Reliability evaluation and optimization method of parallel test fire test system

Abstract

The present invention discloses a reliability evaluation and optimization method for a parallel test ignition test system, which mainly relates to the field of aerospace pyrotechnics ignition test. The method comprises the following steps: S1, regarding the ignition test system of each test point as a multi-stage task system, performing parallel test redundancy analysis on the ignition test system, and performing module backup; S2, building a reliability model of the parallel test system using a dynamic Bayesian network based on the backup parameters of each module in each stage obtained by the analysis in step S1; S3, constructing a parallel test optimization problem for the ignition test system, establishing a mathematical model for the highest efficiency and the best reliability for the parallel test task, and then obtaining the optimal solution through a parallel test reliability optimization method based on an ant colony algorithm. The present invention can solve the problem of low efficiency of a single test point when the traditional method is used for large-scale test tasks, and complete large-scale test tasks more quickly and accurately.

Description

Technical Field

[0001] The present invention relates to the field of aerospace pyrotechnics ignition testing, and in particular to a reliability evaluation and optimization method for a parallel ignition testing system. Background Art

[0002] A fire test system is specifically designed for testing the quality of explosive products such as pyrotechnics. It determines product performance indicators by conducting fire tests on the products. Due to the unique application scenarios of military and aerospace pyrotechnics, product reliability requirements are extremely high. Therefore, rigorous testing is required to ensure product quality and safety, striving for 100% reliability.

[0003] For smaller test tasks, a single test point can meet the requirements. Therefore, traditional methods often focus on testing at a single test point, failing to consider the scale and efficiency of the test task. However, for larger and more urgent test tasks, continuing to use a single test point reduces test efficiency and significantly increases test time. Therefore, parallel testing is generally used to address the needs of large-scale, urgent test tasks. The reliability of the test results from parallel testing is determined by the reliability of the parallel test system, making the reliability of the parallel test system extremely important.

[0004] Although the parallel test system can perform fire tests on multiple products at the same time, greatly improving the test efficiency, the reliability of the parallel test system is also a very important issue. If the parallel test system fails, it may affect the entire test process and may lead to incorrect test results. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems existing in the prior art and provide a reliability evaluation and optimization method for a parallel test ignition test system, which can solve the problem of low efficiency of a single test point when the traditional method is used for large-scale test tasks, and complete large-scale test tasks faster and more accurately.

[0006] To achieve the above-mentioned purpose, the present invention is implemented through the following technical solutions:

[0007] A parallel test fire test system reliability evaluation and optimization method includes the following steps:

[0008] S1. Treat the ignition test system of each test point as a multi-stage task system, conduct parallel testing and redundancy analysis on the ignition test system, and perform module backup;

[0009] S2. Based on the backup parameters of each module in each stage obtained from the analysis in step S1, a parallel test system reliability model is constructed using a dynamic Bayesian network;

[0010] S3. Construct the parallel test optimization problem of the fire test system, establish a mathematical model for the parallel test task to achieve the highest efficiency and the best reliability, and then obtain the optimal solution through the parallel test reliability optimization method based on the ant colony algorithm.

[0011] Preferably, in step S3, the task scheduling optimization objective function of the parallel test is as follows:

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022] ξ+ψ=1

[0023] Where: J = maxF is the optimization objective function of the parallel test task in the optimization time domain; is the optimization objective function of test point j in the optimization time domain; bl is the balance index function, f j is the target optimization function without considering the balance; i,z∈N, i,z are the batch number and the number of test points of the test task respectively.

[0024] Preferably, in step S3, the total test task time T is determined by the maximum time required for the measurement device in each batch task and the maximum test time required for each test point, and its mathematical model is as follows:

[0025] T=max(T mes )*max(T site )+T switch ;

[0026] Among them, T mes The maximum time required to measure devices in each batch task, T site The maximum test time required for the test point, T switchThe time required for switching the ignition test system components corresponding to this batch of tasks.

[0027] Preferably, in step S3, the parallel test task scheduling algorithm improved based on the ant colony algorithm is as follows:

[0028] 1) Heuristic function η ij , which represents the test batch task t j Test point m i Select the desired level:

[0029] r j ∈m i Test task set;

[0030] Where Q represents the strength of the expected degree of test task selection by the test point; f ji Represents the tested batch task r j At the test point m i The objective optimization function of

[0031] 2) State transition probability This function represents the probability that ant k will choose the next task j after completing the current test task i:

[0032]

[0033] Among them: α is the information heuristic factor, which represents τ ij The greater its value, the more likely the ants are to choose the routes taken by other ants, and the higher the probability that all feasible solutions obtained by ants will converge together; τ ij The pheromone concentration of the test batch task is selected for the test point; β = 1-α is η ij The greater the value, the greater the influence of the state transition probability on the prior objective factors; k (i) is ant k at test point m i The taboo table indicates the test batch tasks that the test point has completed, and s is the batch task number that does not belong to the taboo table;

[0034] 3) Pheromone update rule τ ij : After completing a full iteration, consider the path selection of all ants for the test task selection of any test point and calculate:

[0035] τ ij =(1-ρ)τ ij +τ′ ij

[0036]

[0037] Where: (1-ρ) is the pheromone residual factor; n is the number of ants; is the pheromone increment of ant k from test batch task i to test batch task j, which is calculated as follows:

[0038]

[0039] Where: τ0 is the initial value of pheromone concentration; f s is the value of the objective function generated by the path passed by ant k, is its average value; L is the minimum objective function value for ants to select test batch tasks; k∈i→j means that ant k selects test task j based on the current test task i.

[0040] Preferably, in step S3, a solution with similar test time for multiple test points is selected from the existing optimal solutions for multiple test points, and is expressed using the following formula:

[0041] F=min(F s )

[0042]

[0043] Among them, F s is the objective function corresponding to the test point parameter s, z is the number of test points, T v is the test time of test point v, is the average time of z test points.

[0044] Preferably, in step S2, the parallel test system is formed by connecting ignition test systems of multiple test points in series, and the reliability of the entire test system is formed by combining the reliability of each test point.

[0045] Preferably, in step S1, redundant backup is performed only on the dynamic modules in the ignition test system.

[0046] Compared with the prior art, the beneficial effects of the present invention are:

[0047] This paper uses a dynamic Bayesian network to build a reliability model for a multi-stage task system. Then, using a parallel test reliability optimization method based on an ant colony algorithm, a heuristic function, state transition probabilities, and global pheromone update rules are designed based on the structural characteristics of the ignition test system. This allows the system to achieve optimal results in terms of test time and reliability. Compared to conventional methods, this paper addresses the low efficiency of single test points when used for large-scale testing tasks, enabling faster and more accurate completion of large-scale testing tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a flow chart of the present invention;

[0049] Figure 2 A block diagram of the reliability of the parallel test system and a schematic diagram of the Bayesian network of the present invention;

[0050] Figure 3 This is a flowchart of the parallel test task scheduling optimization algorithm of the present invention. DETAILED DESCRIPTION

[0051] Below in conjunction with specific embodiment, further set forth the present invention.Should be understood that these embodiments are only used to illustrate the present invention and are not used in limiting the scope of the present invention.In addition, should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the application equally.

[0052] Example: As shown in the attached Figure 1-3 As shown, the present invention describes a reliability evaluation and optimization method for a parallel test ignition test system. A reliability model of a parallel test ignition test system is constructed. Then, through a parallel test reliability optimization method based on an ant colony algorithm, a heuristic function, a state transition probability, and a global pheromone update rule are designed according to the structural characteristics of the ignition test system, so that the system can obtain the optimal solution for test time and reliability.

[0053] Including steps:

[0054] S1. Redundancy analysis of parallel testing of ignition test system:

[0055] Because parallel testing involves multiple test points sharing the entire test task, component wear in each ignition test system is lower than that in a single test point. Therefore, in parallel testing, the static modules (measurement devices that remain operational throughout the entire test process) and special modules (measurement devices whose operational state involves human intervention and whose matrix switch channel configuration is affected by dynamic modules) of the ignition test system are considered to have very low component wear and tear. Therefore, redundant backup is not implemented for these modules. Only the dynamic modules (measurement devices whose device configuration changes at each stage of the task based on the type and number of test targets) in the ignition test system are redundantly backed up.

[0056] S2. Dynamic Bayesian network builds parallel test system reliability model.

[0057] The parallel test system is composed of multiple test point ignition test systems in series. The reliability of the test system as a whole is composed of the reliability of each test point. Its reliability block diagram and Bayesian network are as follows: Figure 2 shown.

[0058] S3. Construct and solve the parallel test optimization problem of the ignition test system, which specifically includes the following steps:

[0059] S31, constructing a fire test system parallel test optimization problem;

[0060] In response to the production-driven problem of the fire test task, the parallel test task is optimized in a rolling time domain. That is, the test can be carried out while the task is being produced, and the parallel test optimization is performed within a certain time interval, and the time interval is in a rolling change. Assume that the length of the rolling time domain optimization time interval is h, the initial test task batch number is t, and the current test task batch number is i.

[0061] The optimization problem of parallel testing includes optimizing the reliability of the firing test system, minimizing the switching frequency of the test system, minimizing the test time, and considering the priority of the test tasks. The mathematical model of its objective function is shown in the following formula:

[0062]

[0063] where f j is the target optimization function for test point j, i,j∈N, j≤z; To test the reliability of the system by firing batch i+1 at test point j in the optimized time domain; is the number of measurement device switches between test batches i and i+1 in the ignition test system in the optimized time domain; is the time required to test the i+1 batch task of test point j in the optimized time domain; p t,h+t is the task priority of batch i+1 relative to batch i in the optimization time domain; ξ is the reliability weight coefficient, ψ is the test time weight coefficient, and ξ+ψ=1.

[0064] When performing parallel testing, each test batch task can only select one test point for testing. Therefore, the constraint mathematical model is as follows:

[0065]

[0066] Among them, b j is the flag of the test batch task at test point j, indicating whether the test task is tested at this test point, and z is the number of test points.

[0067] During the fire test, the test point test time consists of two parts, mainly including the device use time and the device switching time. Its mathematical model is as follows:

[0068]

[0069] in, The time required to measure the device in the i+1 batch task in the optimized time domain, To test the switching device time of batch i+1 in the optimized time domain, They are the switch flags of the test batch task at the test point, indicating whether the test batch task has been switched.

[0070] To address the measurement device switching issue in a fire test system and reduce the switching rate of measurement devices, a test system device switching optimization problem was constructed. In a fire test system, switching measurement devices results in switching measurement channels in a matrix switch. Therefore, this problem can be transformed into a matrix switch measurement channel switching problem.

[0071] The matrix switch in the fire test system of this embodiment adopts an m×n structure, and its mathematical model can be a two-dimensional matrix. Indicates the configuration of the matrix switch measurement channel of the test system at batch i, The configuration of the matrix switch measurement channels of the test system for batch i+1 is shown in the following two-dimensional matrix mathematical model:

[0072]

[0073]

[0074] Among them, w t,h+t ∈{0,1}, w=0 indicates that the relay in the matrix switch is in the disconnected state, w t,h+t =1 indicates that the relay in the matrix switch is in the closed state. To calculate the switching of the measuring device of the ignition test system between two test batch tasks, the present invention uses a two-dimensional matrix The number of -1s in the value is used to measure the switching of the measurement device between the two test tasks. The smaller the number of -1s, the lower the switching degree. It is expressed by the following formula:

[0075]

[0076] To solve the problem of priority of each test task, this embodiment classifies all test tasks into four priority levels (p0, p1, p2, p3). The smaller the subscript, the greater the priority. Among them, p0 is set for emergency tasks and has the highest priority.

[0077] To prevent low-priority batch test tasks from not being tested, the present invention sets a task waiting time threshold T for each test task. max , if the waiting time of the test task is greater than or equal to T max Then arrange the test first.

[0078] If T i ≥T max Then p i =p0

[0079] Among them, T i is the waiting test time of batch task i, p i is the test priority of batch task i.

[0080] In order to achieve the conditions of optimal reliability of the parallel test system and minimize test time while enabling balanced testing of each test point, the present invention proposes a balance index between test points, the mathematical model of which is shown in the following formula:

[0081]

[0082] Among them, bl is the balance indicator function.

[0083] The task scheduling optimization objective function for parallel testing is as follows:

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094] ξ+ψ=1

[0095] Where: J = maxF is the optimization objective function of the parallel test task in the optimization time domain; is the optimization objective function of test point j in the optimization time domain; bl is the balance index function, f j is the target optimization function without considering the balance; i,z∈N, i,z are the batch number and the number of test points of the test task respectively.

[0096] The total test task time T is determined by the maximum time required to measure the device in each batch task and the maximum test time required for each test point. Its mathematical model is as follows:

[0097] T=max(Tmes )*max(T site )+T switch

[0098] Among them, T mes The maximum time required to measure devices in each batch task, T site The maximum test time required for the test point, T switch The time required for switching the ignition test system components corresponding to this batch of tasks.

[0099] S32, solving the parallel test optimization problem of the ignition test system;

[0100] To improve the reliability of a parallel fire test system and the efficiency of the entire test process under constraints, this embodiment proposes a parallel test reliability optimization method based on an ant colony algorithm. Based on the structural characteristics of the fire test system, a heuristic function, state transition probabilities, and global pheromone update rules are designed. However, the solution obtained by this algorithm may not be unique. Therefore, this embodiment considers the evaluation criterion of test point balance, and can use this criterion to obtain the optimal solution.

[0101] The parallel test task scheduling optimization algorithm for the fire test system is improved based on the ant colony algorithm. The present invention combines the principle of the ant colony algorithm with the characteristics of the fire test system test process to obtain the matrix F with the shortest test time and the best reliability of the fire test system. The rows in the matrix represent the test points, the columns represent the test order of the test points for the test tasks, and the elements within the matrix represent the test batch task numbers:

[0102]

[0103] in: m in i (i=1,2,…,x) is the test point number, if It means that the test point is idle at this time, and the test batch task can go to this test point for testing; j (j = 1, 2, 3, ..., y) is the test batch task sequence number.

[0104] Considering the characteristics of the fire test system test process, the parallel test task scheduling algorithm improved by the ant colony algorithm is designed as follows:

[0105] 1) Heuristic function η ij This function represents the test batch task t j Test point m i Select the desired level:

[0106] r j ∈m i Test task set;

[0107] Where Q represents the strength of the expected degree of test task selection by the test point; f ji Represents the tested batch task r j At the test point m i The objective optimization function.

[0108] The value of the heuristic function indicates the influence of a priori objective factors. Based on the principle of greed, the larger the value of the objective function at a test point, the smaller the value of the heuristic function and the lower the transition probability, and vice versa. However, this process can lead to premature convergence. Therefore, to avoid falling into local optimal solutions, a pheromone parameter is incorporated into task selection.

[0109] 2) State transition probability This function represents the probability that ant k will choose the next task j after completing the current test task i:

[0110]

[0111] Among them: α is the information heuristic factor, which represents τ ij The greater its value, the more likely the ants are to choose the routes taken by other ants, and the higher the probability that all feasible solutions obtained by ants will converge together; τ ij The pheromone concentration of the test batch task is selected for the test point; β = 1-α is η ij The greater the value, the greater the influence of the state transition probability on the prior objective factors; k (i) is ant k at test point m i The taboo table indicates the test batch task that the test point has completed, and s is the batch task number that does not belong to the taboo table. State transition probability It is mainly determined by α and β. Therefore, in order to make the scheduling optimization algorithm of parallel testing have good performance, it is very important to set α and β reasonably.

[0112] 3) Pheromone update rule τ ij : After completing a full iteration, consider the path selection of all ants for the test task selection of any test point and calculate:

[0113] τ ij =(1-ρ)τ ij +τ′ ij

[0114]

[0115] Where: (1-ρ) is the pheromone residual factor; n is the number of ants; is the pheromone increment of ant k from test batch task i to test batch task j, which is calculated as follows:

[0116]

[0117] Where: τ0 is the initial value of pheromone concentration; f s is the value of the objective function generated by the path passed by ant k, is its average value; L is the minimum objective function value for ants to select test batch tasks; k∈i→j means that ant k selects test task j based on the current test task i.

[0118] The above steps can be used to obtain the task scheduling optimization matrix for parallel testing. However, there may be multiple solutions. This is because the test order of the test points is different, and the number of stage tasks undertaken by each test point is different. This results in different working intensities at each test point when the overall objective function value is the same. Therefore, in order to minimize the working intensity of the ignition test system between each test point while keeping the overall objective function value the same, it is also necessary to select the optimal solution based on the specific working conditions of the test point. In the case where the optimization result of the optimization algorithm contains multiple optimal solutions at the same time, the present invention aims to balance the test time of each test point and select a single best optimal solution from the multiple optimal solutions.

[0119] Considering parallel testing at multiple test points, the ideal optimization task sequence is to select a test task solution that is balanced across multiple test points, i.e., the test task optimization solution with the highest degree of balance. Therefore, from the existing optimal solutions for multiple test points, a solution with similar test times for multiple test points is selected, which can be expressed using the following formula:

[0120] F=min(F s )

[0121]

[0122] Among them, F s is the objective function corresponding to the test point parameter s, z is the number of test points, T v is the test time of test point v, is the average time of z test points.

Claims

1. Parallel test fire test system reliability evaluation and optimization method, characterized by: Including steps: S1. Treat the ignition test system of each test point as a multi-stage task system, conduct parallel testing and redundancy analysis on the ignition test system, and perform module backup; S2. Based on the backup parameters of each module in each stage obtained from the analysis in step S1, a parallel test system reliability model is constructed using a dynamic Bayesian network; S3. Construct a parallel test optimization problem for the fire test system, establish a mathematical model for the parallel test tasks to maximize efficiency and reliability, and then obtain the optimal solution through a parallel test reliability optimization method based on the ant colony algorithm; The task scheduling optimization objective function for parallel testing is as follows: ξ+ψ=1 Where: J = maxF is the optimization objective function of parallel test tasks in the optimization domain; i, z, j∈N, where i is the batch number of the test task, z is the number of test points, and j is the serial number of the test point; b j It is the flag of the test batch task at test point j, indicating whether the test task is tested at this test point; is the optimization objective function of test point j in the optimization time domain; bl is the balance index function; f j is the target optimization function without considering the degree of balance; ξ is the reliability weight coefficient; ψ is the test time weight coefficient; To test the reliability of the system by firing batch i+1 at test point j in the optimized time domain; is the number of measurement device switches between test batches i and i+1 in the ignition test system in the optimized time domain; is the time required to test the i+1 batch task of test point j in the optimized time domain; p t,h+t is the task priority of batch i+1 relative to batch i in the optimization time domain; To use a two-dimensional matrix The number of -1s in the value is used to measure the switching of the measurement device between the two test tasks. The smaller the number of -1s, the lower the switching degree. Indicates the configuration of the matrix switch measurement channels of the test system at batch i; Indicates the configuration of the matrix switch measurement channel of the test system in batch i+1; w t,h+t ∈{0,1}, w=0 indicates that the relay in the matrix switch is in the disconnected state, w t,h+t =1 indicates that the relay in the matrix switch is in the closed state; The time required to measure the device in the i+1 batch task in the optimized time domain; The switching device time of test batch i+1 in the optimized time domain; It is the switch flag of the test batch task at the test point, indicating whether the test batch task has been switched.

2. The parallel test fire test system reliability evaluation and optimization method according to claim 1, wherein: In step S3, the total test task time T is determined by the maximum time required for measuring devices in each batch task and the maximum test time required for each test point. Its mathematical model is as follows: T=max(T mes )*max(T site )+T switch ; Among them, T mes The maximum time required to measure devices in each batch task, T site The maximum test time required for the test point, T switch The time required for switching the ignition test system components corresponding to this batch of tasks.

3. The parallel test fire test system reliability evaluation and optimization method according to claim 2, wherein: In step S3, the parallel test task scheduling algorithm improved based on the ant colony algorithm is as follows: 1) Heuristic function η ij , which represents the test batch task t j Test point m i Select the desired level: Where Q represents the strength of the expected degree of test task selection by the test point; f ji Represents the tested batch task r j At the test point m i The objective optimization function of 2) State transition probability This function represents the probability that ant k will choose the next task j after completing the current test task i: Among them: α is the information heuristic factor, which represents τ ij The greater its value, the more likely the ants are to choose the routes taken by other ants, and the higher the probability that all feasible solutions obtained by ants will converge together; τ ij The pheromone concentration of the test batch task is selected for the test point; β = 1-α is η ij The greater the value, the greater the influence of the state transition probability on the prior objective factors; k (i) is ant k at test point m i The taboo table indicates the test batch tasks that the test point has completed, and s is the batch task number that does not belong to the taboo table; 3) Pheromone update rule τ ij : After completing a full iteration, consider the path selection of all ants for the test task selection of any test point and calculate: t ij =(1-ρ)τ ij +t′ ij Where: (1-ρ) is the pheromone residual factor; n is the number of ants; is the pheromone increment of ant k from test batch task i to test batch task j, which is calculated as follows: Where: τ0 is the initial value of pheromone concentration; f s is the value of the objective function generated by the path passed by ant k, and f is its average value; L is the minimum objective function value for the ant to select the test batch task; k∈i→j means that ant k selects test task j based on the current test task i.

4. The parallel test fire test system reliability evaluation and optimization method according to claim 3, wherein: In step S3, a solution with similar test time for multiple test points is selected from the existing optimal solutions for multiple test points, and is expressed using the following formula: F=min(F s ) Among them, F s is the objective function corresponding to the test point parameter s, z is the number of test points, T v is the test time of test point v, is the average time of z test points.

5. The parallel test fire test system reliability evaluation and optimization method according to claim 1, wherein: In step S2, the parallel test system is formed by connecting the ignition test systems of multiple test points in series, and the reliability of the entire test system is formed by combining the reliability of each test point.

6. The parallel test fire test system reliability evaluation and optimization method according to claim 1, wherein: In step S1, only the dynamic modules in the ignition test system are backed up redundantly.

Images