Optimal test strategy generation method for complex communication system

Abstract

The invention discloses an optimal test strategy generation method for a complex communication system, and the method comprises the steps: firstly constructing a fault test model of a to-be-tested system, then generating an optimal complete graph in combination with the fault test model, and reducing the optimal complete graph into a binary tree which serves as an optimal test strategy, thereby solving a problem that a space structure is difficult to solve when the space structure is complex; in the process of generating the optimal complete graph, an intelligent search algorithm is selected, and a complete graph solution set is continuously iterated and updated according to an accelerated convergence rule to fit an optimal test strategy.

Description

Technical Field

[0001] This invention belongs to the field of communication equipment test strategy generation technology, and more specifically, relates to an optimal test strategy generation method for complex communication systems. Background Technology

[0002] Communication systems are the core infrastructure of modern information society, undertaking key functions such as data transmission, signal interaction, and resource scheduling. They are widely used in broadcasting, emergency rescue, and live broadcast support for major events. Throughout the entire lifecycle of a communication system, fault diagnosis is a crucial link in ensuring signal transmission and stable system operation. Finding the optimal diagnostic strategy is a core technology for improving the efficiency of fault location in broadcasting and other equipment, reducing the operation and maintenance costs of communication system facilities, and minimizing accident losses.

[0003] In equipment fault diagnosis, the steps to find the optimal diagnostic strategy are typically as follows: first, establish a multi-signal flow graph model; then, describe the strategy generation problem as an optimal binary search tree (decision tree) generation problem (OBP); and finally, solve the OBP. In the OBP, the algorithm progressively distinguishes fault states based on the test sequence, ultimately forming a binary tree. However, the selection of tests and the classification of fault states are closely related. The architectural pattern of this problem leads to the mutual coupling of decision variables, making the solution space complex and difficult to search. Furthermore, as the system scale increases, the scale and dimension of the solution space increase significantly, making the problem-solving process cumbersome and reducing the efficiency of problem solving.

[0004] Existing algorithms for solving OBP can be categorized into three types: ① Backtracking algorithms (such as Rollout and AO*), which theoretically can find the global optimum, but have extremely high computational complexity; ② Update algorithms (such as SADPSO and QGA), which use metaheuristic algorithms for iteration, but the strong coupling characteristics of the binary tree structure easily violate constraints and lead to local optima; ③ Greedy algorithms (such as Growing), which are computationally fast but suffer from short-sightedness, where the current optimal test may lead to subsequent high-cost tests. Furthermore, finding the optimal test strategy (OTS) is essentially an NP-hard problem, and the theoretically optimal solution is unattainable, further highlighting the limitations of traditional OBP solutions.

[0005] To overcome the above problems, this invention is based on a multi-signal flow graph model. First, it constructs the Optimal Separation Matrix Generation Problem (OSP) as an intermediate bridge, proving that OSP and OBP are equivalent. Then, by relaxing the redundancy constraints of OSP and combining it with a projection algorithm, it constructs the Optimal Complete Graph Generation-Projection Problem (OGPP). OGPP and OBP have the same solution and the decision variables are independent of each other, resulting in a simpler solution space structure and significantly improved solvability.

[0006] The optimal test strategy generation method for complex communication systems solves the coupling and complexity of the traditional optimal binary tree generation problem by utilizing the optimal complete graph generation problem, optimizing the search process, and improving the feasibility and efficiency of the solution. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide an optimal test strategy generation method for complex communication systems. This method reconstructs the binary tree by generating an optimal complete graph, and provides a solution based on decision variable decoupling to address the difficulty in solving complex solution spaces. The method iteratively updates the complete graph set according to the accelerated convergence rule to fit the optimal test strategy.

[0008] To achieve the above-mentioned objectives, the present invention provides a method for generating optimal test strategies for complex communication systems, characterized by comprising the following steps:

[0009] (1) Constructing a fault test model :

[0010] ;

[0011] in, This represents the set of potential faults in the system under test. , Indicates the first Types of fault states, Indicates a fault-free state. Indicate the type of fault; This represents all available test sets for the system under test. , Indicates the first One available test, D represents the total number of available tests; D represents the fault dependency matrix, specifically:

[0012] ;

[0013] in, express exist The following test information, or ,when When, it means able to pass When detected, When, it means Unable to pass It was detected;

[0014] Is with The corresponding prior probability set, , express The prior probability; Is with The corresponding set of test costs, , express The cost of testing;

[0015] (2) Parameter definition;

[0016] Define the fault ordering weight vector :

[0017] For each fault Randomly generate weight values , ,Will Each weight value Composition of fault ranking weight vector , recorded as ;

[0018] According to each fault Corresponding weight value For fault sets The fault sets are sorted in descending order of weight, and the sorted fault sets are denoted as follows: , Indicates the sorted order of the first... One fault;

[0019] Define the family of relationships between faults : , Indicates all that can and The set of separate tests, namely: ;

[0020] Define the generation probability matrix : , where the probability vector satisfy: , No. element express and Depend on The probability of separation;

[0021] (3) Generating a binary tree using the particle swarm optimization algorithm:

[0022] (3.1) Set the population size as The maximum number of iterations is ;

[0023] (3.2) Initialize the particle swarm, assigning each individual in the swarm the number 1. particles Generate fault ranking weight vector Fault Relationships Family probability matrix and separation matrix , where the separation matrix The size is The diagonal elements are all 0, and the upper triangular elements are all 0. The lower triangle element is obtained by transposing the upper triangle element;

[0024] (3.3) Iterative Updates:

[0025] (3.3.1) Initialize the number of iterations ;

[0026] (3.3.1.1) Initialize individual ID ;

[0027] (3.3.1.2) Generate test set family ;

[0028] ;

[0029] in, Indicates will A set of tests that distinguish it from other faults. Indicates the ability to and A separate test;

[0030] In this embodiment, the test The determination is based on the principles of roulette, specifically as follows:

[0031] Generate vectors , Indicates a fault With fault By test The probability of separation;

[0032] Determine elements exist The corresponding interval in the middle, when the element Located in the interval At that time, then corresponding Set as ,Right now: ;

[0033] (3.3.1.3) Update the test set family using the accelerated convergence rule. ;

[0034] (3.3.1.3.1) Initialization ;

[0035] (3.3.1.3.2) Execute redundancy constraint judgment:

[0036] 1) Initialization ;

[0037] 2) Remember The first in The elements are ;

[0038] 3) Comparison and The size, if If so, proceed directly to step 4); , then it means separable and ,use Replace elements in Record the replacement for After replacement, based on symmetry constraints, let In Record the replacement for ;

[0039] Calculate the variables based on the results of the substitution. Then proceed to step 4).

[0040] 4) Determine the current iteration number Does it meet the requirements? If satisfied, proceed to step 5; otherwise, let Then return to step 2);

[0041] 5) Minimum value corresponding to replace elements in After the replacement was finally obtained ;

[0042] (3.3.1.3.3), Order Then return to step (3.3.1.3.2) until... ;

[0043] (3.3.1.3.4) Obtain the complete diagram :

[0044] ;

[0045] (3.3.1.4) Calculation fitness ;

[0046] (3.3.1.5) Order Then return to step (3.3.1.2) until... ;

[0047] (3.3.2) Iteratively update according to the fitness using the particle swarm optimization algorithm;

[0048] (3.3.3) Then return to step (3.3.1.1) until... fitness smallest The best complete diagram is denoted as ;

[0049] (3.4) will Restore to a binary tree :

[0050] (3.4.1) Initialize the set of nodes to be processed Initial node ;

[0051] (3.4.2) Selection Select a node and denote it as , will node As The root node of the current state;

[0052] (3.4.3) Extraction Contains Given the nodes and edges, generate a graph. ;

[0053] (3.4.4) Select according to the following formula Tests in and will As Current node The following test;

[0054] ;

[0055] (3.4.5) According to the test Will The corresponding fault set is divided into two subsets, which are represented by nodes. and ;

[0056] Determine the node Does it contain more than 1 fault? Then let ,otherwise, Remain unchanged;

[0057] Determine the node Does it contain more than 1 fault? Then let ,otherwise, Remain unchanged;

[0058] treat and After all judgments are completed, from Delete;

[0059] (3.4.6) Repeat steps (3.4.2)-(3.4.5) until... , After restoration is complete, proceed to step (4).

[0060] (4) The restored This is the optimal testing strategy.

[0061] The objective of this invention is achieved as follows:

[0062] This invention provides a method for generating optimal test strategies for complex communication systems. First, a fault test model of the system under test is constructed. Then, the optimal complete graph is generated by combining the fault test model and restored to a binary tree that serves as the optimal test strategy, thus solving the problem of difficulty in solving when the solution space structure is complex. In the process of generating the optimal complete graph, an intelligent search algorithm is selected, and the complete graph set is continuously updated iteratively according to the acceleration convergence rule to fit the optimal test strategy.

[0063] The optimal test strategy generation method for complex communication systems proposed in this invention also has the following beneficial effects:

[0064] (1) By using the OSP intermediate bridge and projection algorithm, the complete graph set is ensured to meet the fault isolation constraint. The OGPP is solved by the particle swarm algorithm to improve the reliability of the solution. It should be noted that the intelligent search algorithm used in this invention can be selected from a variety of options, such as genetic algorithm, particle swarm algorithm, ant colony algorithm, etc.

[0065] (2) This invention provides a quantitative standard for generating the optimal test strategy by defining the expected cost of the diagnostic strategy and combining the prior probability of the fault state with the test cost;

[0066] (3) This invention optimizes the update process of the complete graph set by defining an accelerated convergence rule, and improves computational efficiency by accelerating solution convergence through local exchange and update;

[0067] (4) The present invention determines the decision variables by random number roulette based on different weights, making low-cost tests more likely to be selected to optimize the diagnostic path. Attached Figure Description

[0068] Figure 1 This is a flowchart of an optimal test strategy generation method for complex communication systems according to the present invention;

[0069] Figure 2 This is a schematic diagram of the complete set of diagrams;

[0070] Figure 3 It is a complete graphical reconstruction of a binary tree;

[0071] Figure 4 This is a diagram illustrating the comparison of testing costs between the control algorithm and the proposed method on the validation dataset. Detailed Implementation

[0072] The specific embodiments of the present invention will now be described with reference to the accompanying drawings to enable those skilled in the art to better understand the invention. It should be particularly noted that in the following description, detailed descriptions of known functions and designs that might obscure the main content of the invention will be omitted here.

[0073] Example

[0074] In this embodiment, as Figure 1 As shown, the present invention provides a method for generating optimal test strategies for complex communication systems, comprising the following steps:

[0075] (1) Constructing a fault test model :

[0076] ;

[0077] in, This represents the set of potential faults in the system under test. , Indicates the first Types of fault states, Indicates a fault-free state. Indicate the type of fault; This represents all available test sets for the system under test. , Indicates the first One available test, D represents the total number of available tests; D represents the fault dependency matrix, specifically:

[0078] ;

[0079] in, express exist The following test information, or ,when When, it means able to pass When detected, When, it means Unable to pass It was detected;

[0080] Is with The corresponding prior probability set, , express The prior probability; Is with The corresponding set of test costs, , express The cost of testing;

[0081] (2) Parameter definition;

[0082] Define the fault ordering weight vector :

[0083] For each fault Randomly generate weight values , ,Will Each weight value Composition of fault ranking weight vector , recorded as ;

[0084] According to each fault Corresponding weight value For fault sets The fault sets are sorted in descending order of weight, and the sorted fault sets are denoted as follows: , Indicates the sorted order of the first... One fault;

[0085] Define the family of relationships between faults : , Indicates all that can and The set of separate tests, namely: ;

[0086] Define the generation probability matrix : , where the probability vector satisfy: , No. element express and Depend on The probability of separation is calculated using the following formula:

[0087] ;

[0088] (3) Generating a binary tree using the particle swarm optimization algorithm:

[0089] (3.1) Set the population size as The maximum number of iterations is ;

[0090] (3.2) Initialize the particle swarm, assigning each individual in the swarm the number 1. particles Generate fault ranking weight vector Fault Relationships Family probability matrix and separation matrix , where the separation matrix The size is The diagonal elements are all 0, and the upper triangular elements are all 0. The lower triangle element is obtained by transposing the upper triangle element;

[0091] (3.3) Iterative Updates:

[0092] (3.3.1) Initialize the number of iterations ;

[0093] (3.3.1.1) Initialize individual ID ;

[0094] (3.3.1.2) Generate test set family ;

[0095] ;

[0096] in, Indicates will A set of tests that distinguish it from other faults. Indicates the ability to and A separate test;

[0097] (3.3.1.3) Update the test set family using the accelerated convergence rule. ;

[0098] (3.3.1.3.1) Initialization ;

[0099] (3.3.1.3.2) Execute redundancy constraint judgment:

[0100] 1) Initialization ;

[0101] 2) Remember The first in The elements are ;

[0102] 3) Comparison and The size, if If so, proceed directly to step 4); , then it means separable and ,use Replace elements in Record the replacement for After replacement, based on symmetry constraints, let In Record the replacement for ;

[0103] Calculate the variables based on the results of the substitution. Then proceed to step 4).

[0104] 4) Determine the current iteration number Does it meet the requirements? If satisfied, proceed to step 5; otherwise, let Then return to step 2);

[0105] 5) Minimum value corresponding to replace elements in After the replacement was finally obtained ;

[0106] (3.3.1.3.3), Order Then return to step (3.3.1.3.2) until... ;

[0107] (3.3.1.3.4) Obtain the complete diagram :

[0108] ;

[0109] In this embodiment, The test set family in the OGPP corresponds to the separation matrix, and the elements of the separation matrix are... and The intersection elements, such as Figure 2 The image shows an example of a complete diagram. There are three complete diagrams in total, from left to right: The three solution sets share the same nodes, but different tests are used on the paths, resulting in different costs. The cost relationship between the three complete solutions in the diagram is as follows: ;

[0110] (3.3.1.4) Calculation fitness ;

[0111] (3.3.1.5) Order Then return to step (3.3.1.2) until... ;

[0112] (3.3.2) Iteratively update according to the fitness using the particle swarm optimization algorithm;

[0113] (3.3.3) Then return to step (3.3.1.1) until... fitness smallest The best complete diagram is denoted as ;

[0114] (3.4) will Restore to a binary tree :

[0115] (3.4.1) Initialize the set of nodes to be processed Initial node ;

[0116] (3.4.2) Selection Select a node and denote it as , will node As The root node of the current state;

[0117] (3.4.3) Extraction Contains Given the nodes and edges, generate a graph. ;

[0118] (3.4.4) Select according to the following formula Tests in and will As Current node The following test;

[0119] ;

[0120] (3.4.5) According to the test Will The corresponding fault set is divided into two subsets, which are represented by nodes. and ;

[0121] Determine the node Does it contain more than 1 fault? Then let ,otherwise, Remain unchanged;

[0122] Determine the node Does it contain more than 1 fault? Then let ,otherwise, Remain unchanged;

[0123] treat and After all judgments are completed, from Delete;

[0124] (3.4.6) Repeat steps (3.4.2)-(3.4.5) until... , After restoration is complete, proceed to step (4).

[0125] In this embodiment, Figure 3 This is an example illustrating the complete reconstruction of a binary tree diagram. The binary tree is obtained through the restoration step. .

[0126] (4) The restored This is the optimal testing strategy.

[0127] Figure 4 This is a diagram illustrating the comparison of testing costs between the control algorithm and the proposed method on the validation dataset.

[0128] In this embodiment, as Figure 4 As shown, different algorithms for directly solving OBP to generate optimal test strategies and solving OGPP to generate optimal test strategies are presented at different scales. A comparison of testing costs is presented. The datasets are derived from physical experiments of six different communication systems: A: a satellite communication link transmission system (15×22); B: a superheterodyne receiver system (22×36); C: an electronic system (40×14); D: a multi-antenna MIMO system (61×28); E: a radar RF transceiver front-end (73×50); and F: a signal processing system (76×63). In the figure, Rollout to DIG represent direct solution of OBP, and GA to HHO represent solution of OGPP. Under the same scale D, the testing cost of solving OGPP is less than that of solving OBP.

[0129] Although the illustrative specific embodiments of the present invention have been described above to enable those skilled in the art to understand the invention, it should be understood that the invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the invention as defined and determined by the appended claims, and all inventions utilizing the concept of the present invention are protected.

Claims

1. A method for generating optimal test strategies for complex communication systems, characterized in that, Includes the following steps: (1) Constructing a fault test model : ； in, This represents the set of potential faults in the system under test. , Indicates the first Types of fault states, Indicates a fault-free state. Indicate the type of fault; This represents all available test sets for the system under test. , Indicates the first One available test, D represents the total number of available tests; D represents the fault dependency matrix, specifically: ； in, express exist The following test information, or ,when When, it means able to pass When detected, When, it means Unable to pass It was detected; Is with The corresponding prior probability set, , express The prior probability; Is with The corresponding set of test costs, , express The cost of testing; (2) Parameter definition; Define the fault ordering weight vector : For each fault Randomly generate weight values , ,Will Each weight value Composition of fault ranking weight vector , recorded as ; According to each fault Corresponding weight value For fault sets The fault sets are sorted in descending order of weight, and the sorted fault sets are denoted as follows: , Indicates the sorted order of the first... One fault; Define the family of relationships between faults : , Indicates all that can and The set of separate tests, namely: ; Define the generation probability matrix : , where the probability vector satisfy: , No. element express and Depend on The probability of separation; (3) Generating a binary tree using the particle swarm optimization algorithm: (3.1) Set the population size as The maximum number of iterations is ; (3.2) Initialize the particle swarm, assigning each individual in the swarm the number 1. particles Generate fault ranking weight vector Fault Relationships Family probability matrix and separation matrix , where the separation matrix The size is The diagonal elements are all 0, and the upper triangular elements are all 0. The lower triangle element is obtained by transposing the upper triangle element; (3.3) Iterative Updates: (3.3.1) Initialize the number of iterations ; (3.3.1.1) Initialize individual ID ; (3.3.1.2) Generate test set family ; ； in, Indicates will A set of tests that distinguish it from other faults. Indicates the ability to and A separate test; (3.3.1.3) Update the test set family using the accelerated convergence rule. ; (3.3.1.3.1) Initialization ; (3.3.1.3.2) Execute redundancy constraint judgment: 1) Initialization ; 2) Remember The first in The elements are ; 3) Comparison and The size, if If so, proceed directly to step 4); , then it means separable and ,use Replace elements in Record the replacement for After replacement, based on symmetry constraints, let In Record the replacement for ; Calculate the variables based on the results of the substitution. Then proceed to step 4). 4) Determine the current iteration number Does it meet the requirements? If satisfied, proceed to step 5; otherwise, let Then return to step 2); 5) Minimum value corresponding to replace elements in After the replacement was finally obtained ; (3.3.1.3.3), Order Then return to step (3.3.1.3.2) until... ; (3.3.1.3.4) Obtain the complete diagram : ； (3.3.1.4) Calculation fitness ; (3.3.1.5) Order Then return to step (3.3.1.2) until... ; (3.3.2) Iteratively update according to the fitness using the particle swarm optimization algorithm; (3.3.3) Then return to step (3.3.1.1) until... fitness smallest The best complete diagram is denoted as ; (3.4) will Restore to a binary tree : (3.4.1) Initialize the set of nodes to be processed Initial node ; (3.4.2) Selection Select a node and denote it as , will node As The root node of the current state; (3.4.3) Extraction Contains Given the nodes and edges, generate a graph. ; (3.4.4) Select according to the following formula Tests in and will As Current node The following test; ； (3.4.5) According to the test Will The corresponding fault set is divided into two subsets, which are represented by nodes. and ; Determine the node Does it contain more than 1 fault? Then let ,otherwise, Remain unchanged; Determine the node Does it contain more than 1 fault? Then let ,otherwise, Remain unchanged; treat and After all judgments are completed, from Delete; (3.4.6) Repeat steps (3.4.2)-(3.4.5) until... , After restoration is complete, proceed to step (4). (4) The restored This is the optimal testing strategy.

2. The method for generating optimal test strategies for complex communication systems according to claim 1, characterized in that, The probability The calculation method is as follows: 。 3. The method for generating an optimal test strategy for a complex communication system according to claim 1, characterized in that, The test The determination is based on the principles of roulette, specifically as follows: Generate vectors , Indicates a fault With fault By test The probability of separation; Determine elements exist The corresponding interval in the middle, when element Located in the interval At that time, then corresponding Set as ,Right now: .

4. The method for generating an optimal test strategy for a complex communication system according to claim 1, characterized in that, The fitness The calculation method is as follows: 。

Images