due to the increase in the complexity of the reliability calculation formula, the solution time of the model increases significantly, making the model unable to guarantee the real-time solution of the optimization model while maintaining the cross-domain overall optimization ability. The flexible manufacturing production mode of multi-variety and small-batch has the characteristics of frequent product model switching and frequent batch changes, which puts high standards and strict requirements on the rapid adaptation of the test strategy design of the test line. In this context, the existing extended model is difficult to meet the requirement of rapid test strategy design in the main board function test process, forming a new technical bottleneck. Summary of the Invention

[0004] Based on the technical problems existing in the background art, the present invention proposes a cross-domain overall optimization method for high real-time function testing of human-machine collaboration, which solves the problem of real-time solution of the model by integrating expert experience in the reliability analysis process, and further improves the cross-domain overall optimization ability of the model.

[0005] The cross-domain optimization method for high real-time function testing of human-machine collaboration proposed by the present invention includes: Expand the discrete optimization model into a continuous optimization model, establish a fault tree, and set the probability of simultaneous failures of multiple intermediate events in the fault tree to the minimum of the test ratios corresponding to all intermediate events included in the intersection term, thereby obtaining a continuous optimization model considering the intersection term; Through theoretical analysis of the formula for calculating the inherent reliability of the system and truncating the formula for calculating the inherent reliability of the system in combination with expert experience, a continuous optimization model considering real-time performance is obtained; By additionally establishing the relationship between the test ratio of each test item and the inherent reliability of the system, a continuous optimization model with additional constraints is obtained; Solve the continuous optimization model with additional constraints and screen the test strategy based on the training set to obtain the optimal test strategy.

[0006] Further, in the fault tree, regard the motherboard failure as the top event, regard the failure of the test item as the intermediate event, and regard the underlying cause of the test item failure as the bottom event of the fault tree.

[0007] Further, in the process of expanding the discrete optimization model into a continuous optimization model, specifically: Through reliability analysis of the motherboard function test process, determine the relationship between the fault coverage rate and the inherent reliability of the system; Based on fault tree analysis, analyze the impact on the inherent reliability of the system when intermediate events containing the same bottom event fail concurrently; Derive the system reliability of the motherboard function test process, thereby constructing a continuous optimization model.

[0008] Further, through theoretical analysis of the formula for calculating the inherent reliability of the system and truncating the formula for calculating the inherent reliability of the system in combination with expert experience, a continuous optimization model considering real-time performance is obtained, specifically: Since the fault tree analysis method analyzes the system reliability from the perspective of the system unreliability, and the sum of the inherent reliability of the system and the inherent unreliability of the system is always equal to 1, therefore, analyzing the formula for calculating the inherent unreliability of the system is equivalent to analyzing the formula for calculating the inherent reliability of the system. They are essentially equivalent.

[0009] Divide the formula for calculating the inherent unreliability of the system into sub-items, the number of sub-items is the same as the number of test items, the th sub-item contains small items, the number of small items contained in the th sub-item and the th sub-item is the same, and divide the two sub-items into a group; By analyzing the expression of a group of sub-items after division, it is obtained that there is a phenomenon of mutual cancellation of small items contained in some sub-items in the formula for calculating the inherent unreliability of the system; The calculation formula of the system's inherent reliability is truncated by combining expert experience to obtain a continuous optimization model considering real-time performance, thereby reducing the complexity of the continuous optimization model while retaining the relevant information between measurement items.

[0010] Furthermore, by additionally establishing the relationship between the test ratio of each measurement item and the system's inherent reliability, a continuous optimization model based on additional constraints is obtained, specifically: Based on the maintenance data of historical test data, the occurrence probability of the basic event is obtained as the unreliability of the basic event. Combining expert experience, the mean and deviation of the occurrence probability of the basic event are obtained, and then the reliability range of the basic event is determined; Based on the reliability range of the basic event, the system's inherent reliability range of the intermediate event corresponding to each measurement item is quantitatively calculated using the fault tree, and based on this, the optimization interval of the test ratio of each measurement item is set as a function based on the system's inherent reliability.

[0011] Furthermore, the optimization interval of the test ratio is specifically: ; where and are respectively the lower limit and upper limit of the system's inherent reliability of the th measurement item, and are the proportionality coefficients of the lower limit and upper limit of the test ratio and the system's inherent unreliability, which are set according to expert experience.

[0012] Furthermore, the continuous optimization model based on additional constraints is specifically: ; ; ; where is the test ratio, is the system's inherent reliability obtained from the truncated system's inherent reliability calculation formula, represents the average test time of the th measurement item, represents the test time threshold of each main board, is the total number of measurement items, is the test ratio of the th measurement item, and are respectively the lower limit and upper limit of the system's inherent reliability, and are the proportionality coefficients of the lower limit and upper limit of the test ratio and the system's inherent unreliability, which are set according to expert experience, Represents an all - 1 array of length .

[0013] Furthermore, three evaluation metrics are set to verify the effectiveness of the cross - domain overall optimization method for motherboard function testing. The evaluation metrics include the average motherboard test time, the false negative rate, and the total average motherboard test cost.

[0014] Furthermore, the formula for the false negative rate is as follows: ; where represents the correspondence between the test strategy randomly generated at the test ratio and the test status of each test item of each motherboard for the th group, is the false negative rate of the th group correspondence corresponding to the test ratio on the training set or test set.

[0015] Furthermore, the formula for the average motherboard test time is as follows: ; where represents the correspondence between the test strategy randomly generated at the test ratio and the test status of each test item of each motherboard for the th group, is the average test time of the th group correspondence corresponding to the test ratio on the training set or test set.

[0016] A computer device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above - described method is implemented.

[0017] A computer - readable storage medium stores a number of classification programs, and the number of classification programs is used to be called by a processor and execute the above - described method.

[0018] ​The advantages of the cross-domain optimization method, device and medium for high-real-time human-machine collaborative function testing provided by the present invention are as follows: expanding the discrete optimization model into a continuous optimization model, selecting a reasonable probability value for the simultaneous failure of relevant modules, and reasonably truncating the formula for calculating the inherent reliability of the system by incorporating expert experience into the reliability analysis process, significantly improving the real-time performance of the solution of the optimization model while ensuring the calculation accuracy of the reliability, and making the test strategy design more reasonable; providing a feasible path for quickly designing a motherboard function test strategy with cross-domain overall optimization ability, and contributing to cost reduction and efficiency improvement in the flexible manufacturing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flowchart of the present invention; Figure 2 is a fault tree diagram. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] Next, the technical solutions of the present invention will be described in detail through specific embodiments. Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0021] As Figure 1 and 2 shown, a cross-domain overall optimization method for high-real-time motherboard function testing with human-machine collaboration proposed by the present invention includes steps one to four: Step 1: Expand the discrete optimization model into a continuous optimization model, establish a fault tree, and set the probability of simultaneous failure of multiple intermediate events in the fault tree to the minimum value of the test ratios corresponding to all intermediate events included in the intersection term, to obtain a continuous optimization model considering the intersection term; Step 2: Through theoretical analysis of the formula for calculating the inherent reliability of the system, and combining expert experience to truncate the formula for calculating the inherent reliability of the system, to obtain a continuous optimization model considering real-time performance; Step 3: By additionally establishing the relationship between the test ratio of each test item and the inherent reliability of the system, to obtain a continuous optimization model based on additional constraints; Step 4: Solve the continuous optimization model based on additional constraints, and screen the test strategy based on the training set to obtain the optimal test strategy.

[0022] In view of the problem that the existing methods are difficult to ensure the real-time solution of the optimization model while maintaining the cross-domain overall optimization ability, resulting in difficulty in meeting the requirements of the main board function test process for rapid strategy design, this embodiment proposes a cross-domain overall optimization method for main board function test. By integrating expert experience into the reliability analysis process and reasonably truncating the calculation formula of the inherent reliability of the system, the real-time solution of the optimization model is significantly improved while ensuring the calculation accuracy of the reliability, and the test strategy design is made more reasonable. This embodiment provides a feasible path for rapidly designing a main board function test strategy with cross-domain overall optimization ability, which helps to reduce costs and increase efficiency in the flexible manufacturing process.

[0023] In one embodiment, step 1: Expand the discrete optimization model into a continuous optimization model, establish a fault tree, and set the intersection result of two intermediate events in the fault tree as the minimum value of the test ratios corresponding to all intermediate events included in the intersection term, to obtain a continuous optimization model considering the intersection term, specifically: Indicators such as the yield often only reflect the reliability of the main board components in a short period. Although the yield fluctuates frequently, the reliability of the main board components determined by the fixed manufacturing process has a definite value. Considering that the reliability of the main board function test process depends on the reliability of the tested main board components, if the fault coverage rate and other indicators are calculated based on the reliability of the main board function test process and an optimization model is established on this basis, the obtained optimal solution is more likely to have stronger generalization ability than the existing methods.

[0024] The function test is divided into two stages: the main board function test and the finished product function test. In the main board function test stage, the test strategy determines the test status of each test item. At the same time, according to the test strategy, the test items are divided into two categories: the tested items and the untested items. Only when all the tested items of the main board pass the test, the main board will be judged as a good product; if there is any test item that fails, it will be judged as a defective product and sent to the repair center for subsequent repair.

[0025] It should be noted that if there are untested fault items on the main board (i.e., potential defects not covered by the test strategy), the main board will be misjudged as a good product, and this situation is called a false negative main board. Such main boards will be assembled into finished products in the subsequent production process and detected as defective products in the finished product function test stage. Since the finished products need to be disassembled and repaired, and the repair cost of false negative main boards is significantly higher than that of regular defective products, an excessively high false negative rate will directly lead to an increase in the overall test cost. The finished product function test stage will conduct a full-function test on all assembled finished products to ensure that there is no missed detection in this stage and all function defects can be effectively detected.

[0026] In the main board functional test stage, each test item adopts a serial test method and is independent of each other. Therefore, the entire test system can be modeled as a reliability model consisting of two series subsystems: one subsystem contains all the tested items, and the other subsystem contains all the untested items. The overall test reliability of the main board depends on the combined action of these two subsystems.

[0027] Analysis shows that if a fault of a certain main board occurs exactly in the tested items, then this fault will surely be detected, and the main board will be accurately judged as a defective product; on the contrary, the generation of false-negative main boards completely stems from the faults in the untested items not being detected. Therefore, for the finished product functional test, the tested item subsystem in the main board functional test can be regarded as completely reliable (reliability is 100%), and the unreliability of the entire test process is only determined by the untested item subsystem. Therefore, the inherent reliability of the system is calculated as follows ; (1) where is the total number of test items included in the test strategy in, is the test item index, is the test strategy in the th test item, represents the th test item whose test start signal is not activated.

[0028] In step one, through the reliability analysis of the functional test process, it is proposed that using the inherent reliability of the system as a characterization index of the fault coverage rate is more accurate. Based on the basic principles of reliability theory, the inherent reliability of the system can be expressed as the product of the reliabilities of each component unit. However, there is a limitation in the existing analysis method when calculating the inherent reliability of the system: it fails to fully consider the impact of concurrent faults of modules containing the same basic event on the overall reliability of the system.

[0029] Specifically: In the fault tree modeling, we set the main board fault as the top event, the test item fault as the intermediate event, and the underlying factors causing the test item fault as the basic events. When using the fault tree method for quantitative calculation, the intermediate events are regarded as independent of each other, ignoring the influence of repeated events in the minimum cut sets on the calculation results. This simplified treatment will lead to significant deviations in the calculation of the inherent reliability of the system and ultimately affect the optimization effect of the test strategy.

[0030] Taking Figure 2 the shown fault tree as an example, specifically illustrate the calculation method of the top event occurrence probability considering repeated events. In this fault tree, each event is connected by an OR gate, and the top event The failure of may be caused by the failures of events M1, M2, or M3. By analyzing the fault tree structure, it can be found that events M1 and M2 contain the same event M5. Similarly, events M1, M2, and M3 also all contain event M5.

[0031] To solve the problem of calculating the concurrent failure probability of modules containing the same basic events, the module association information shown in the fault tree is utilized. Specifically, the occurrence probability of event M5 is used to represent the probability that M1 and M2 occur simultaneously. This method effectively solves the calculation deviation problem in the existing methods.

[0032] In this embodiment, by deeply analyzing the fault tree structure characteristics, quantitatively analyzing the impact of the concurrent failures of modules containing the same basic events on the system reliability, the following formula for calculating the inherent unreliability of the system in the functional test process is derived: ; (1) ; (2) Among them, is the top event is the occurrence probability, that is, the inherent unreliability of the system, is the total number of test items, is the test item index, is the th test item, is that the test start signal of the th test item is not activated, are respectively that the test start signals of the th test item are not activated.

[0033] On this basis, the following optimization model is established: ; (3) ; (4) ; (5) Among them, is the test strategy, represents the average test time of the th test item, represents the test time threshold for each main board, is that the test start signal of the th test item is activated.

[0034] Since the test status in the test item selection method is single, and each test item only has two states: tested and not tested, this binary classification of test status is effective for test items that are very necessary to test and those that are not necessary to test. However, for test items whose test necessity lies between the two, whether they are set to be tested or not, it is likely to make it difficult to further reduce the overall test cost. Therefore, this embodiment proposes an overall cross-domain optimization method for motherboard function testing, which improves the discrete test to a continuous test, and each test item tests a part of the motherboard.

[0035] Specifically: Let be the test ratio of each test item, then the ratio of the th test item that is not tested is , Let and be the test ratios of the th and th test items respectively, then the unreliability of the untested part of the th test item can be expressed as: ; (6) Among them, represents the inherent reliability of the system of the

[0036] th test item, which is determined by the reliability of the motherboard components corresponding to the test item, and is expressed as the probability that the

[0037] th test item passes the test. Set the defective motherboard in the untested part of the motherboard function test process as the top event , and the occurrence of a fault in the untested part of the th test item as the bottom event , and the occurrence probability of the bottom event is denoted as , then the occurrence probability of the top event is expressed as: ; (8) ; (9) After the evolution of formulas (7) to (9), the formula for calculating the inherent unreliability of the system of formula (9) is finally obtained. Among them, All are measurement item indexes, It means that taking the failure of the untested part in the th measurement item as the corresponding basic event, It refers to the test ratio corresponding to the simultaneous occurrence of the and th measurement items, In it represents the test ratio when the events fail simultaneously, It represents the occurrence probability of the simultaneous occurrence of the measurement item , is the basic event occurring simultaneously, It represents the probability of the th measurement item failing, which is used to represent the estimated value of the unreliability of the th intermediate event , The relationship with the reliability of the measurement item is as follows: ; (10) Among them, represents the test ratio corresponding to the simultaneous occurrence of multiple events, which is closely related to the occurrence probabilities of all events containing the same basic event in this measurement item (i.e., intermediate event). Taking in formula (9) as an example for illustration. It represents the test ratio corresponding to the simultaneous occurrence of the measurement item . Since the test ratios of the measurement items corresponding to the intermediate events (measurement items) , are different, it is impossible to accurately determine the test ratio corresponding to the intersection result of the two intermediate events. Therefore, it is set to the minimum value of the test ratios of all intermediate events included in the intersection item, which can be expressed as: ; (11) are respectively the test ratios of the and th measurement items (intermediate events).

[0038] Based on the above analysis, a continuous optimization model considering the intersection item is established: ; (12) ; (13) ; (14) Among them, is the test ratio, is the inherent reliability of the system, that is, formula (12) is the reliability calculation formula, It represents the The average test time of each test item, represents the test time threshold for each main board, is the total number of test items, is the test ratio of the th test item,

[0039] By expanding the discrete test state to a continuous test state and selecting reasonable probability values for the simultaneous failures of relevant modules, the continuous optimization model considering the intersection terms realizes the high-precision calculation of the system reliability, further optimizes the test strategy, and further reduces the test cost.

[0040] In one embodiment, in step two, through theoretical analysis of the formula for the inherent reliability of the system and truncating the formula for the inherent reliability of the system in combination with expert experience, a continuous optimization model considering real-time performance is obtained, specifically: To achieve high-precision calculation of reliability, in step one, the discrete optimization model is expanded into a continuous optimization model, which increases the computational complexity of the formula for the inherent reliability of the system, and the computational scale increases with the increase in the number of test items, resulting in a long solution time for the continuous optimization model, seriously affecting the real-time performance of obtaining the test strategy and making it difficult to meet the requirement of the main board function test process for rapid strategy design.

[0041] Since the fault tree analysis method analyzes the system reliability from the perspective of the system unreliability, and the sum of the system inherent reliability and the system inherent unreliability is always equal to 1, analyzing the formula for the system inherent unreliability is essentially equivalent to analyzing the formula for the system inherent reliability. Therefore, to solve the problem of balancing the contradiction between the calculation accuracy and the computational complexity of the system reliability, based on expert guidance, the formula (9) for the system inherent unreliability is theoretically analyzed, and the analysis process is as follows: First, divide the structure of formula (9). Using the symbol "∑" as the distinction, the formula is divided into sub-items, and the number of sub-items is the same as the number of test items. The th sub-item contains small items. The number of small items in the th sub-item and the th sub-item is the same, and the two sub-items are divided into a group. Taking the th item, the th item and the th item, the th item as an example for analysis, where .

[0042] The expression of the th sub-item is: ; (15) Expanding formula (15) gives formula (16): ; (16) The -th term expression is:[[]] ; (17) Expanding formula (17) gives formula (18): ; (18) The -th term expression is:[[]] ; (19) Expanding formula (19) gives formula (20): ; (20) The -th term expression is:[[]] ; (21) Expanding formula (21) gives formula (22): ; (22) The relationship of the above four terms in formula (9) is:[[]] ; (23) Combining formula (23) gives formula (24): ; (24) For ease of understanding, formula (16) is the expanded form of formula (15), is the index of the minor term in the sub-term, , that is , corresponding to the first term in formula (16), that is . The -th sub-term and the -th term contain the same number of sub-terms. According to formula (16) and formula (22), it can be seen that the result corresponding to the intersection term contained in formula (22) is a subset of the result corresponding to the intersection term contained in formula (16). That is to say, the occurrence probability of the intersection term in formula (22) is less than or equal to the occurrence probability of the intersection term in formula (16), and at the same time, the test ratio of the intersection term in formula (22) is less than or equal to the test ratio of the intersection term in formula (16). Therefore, the overall result of the first part of formula (24) is greater than or equal to 0. Similarly, the result of the second part of formula (24) is less than or equal to 0, so the overall calculation result is close to 0. Among them, the first part of formula (24) is:[[]] , and the second part of formula (24) is:[[]] 。

[0043] Based on the above analysis, it is known that during the actual calculation process, there is a phenomenon of mutual cancellation of the minor terms included in some subterms in formula (9). Therefore, combining the above analysis and expert experience, the formula for the inherent unreliability of the system is reasonably truncated. While retaining the relevant information between the measurement items, the complexity of the model is reduced. The truncated result is as follows: ; (25) ; (26) ; (27) Based on the above analysis, formulas (12)-(14) of the continuous optimization model considering intersection terms are improved to a continuous optimization model considering real-time performance, specifically as follows: ; (28) ; (29) ; (30) Among them, is the inherent reliability of the system obtained based on the formula for the inherent reliability of the truncated system.

[0044] Through the analysis of the formula for the inherent reliability of the system in the continuous optimization model considering real-time performance, it is known that the influence of multiple events occurring simultaneously on the inherent reliability of the system is very small. Therefore, truncating the formula for the inherent reliability of the system not only ensures the calculation accuracy of the system reliability but also can significantly reduce the model solution time. When facing demand changes, the test strategy can be quickly adjusted.

[0045] In one of the embodiments, in step three, by additionally establishing the relationship between the decision variable and the inherent reliability of the system, a continuous optimization model based on additional constraints is obtained, specifically as follows: The inherent reliability of the intermediate event (measurement item) essentially reflects the qualified rate of the product corresponding to the measurement item. The product corresponding to the measurement item with a high inherent reliability of the system has a relatively high qualified rate and a relatively low possibility of having a faulty main board. On the contrary, the product corresponding to the measurement item with a low inherent reliability of the system has a relatively low qualified rate and a relatively high possibility of having a faulty main board. Therefore, during the testing process, in order to detect as many faulty main boards as possible, the design of the test strategy should follow the principle of setting a relatively low test ratio for the measurement items with a high inherent reliability of the system and a relatively high test ratio for the measurement items with a low inherent reliability of the system, so as to design the best test strategy that ensures both efficiency and quality.

[0046] The value range of the decision variable of the continuous optimization model is , for the test items with relatively low inherent reliability of the system and very high average test time, a very low test ratio is often set. When the yield fluctuates, it is easy to cause the test ratio obtained by optimization to have poor generalization ability on the test set and the effect of reducing the test cost is weakened.

[0047] Since the test ratio is closely related to the inherent reliability of the system of the test item, the test item with high inherent reliability of the system should be tested at a lower ratio, while the test item with low inherent reliability of the system should be tested at a higher ratio. Therefore, adjusting the test ratio based on the process mechanism of the optimization interval, the specific calculation process is as follows: Based on the maintenance data of historical test data, obtain the occurrence probability of the bottom event as the unreliability of the bottom event. To consider the calculation error, obtain the mean and deviation of the occurrence probability of the bottom event through mathematical statistics or combined with expert experience, and determine the reliability range of the bottom event; Based on the reliability range of the bottom event, use the fault tree to quantitatively calculate the reliability range of the corresponding intermediate event of each test item, and then obtain the inherent reliability of the system of each test item, and accordingly set the optimization interval of the test ratio of each test item as a function based on the inherent reliability of the system: ; (31) where is the test ratio of the th test item, and are respectively the lower limit and upper limit of the inherent reliability of the system of the th test item, and are respectively the test ratio coefficients of the lower limit and upper limit of the test ratio and the inherent unreliability of the system, which are set according to expert experience.

[0048] Based on the above analysis, improve the formulas (28)-(30) of the continuous optimization model considering real-time performance to the continuous optimization model with additional constraints, specifically: ; (32) ; (33) ; (34) where, is the test ratio, is the inherent reliability of the system obtained based on the formula for calculating the inherent reliability of the truncated system, represents an all-1 array with a length of .

[0049] It should be noted that since the total number of set measurement items is , and represent the lower and upper limits of the inherent system reliability of the measurement items and are arrays of length m. Here, it is necessary to calculate the inherent system unreliability of the measurement items, so a full-1 array of length is required.

[0050] The continuous optimization model (32)-(34) based on additional constraints ensures that each measurement item can be allocated a reasonable proportion of test resources by introducing additional constraint conditions for the decision variables. This improvement can effectively improve the fault detection rate and reduce the test cost even for measurement items with low inherent system reliability and long average test time.

[0051] In one embodiment, to verify the effectiveness of the proposed method, three evaluation indicators are used in this embodiment: the average mainboard test time, the false negative rate (FNR), and the total average mainboard test cost. The average mainboard test time and the defect level can evaluate the test efficiency and test quality of the test strategy respectively. The calculation process of the indicators is as follows: First, calculate the number of actually qualified mainboards as: ; (35) where represents the mainboard index, is the number of batch mainboards for testing, represents the th measurement item test result of the th mainboard of the th product, is true for the logical AND operation.

[0052] Then the number of actually defective mainboards is: ; (36) When using the test strategy , the number of defective mainboards detected is: ; (37) where represents that the th measurement item of the

[0053] th product fails the test. Then the number of undetected defective mainboards under the test strategy .(38) The false negative rate represents the proportion of undetected defective motherboards among all defective motherboards, and evaluates the detection ability of the test strategy for defective motherboards. Under the test strategy , the formula for calculating the false negative rate FNR is as follows: ;(39) where is the false negative rate under the test strategy .

[0054] The average test time represents the average test time for testing one motherboard and is used to evaluate the test efficiency of the test strategy. Under the test strategy , the formula for the average test time is as follows: ;(40) where represents the test time of the th test item of the th product.

[0055] Average test cost of motherboard is the sum of the test time and repair time of one motherboard and is used to evaluate the overall test effect of the test strategy. The calculation formula is as follows: ;(41) where represents the average repair time of the motherboard, represents the average repair time of the notebook finished product, is the number of faulty motherboards detected by the test strategy in the motherboard functional test stage. Among them, the repair times of the motherboard and the finished product are determined by comprehensive statistical analysis of the actual test process and combined with the experience evaluation of domain experts.

[0056] There is randomness in the high-precision test strategy during the actual test process, specifically manifested in the process of determining the test status according to the test strategy. The test status of the test strategy and each test item of each motherboard can be expressed as: ;(42) where represents the test status of the th test item of the th motherboard, ⌊ ⌉ represents taking the integer value closest to , and

[0057] For the same test ratio , the test status may be different, resulting in different values of the evaluation indicators calculated. To determine the overall test effect of a test ratio, for each test ratio , randomly generate times of test statuses, and calculate the evaluation indicator for each time, so as to use the average value of the times of evaluation indicators to represent the overall test effect of the test ratio.

[0058] The average test cost is used to evaluate the comprehensive test effect of the test strategy. The average test cost under continuous planning is expressed as: ; (43) where represents the th group of test strategies randomly generated by using formula (42) for the test ratio , represents the comprehensive evaluation indicator value of the test status on the training set or the test set, which is calculated by using formula (41).

[0059] The average test time is used to evaluate the test efficiency. The average test time under continuous planning is expressed as: ; (44) where is the average test time of the test status on the training set or the test set, which is calculated by using formula (40).

[0060] FNR is used to evaluate the test quality of the test strategy. The calculation formula under continuous planning is:

[0061] where is the false negative rate of the test status on the training set or the test set, that is, the fault undetected rate, which is calculated by using formula (39).

[0062] In this embodiment, the discrete optimization model is extended to a continuous optimization model to select a reasonable probability value for the simultaneous faults of relevant modules, and by integrating expert experience in the reliability analysis process, the calculation formula of the inherent reliability of the system is reasonably truncated, significantly improving the real-time solution of the optimization model while ensuring the calculation accuracy of the inherent reliability of the system; by additionally establishing the relationship between the decision variable and the system reliability, necessary constraints are added to the optimization model to further enhance the generalization ability of the model.

[0063] To verify the effectiveness of this embodiment, the following comparison methods are set: (1) The continuous optimization method without truncation processing, which is different from this embodiment in that no truncation processing is performed on the formula for calculating the inherent reliability of the system.

[0064] (2) The continuous optimization method without additional constraints, which is different from this embodiment in that the constraint on the optimization range of the test ratio is cancelled, and the optimization range of the test ratio for each test item is set to .

[0065] (3) The continuous optimization method without fusing correlation information, which is different from this embodiment in that when analyzing the reliability of the functional test process, the test items are regarded as independent components in the system, and the connections between the test items are ignored.

[0066] (4) The binary optimization method considering the instability of reliability, which is different from this embodiment in that there are only two settings for the test status of the test items, namely 0 and 1.

[0067] (5) The binary optimization method based on budget maximum coverage, which is different from this embodiment in that the optimization model objective function is to maximize the fault coverage rate in the training set, and at the same time, there are only two settings for the test status of the test items, namely 0 and 1.

[0068] Among them, maximizing the fault coverage rate in the training set as the optimization model objective function is an existing method. For details, please refer to the literature PAN R, ZHANG Z, LI X, et al. Black-Box Test-Coverage Analysis and Test-Cost Reduction Based on a Bayesian Network Model[C] / / 2019 IEEE 37th VLSI Test Symposium(VTS). 2019: 1-6.

[0069] The evaluation indexes of the method of this embodiment and its comparison methods on the test set are shown in Table 1, and the model solution times of each method are shown in Table 2; Table 1 Evaluation indexes of the method of the embodiment and its comparison methods on the test set

[0070] Table 2 Model solution times of the method of this embodiment and its comparison methods

[0071] Comparing the proposed method in Table 1 with the continuous optimization method without truncation processing, it can be seen that truncating the formula for calculating the inherent reliability of the system results in a slightly higher average total test cost of the obtained test strategy by about 1% due to the loss of some relevant information to a certain extent. However, as can be seen from Table 2, the model solving time of the proposed method only accounts for about 4% of that of the continuous optimization method without truncation processing. This processing significantly reduces the model solving time, enabling faster adjustment of the test strategy and reducing unnecessary test costs when facing the need for rapid strategy design.

[0072] From the data in Tables 1 and 2, it can be seen that the method of this embodiment can effectively reduce the undetected rate of defective motherboards, thereby reducing the total motherboard test cost by up to 7.58% at most. By adding test states, the flexibility of test strategy adjustment is enhanced, thereby increasing the detection rate of faulty motherboards and further reducing the test cost by 2.53%. Moreover, based on expert guidance, the relationship between decision variables and system reliability is established additionally to impose necessary constraints on the optimization model, making the optimization result of the model more reasonable and further enhancing the generalization ability of the model.

[0073] As an embodiment; Taking a typical notebook motherboard as an object, by changing the binary state of test items to a continuous test state, the discrete optimization model is extended to a continuous optimization model, achieving high-precision calculation of system reliability. On this basis, a human-machine collaboration method is used to reasonably truncate the formula for calculating the inherent reliability of the system, significantly improving the calculation real-time performance while ensuring the calculation accuracy. Further, a mapping relationship between decision variables and system reliability is constructed based on the experience of domain experts, and the optimization model architecture is improved by introducing key constraint conditions, which not only enhances the generalization ability of the model in different production scenarios but also designs a test strategy that meets the requirements of flexible manufacturing. The specific implementation steps are as follows: A1. According to historical maintenance data, statistically calculate the occurrence probability of bottom events and calculate the reliability range of bottom events; A2. According to the established fault tree, quantitatively calculate the unreliability range of intermediate events corresponding to test items; A3. Combining expert experience, set the lower limit test ratio coefficient in formula (31) to 100 and the upper limit test ratio coefficient to 2500, and then obtain the optimization range of the test ratio , that is, the optimization range of decision variables, and the results are shown in Table 3; Table 3 Optimization interval of test ratio of

[0074] A4. Based on the test ratio Optimization range, and establish the following continuous optimization model: ; (46) ; (47) ; (48) Wherein, is the lower limit of the test ratio and is the upper limit of the test ratio .

[0075] A5. Based on the progress analysis of the motherboard manufacturing process and expert experience evaluation, set the test time threshold for each motherboard to 75 seconds; A6. Combining the statistical analysis of historical repair data and comprehensive evaluation of expert experience, set the motherboard repair cost to 0.5h, and the finished product repair cost to 4h; A7. Solve the continuous optimization model established in A4 to obtain the optimal test strategy, and calculate the evaluation indicators corresponding to the test strategies solved by the method proposed in the present invention and the comparative method respectively. The detailed results are shown in Table 1. By analyzing the evaluation indicators in Table 1, verify the effectiveness of the method proposed in this embodiment.

[0076] In one embodiment, a computer device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above-mentioned method is implemented. Among them, the processor is responsible for performing calculation and control functions; the memory includes a non-volatile storage medium and a memory. The non-volatile storage medium stores an operating system, a computer program, and a database, and the memory provides a support environment for the operation of the operating system and the program; the network interface realizes network communication between the device and an external terminal. The database is specifically used to store the data required for the relevant method. When the processor executes the computer program, the above-mentioned method can be implemented.

[0077] In one embodiment, the present invention also provides a computer-readable storage medium. A number of classification programs are stored on the computer-readable storage medium, and the number of classification programs is used to be called by a processor and execute the above-mentioned method. Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps including the above method embodiments; and the foregoing storage medium includes: various media such as ROM, RAM, magnetic disk, or optical disk that can store program codes.

[0078] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0079] As mentioned above, the above is only the preferred specific embodiment of the present invention, which is used to help understand the technical solution of the present invention. However, the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A cross-domain optimization method for high-real-time function testing of human-machine collaboration, characterized in that, Including: Extend the discrete optimization model to a continuous optimization model, establish a fault tree, and set the probability of simultaneous failures of multiple intermediate events in the fault tree as the minimum of the test ratios corresponding to all intermediate events included in the intersection term, obtaining a continuous optimization model considering the intersection term; Through theoretical analysis of the formula for calculating the inherent reliability of the system and truncating the formula for calculating the inherent reliability of the system in combination with expert experience, obtain a continuous optimization model considering real-time performance; By additionally establishing the relationship between the test ratio of each test item and the inherent reliability of the system, obtain a continuous optimization model based on additional constraints; Solve the continuous optimization model based on additional constraints, and screen the test strategy based on the training set to obtain the optimal test strategy.

2. The functional test cross-domain optimization method according to claim 1, wherein In the fault tree, consider the motherboard failure as the top event, the failure of the test item as the intermediate event, and the underlying cause of the test item failure as the bottom event of the fault tree.

3. The functional test cross-domain optimization method according to claim 2, characterized in that In the process of extending the discrete optimization model to a continuous optimization model, specifically: Through reliability analysis of the motherboard function test process, determine the relationship between the fault coverage rate and the inherent reliability of the system; Based on fault tree analysis, consider the impact of the concurrent failure of intermediate events containing the same bottom event on the inherent reliability of the system; Deduce the system reliability of the motherboard function test process, thereby constructing a continuous optimization model.

4. The functional test cross-domain optimization method according to claim 1, wherein Through theoretical analysis of the formula for calculating the inherent reliability of the system and truncating the formula for calculating the inherent reliability of the system in combination with expert experience, obtain a continuous optimization model considering real-time performance, specifically: The calculation formula for the inherent unreliability of the system is divided into sub-items, and the number of sub-items is the same as the number of measurement items. The th sub-item contains minor items. The th sub-item and the th sub-item contain the same number of minor items, and the two sub-items are divided into a group; By analyzing the expressions of a group of sub-items after division, it is found that there is a phenomenon of mutual cancellation of the small items contained in some sub-items in the formula for calculating the inherent unreliability of the system; Since the sum of the inherent reliability of the system and the inherent unreliability of the system is always equal to 1, truncate the formula for calculating the inherent reliability of the system in combination with expert experience to obtain a continuous optimization model considering real-time performance, thereby reducing the complexity of the continuous optimization model while retaining the relevant information between test items.

5. The functional test cross-domain optimization method according to claim 2, wherein By additionally establishing the relationship between the test ratio of each test item and the inherent reliability of the system, obtain a continuous optimization model based on additional constraints, specifically: Based on the maintenance data of historical test data, obtain the occurrence probability of the bottom event as the unreliability of the bottom event, combine expert experience to obtain the mean and deviation of the occurrence probability of the bottom event, and then determine the reliability range of the bottom event; Based on the reliability range of basic events, the inherent reliability range of the intermediate events corresponding to each test item is quantitatively calculated using the fault tree, and based on this, the test ratio of each test item The optimization interval of is set as a function based on the inherent reliability of the system.

6. The functional test cross-domain optimization method according to claim 5, wherein Test ratio The optimization range is specifically as follows: ； Among them, and are respectively the lower limit and the upper limit of the inherent reliability of the system for the th measurement item, and are the proportionality coefficients of the test ratio to the lower limit and the upper limit of the inherent unreliability of the system, which are set according to expert experience.

7. The functional test cross-domain optimization method according to claim 1, wherein The continuous optimization model based on additional constraints is specifically: ； ； ； Among them, is the test ratio, is the inherent system reliability obtained from the calculation formula of the inherent system reliability after truncation, represents the average test time of the th test item, is the test time threshold for each main board, is the total number of test items, is the test ratio of the th test item, and are the lower and upper limits of the inherent system reliability respectively, and are the proportionality coefficients of the lower and upper limits of the test ratio and the inherent system unreliability respectively, set according to expert experience, represents an all - 1 array with a length of .

8. The functional test cross-domain optimization method according to claim 1, wherein Set three evaluation indicators to verify the effectiveness of the cross-domain overall optimization method for motherboard function testing. The evaluation indicators include the average motherboard test time, the false negative rate, and the total average motherboard test cost; False negative rate The calculation formula is as follows: ； Among them, represents the correspondence relationship between the test strategy randomly generated at the test ratio and the test status of each test item of each main board for the first group, is the false negative rate of the corresponding first group of correspondence relationships on the training set or test set; Average test time of the main board The calculation formula is as follows: ； Among them, represents the test strategy randomly generated at the test ratio and the corresponding relationship between the th group of the test status of each test item on each main board, is the average test time of the corresponding th group of corresponding relationships on the training set or the test set.

9. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, There are several classification programs stored on the computer-readable storage medium. The several classification programs are used to be called by the processor and execute the method described in any one of claims 1-8.

Images