A software testing method based on an autonomously controllable hyper-converged industrial control platform

Abstract

The application discloses a software testing method based on an autonomous controllable super-fusion industrial control platform, which comprises the following steps: obtaining software testing information of the industrial control platform, taking each functional module in the software as an intelligent agent, and inputting a trigger condition in each intelligent agent; performing functional testing on a single intelligent agent according to the trigger condition; analyzing the trigger condition of each intelligent agent to generate a relationship network among the intelligent agents; generating a hybrid testing signal according to a functional testing result of a single intelligent agent and a characteristic relationship of the relationship network; after completing testing of each hybrid testing signal, generating a new testing signal based on a testing result of the last time, and ending the testing when a maximum testing number is reached or a termination condition is reached. The method can realize efficient and comprehensive testing on multiple functional modules in the super-fusion industrial control platform.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of industrial control system testing and software quality technology, in particular to a software testing method based on an autonomously controllable super-fusion industrial control platform. BACKGROUND

[0002] In modern industrial control systems, with the development trend of device intelligence and platform integration, industrial control platforms have gradually evolved into super-fusion systems that integrate data acquisition, real-time control, communication linkage, and human-computer interaction. Such platforms usually consist of multiple functional modules that work together, with each module handling specific tasks and complex calling relationships, data dependencies, and state linkages between modules. Although this system architecture improves control efficiency and flexible response capabilities, it significantly increases the complexity of system testing.

[0003] Traditional software testing methods mainly focus on static function verification and module-level independent testing, making it difficult to effectively cover the linkage logic between modules and the abnormal propagation path across modules. At the same time, due to the vast parameter space, test cases tend to be concentrated in certain intervals, leading to the exposure of untested implicit defects in actual system operation. In addition, the lack of effective feedback mechanisms in traditional testing processes makes it impossible to dynamically optimize testing strategies based on historical test results, resulting in wasted testing resources or missed critical scenarios.

[0004] Especially in the context of increasing autonomy and localization requirements, software testing for industrial control platforms not only needs to verify the correctness of basic functions, but also needs to fully cover boundary behaviors, interaction dependencies, and system stability. How to efficiently, comprehensively, and controllably complete the testing and verification of complex systems within a limited testing period has become one of the key problems restricting the large-scale deployment and long-term stable operation of super-fusion industrial control platforms. SUMMARY

[0005] In view of the above problems, the present application is proposed.

[0006] Therefore, the technical problem solved by the present application is that existing software testing methods have problems such as difficulty in testing module linkage, insufficient parameter coverage, and inability to adaptively optimize testing strategies.

[0007] To solve the above technical problems, the present application provides the following technical solution: a software testing method based on an autonomously controllable super-fusion industrial control platform, comprising:

[0008] Obtain software testing information of the industrial control platform, treat each functional module in the software as an agent, and record trigger conditions in each agent;

[0009] According to the trigger conditions, perform function testing on a single agent;

[0010] analyzing the trigger condition of each agent to generate a relationship network between the agents;

[0011] generating a hybrid test signal according to the functional test result of a single agent and the characteristic relationship of the relationship network;

[0012] After completing the test of each hybrid test signal, a new test signal is generated based on the last test result until the maximum test number is reached or the termination condition is reached, and the test is ended.

[0013] As a preferred scheme of the software testing method based on the self-controllable super-fusion industrial control platform, the software testing information includes all functional modules in the industrial control platform, and the response parameters and value ranges corresponding to each functional module when triggering a response.

[0014] The functional modules include components for implementing various functions in the platform.

[0015] The response parameters include the data types used by each functional module to trigger a response, and the value ranges used to define the behavior boundaries and functional trigger conditions of the functional modules.

[0016] As a preferred scheme of the software testing method based on the self-controllable super-fusion industrial control platform, the functional test of a single agent includes adjusting the response parameters corresponding to each functional module to the value range that triggers a response, selecting the boundary values and any n×m intermediate values of the value range for each response parameter for testing; if the test process can cause the agent to respond, the test is passed; if the test process does not respond, the parameter interval is determined by the bisection method, and the agent and the non-responsive parameters and their parameter intervals, or the parameter intervals of the parameter combinations are output.

[0017] Wherein, n represents the preset number of selected values, and m represents the interval length of the value range divided by the minimum change step of the corresponding response parameter.

[0018] As a preferred scheme of the software testing method based on the self-controllable super-fusion industrial control platform, the relationship network includes taking each agent as a node, and constructing an edge between any two nodes; if there is an intersection between the response parameters and value ranges of the two nodes, the response parameters and value ranges of the intersection are used as the edge; if there is no intersection between the response parameters and value ranges of the two nodes, the edge between the two nodes is disconnected.

[0019] As a preferred scheme of the software testing method based on the autonomously controllable super-fusion industrial control platform, the generation process of the mixed test signal comprises:

[0020] In the functional test of a single agent, the agent that does not respond is removed from the relationship network, and a second relationship network for generating the mixed test signal is obtained;

[0021] Based on the second relationship network, an attention matrix of the agent is constructed;

[0022] The attention weight in the attention matrix is converted into a selection probability, and the agent is randomly selected according to the selection probability;

[0023] In the selected agent, if there is an edge between two agents, a parameter value is randomly selected in the edge corresponding response parameter and value range, and a parameter value is randomly selected in the response parameter and value range that belongs to the trigger condition of the agent but is outside the edge, as the mixed test signal;

[0024] In the generation process of the mixed test signal, the constraints include that there is no conflict between the test values of each agent response parameter in the mixed test signal; one mixed test signal is generated only once; and the number of agents corresponding to each mixed test signal is not less than 2;

[0025] In the attention matrix, each element represents the attention degree of an agent: the use frequency of the agent in the historical record is multiplied by m, and then normalized.

[0026] As a preferred scheme of the software testing method based on the autonomously controllable super-fusion industrial control platform, the new test signal comprises: if the last test is passed, the attention weight of the agent selected in the last test is evaporated, and the evaporation rate is θ%; at the same time, the attention weight of the agent not selected in the last test is enhanced, and the enhancement ratio is θ%; the selection probability is converted according to the adjusted attention, and the agent is randomly selected according to the selection probability;

[0027] After each test signal is generated, the attention of each response parameter in the agent is redistributed: the interval of parameter i is , contains w steps , if the parameter value t in the interval is selected by the mixed test signal, the attention coefficient is increased from 0 to both ends with as the center, and the attention coefficient is d× when separated by d intervals ;

[0028] After the selection of the agent is completed, each response parameter of all agents is obtained, and each response parameter meets the value range of the second relationship network, in the value range of each response parameter, the attention coefficient is normalized to obtain the attention weight of each parameter value, the attention is converted into probability, and the parameter value is randomly selected according to the probability;

[0029] If the last test fails, the abnormal agent combination is filtered in the test result: the test result after excluding x agents is tested one by one, all combinations are traversed, x is measured from 2, x is increased by 1 each time the traversal is completed and the test fails, until the test passes;

[0030] After the test, the abnormal agent combination is output;

[0031] Wherein, θ% represents a preset adjustment ratio, represents the minimum step length of each change of the parameter i, represents the attention coefficient of one unit of the parameter i.

[0032] As a preferred scheme of the software testing method based on the self-controllable super-fusion industrial control platform, wherein: the termination condition comprises, simultaneously satisfying the conditions:

[0033] Each agent is selected at least twice by the mixed test signal;

[0034] Each agent, in the respective parameter interval, the parameter value selected by the mixed test signal presents discreteness;

[0035] The test number is greater than the minimum number.

[0036] A software testing system based on a self-controllable super-fusion industrial control platform using the method described in the application, wherein:

[0037] The acquisition unit obtains the software testing information of the industrial control platform, takes each function module in the software as an agent, and records a trigger condition in each agent;

[0038] The test unit performs function testing on a single agent according to the trigger condition;

[0039] The analysis unit analyzes the trigger condition of each agent to generate a relationship network between the agents; and generates a mixed test signal according to the function test result of a single agent and the characteristic relationship of the relationship network;

[0040] After each mixed test signal is tested, a new test signal is generated based on the last test result until the maximum test number is reached or the termination condition is reached, and the test is ended.

[0041] A computer device comprises a memory and a processor; the memory stores a computer program, wherein the processor implements the steps of the method of any one of the present application when executing the computer program.

[0042] A computer readable storage medium stores a computer program, wherein the computer program implements the steps of the method of any one of the present application when executed by a processor.

[0043] The software testing method based on the autonomously controllable hyper-converged industrial control platform provided by the present application can realize efficient and comprehensive testing of the multifunctional modules in the hyper-converged industrial control platform. First, by constructing the agent model and the relationship network, the triggering dependency and the response relationship between the functional modules can be accurately depicted, and the test path generation with stronger pertinence can be realized. Second, the attention mechanism is introduced to guide the test combination selection, and the test strategy is dynamically adjusted in combination with the historical feedback to improve the utilization efficiency of the test resources. Third, in the parameter selection process, the disturbance control and the discreteness judgment mechanism are used to effectively avoid the parameter value centralized coverage and improve the exploration ability of the input space boundary. In addition, when the test fails, the present application supports the abnormal combination troubleshooting mechanism, which can quickly locate the fault module combination and shorten the problem troubleshooting time. Finally, the overall test process supports adaptive evolution, which continuously optimizes the agent selection and signal configuration in multiple rounds of testing. The present application is suitable for industrial software systems with a large number of modules, complex linkage logic and high requirements for autonomy and controllability, and significantly improves the intelligent degree of testing and the stability guarantee capability. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1 The overall flowchart of the software testing method based on the autonomously controllable hyper-converged industrial control platform provided by the present application is provided. DETAILED DESCRIPTION

[0046] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the protection scope of the present application.

[0047] REFERENCEFigure 1 For an embodiment of the present application, a software testing method based on an autonomously controllable hyper-fusion industrial control platform is provided, comprising:

[0048] S1: Obtain software testing information of the industrial control platform, take each functional module in the software as an agent, and record trigger conditions in each agent.

[0049] Further, the software testing information includes all functional modules in the industrial control platform, and corresponding response parameters and value ranges when each functional module triggers a response. The functional modules include components for implementing various functions in the platform. The response parameters include data types used by each functional module to trigger a response, and the value ranges are used to define the behavior boundaries and function trigger conditions of the functional modules.

[0050] Modeling each functional module in the industrial control platform as an agent can effectively abstract the test units that can independently respond in the system, so that subsequent testing can be directed at behavior rather than code, with stronger reconfigurability and modularization features, suitable for automated testing control of complex systems. By recording the response parameters and their value ranges of each functional module, the activation conditions, behavior boundaries, and state change ranges of each module can be clearly defined, providing constraint information for the generation of test cases, avoiding invalid or out-of-bound test signals, and improving test accuracy and system stability.

[0051] The obtained module information and trigger conditions are important bases for key algorithm inputs such as subsequent relationship network construction, agent combination strategy formulation, and attention matrix initialization, enabling the system to develop structured, intelligent, and multi-round feedback testing strategy evolution.

[0052] S2: Perform functional testing on a single agent according to the trigger conditions.

[0053] Specifically, the response parameters corresponding to each functional module are adjusted to the value range that triggers a response, and boundary values and any n x m intermediate values of the value range are selected for each response parameter for testing. If the testing process causes the agent to respond, the testing is passed. If the agent does not respond during the testing process, the binary approximation method is used to determine the parameter interval that does not respond, and the agent and the non-responsive parameter and its parameter interval, or the parameter interval of the parameter combination are output.

[0054] Where n represents the preset number of selected values, and m represents the interval length of the value range divided by the minimum change step of the corresponding response parameter.

[0055] By conducting coverage testing at the boundary and intermediate values of the response parameters, the basic functional integrity of the module within the input interval can be verified, confirming whether it can be triggered stably under normal input conditions and effectively reducing the risk of functional loss. When some parameter values are found to be unable to trigger the module response during testing, the bisection method can be introduced to accurately lock the non-responsive input interval or parameter combination, helping developers or testers identify blind spots or input-sensitive areas in the module design, improving the accuracy and efficiency of problem discovery. The non-responsive parameters or parameter combinations identified during testing will be used to eliminate invalid agents, optimize the subsequent relationship network construction logic, and avoid generating invalid test signals during subsequent hybrid testing, improving overall testing efficiency and signal effectiveness.

[0056] S3: Analyzing the trigger conditions of each agent to generate a relationship network between the agents.

[0057] Each agent is treated as a node, and an edge is constructed between any two nodes to generate a relationship network. If there is an intersection between the response parameters and value ranges of two nodes, the response parameters and value ranges of the intersection are used as the edge. If there is no intersection between the response parameters and value ranges of two nodes, the edge between the two nodes is disconnected. The relationship network constructed by taking agents as nodes and parameter intersections as edges is essentially a modeling of the input level coupling between internal modules of the system. This network not only reflects the test coupling relationship between modules, but also can be used for subsequent test signal path selection, agent combination screening, abnormal propagation path backtracking, and other key operations. As the structural basis for test signal combination strategies, the relationship network can be used to screen module combinations that have actual linkage possibilities in terms of parameters, thereby avoiding redundant test signals caused by unrelated module combinations and significantly improving testing efficiency and test signal effectiveness.

[0058] By analyzing the response parameters and their value ranges in the trigger conditions between agents, if there is an intersection, it can be considered that two modules may be triggered simultaneously or influence each other under certain input conditions. This potential linkage can be represented by an edge, achieving the modeling transformation from "independent functions" to "interactive behaviors".

[0059] S4: Generating a hybrid test signal based on the functional test results of a single agent and the characteristic relationships of the relationship network.

[0060] The generation process of the hybrid test signal is as follows:

[0061] The second relationship network for hybrid test signal generation is obtained after the non-responding agent in the functional test of a single agent is removed from the relationship network; the second relationship network is generated after the agent that does not respond in the functional test is removed, which effectively avoids including a module without triggering capability in the hybrid test, reduces the generation of invalid test signals from the source, and ensures that the test covers the real module combination that can respond.

[0062] Based on the second relationship network, an attention matrix of the agent is constructed; the attention matrix of the agent is constructed, and the historical use frequency is introduced into the weight calculation to realize the evolution of the test path from “fixed” to “attention-guided”, dynamically adjust the priority of the test focus module, and make the generation of the test signal more consistent with the system operation characteristics.

[0063] The attention weight in the attention matrix is converted into a selection probability (weight value x 100% to obtain), and the agent is randomly selected according to the selection probability; the attention matrix reflects the “attention degree” of each agent, which is converted into a probability distribution for guiding the random selection process of the test combination, which can effectively improve the pertinence of the test strategy and the resource utilization rate. Compared with the traditional uniform or fixed selection strategy, this mechanism can focus the test on the modules that are more likely to make mistakes or more critical, reduce redundant combinations, and improve the coverage quality. The attention weight is derived from the historical use frequency or response performance, representing the participation and performance characteristics of the agent in the past test. Mapping it to the selection probability can make the system learn and evolve in the test process, realize the test behavior of “high-attention agent priority test, low-active module gradual activation”, and thus construct a test strategy with learning ability. Through the “probability-driven + random selection” mode, the test retains a certain degree of exploratory and uncertainty, avoids the test path from converging and falling into a local optimum due to excessive attention, and helps to discover potential complex linkage abnormalities and boundary behaviors. In the attention matrix, each element represents the attention degree of an agent: the use frequency of the agent in the historical record is multiplied by m, and then normalized to obtain.

[0064] If there is an edge between the selected agents, a parameter value is randomly selected in the response parameter and value range corresponding to the edge, and a parameter value is randomly selected in the response parameter and value range that belongs to the trigger condition of the agent but is outside the edge, as the hybrid test signal;

[0065] To say, when there is an edge between two selected agents, it means that they have intersection or dependency relationship in trigger condition or response parameter. At this time, selecting parameter value in the parameter interval corresponding to the "edge" can ensure that the test signal has actual input coupling, truly reflects the collaborative triggering or data interaction mechanism between modules, and helps to verify the stability and correctness of the system in the case of multi-module linkage. At the same time, random selection in the non-intersection part parameter range of each agent itself is to ensure that the original function trigger condition of the module is completely covered, and to avoid ignoring the single module internal logic test due to attention to module linkage, so as to improve the overall test. This combination of "structure related + module independent" makes the mixed test signal have cross-module interaction semantics and maintain the ability to explore the function boundary of a single module. It simulates the signal state of more complex input combination in the real industrial running scene, and improves the triggering ability of the system to complex exceptions.

[0066] In the generation process of the mixed test signal, the constraints include that in the mixed test signal, there is no conflict between the test values of each agent response parameter; one mixed test signal (exactly the same mixed test signal) is generated only once (if the corresponding parameters are the same but the specific parameter values are different, it is considered as different mixed test signal); the number of agents corresponding to each mixed test signal is not less than 2.

[0067] S5: After completing the test of each mixed test signal, generate a new test signal based on the test result of the last time, until the maximum test number is reached or the termination condition is reached, end the test.

[0068] Further, the new test signal includes that if the last test passes, evaporate the attention weight of the selected agent in the last test, with an evaporation rate of θ%, and enhance the attention weight of the agent not selected in the last test, with an enhancement ratio of θ%; convert the adjusted attention into selection probability, and randomly select the agent according to the selection probability.

[0069] After each test signal is generated, the attention of each agent in the response parameter is redistributed: let the interval of parameter i , contains w steps , if the parameter value t in the interval is selected by the mixed test signal, then take as the center, start from 0 and increase the attention coefficient to both ends, and if is separated by d intervals, the attention coefficient is d .

[0070] After the selection of the agent is completed, each response parameter of all agents is obtained, and each response parameter meets the value range of the second relationship network. In the value range of each response parameter, the attention coefficient is normalized to obtain the attention weight of each parameter value. The attention is converted into a probability, and the parameter value is randomly selected according to the probability.

[0071] It is known that by introducing an "evaporation-enhancement mechanism", when a certain round of testing is passed, the attention of the agent participating in the test is reduced (evaporation θ%), and the attention of the agent not participating in the test is increased (enhancement θ%), so as to realize the tilt of the test resource to the uncovered area, thereby avoiding repeated combination and expanding the test coverage, and making the test strategy have the dual ability of "memory" and "exploration". Through the attention redistribution of the local interval centered on the selected parameter value, the parameter attention weight gradient of "strong center and weak edge" is constructed, the priority attention of the system to the area near the activated parameter is simulated, and the possibility of exploring the boundary is reserved, so as to improve the test density of the key boundary area and effectively trigger potential abnormalities. The step-level attention coefficient of each response parameter is normalized to a probability distribution, and the probability control of the parameter value selection is realized instead of uniform sampling. This way not only maintains the randomness of the test signal, but also enables the signal to concentrate in the high-attention area, improving the sensitivity and triggering efficiency of the test signal.

[0072] In the two levels of agent selection and parameter value selection, the attention adjustment and probability sampling mechanism is introduced, forming a sustainable optimization and self-evolution test generation process, so that the system test goes from a static test set to a dynamic strategy system, which is more suitable for multi-round linkage test scenarios in complex systems.

[0073] By integrating feedback adjustment, local guidance and probability distribution control mechanism, fine control and strategy evolution of test signal generation are realized, so as to improve the comprehensiveness, efficiency and intelligence level of the test process.

[0074] If the last test is not passed, the abnormal agent combination is filtered in the test result: the test result after excluding x agents is tested one by one, and all combinations are traversed. X is measured from 2, and x is increased by 1 each time the traversal is completed and the test is not passed, until the test is passed.

[0075] After the test, the abnormal agent combination is output.

[0076] wherein θ% represents a preset adjustment ratio, represents the minimum step length of the change of parameter i each time, represents the attention coefficient of one unit of parameter i.

[0077] To say, when a mixed test fails, the system may be due to the linkage between multiple modules, parameter conflict or implicit coupling causes abnormal. Through the way of "one by one to exclude x agents", the combination test tour can automatically identify the smallest combination subset that triggers the exception without relying on manual analysis, and realize the automatic tracing and diagnosis positioning of test exception. Starting from excluding 2 agents, gradually increasing the number of exclusions x, testing all combinations each time until the test passes, which can effectively narrow the scope of the exception and avoid the computational overhead of brute force enumeration of all subsets. This mechanism not only takes into account the completeness of test coverage, but also improves the convergence speed of exception positioning. This mechanism can not only find problems in a single module, but also find combination-type exceptions that only occur when multiple modules are combined. Such problems are usually difficult to capture in traditional testing, and designing this mechanism can effectively uncover cross-module implicit fault modes and enhance the reliability analysis capability of the system. The agents identified as abnormal combinations can be fed back to the relationship network or attention matrix as input to dynamically adjust the combination weight or disconnect related edge connections, thereby optimizing the subsequent test strategy and avoiding known abnormal paths, forming a self-correcting closed loop.

[0078] To say, the termination condition includes, while satisfying the condition: each agent is selected at least twice by the mixed test signal. Each agent, within its respective parameter interval, has a discrete parameter value selected by the mixed test signal. The number of tests is greater than the minimum number.

[0079] To know, by setting the condition that "each agent is selected at least twice by the mixed test signal", it can ensure that all functional modules in the platform are covered by testing, prevent some modules from not participating in testing due to low probability distribution, and improve the integrity and fairness of testing. The introduction of the judgment condition of "parameter value presents discreteness" (and the setting of the judgment threshold index of discreteness) is to prevent the test signal from concentrating in a certain local range of the parameter interval. By covering multiple representative discrete values, the response characteristics of the module in the boundary, extreme value, and transition region can be better detected, thereby improving the ability to discover implicit defects or boundary abnormalities. Setting the test number to be greater than a certain minimum threshold is to ensure the basic size of the test coverage, avoid premature termination of the test due to the satisfaction of other conditions, and ensure that the overall test has sufficient data support and representativeness.

[0080] In this embodiment, the calculation of "parameter value presents discreteness" is realized by the following steps:

[0081] Step 1: Divide the parameter interval into k equal subintervals.

[0082] Step 2: Count the number of parameter values (frequency) falling into each subinterval.

[0083] Step 3: Calculate the standard deviation σ and the average value μ of the frequency distribution.

[0084] Step four: Calculate the discreteness index: RDI = 1 - σ / μ.

[0085] In other optional embodiments, the calculation method of "parameter value presents discreteness" is not limited to the form of the ratio of standard deviation to mean value, but can also use a variety of alternative or supplementary indicators to adapt to different test requirements and data distribution characteristics. For example, the distribution uncertainty of parameter values in the interval can be measured by the method of information entropy, that is, the parameter interval is divided into several subintervals, the distribution frequency of parameter values in each interval is counted, and the entropy value is calculated. If the entropy value is close to the theoretical maximum entropy, it means that the distribution is more uniform and the discreteness is better. Another optional method is to judge based on interval coverage, that is, only the ratio of the number of subintervals where the parameter value falls to the total number of intervals is counted. If most of the intervals are covered, it means that the distribution has good discreteness. This method is suitable for quickly evaluating the test breadth. In addition, the minimum interval between adjacent parameter values can also be used for calculation to determine whether the values are excessively concentrated in a certain area, thereby avoiding "clustered" coverage. By sorting the selected parameter values, the range or standard deviation of adjacent differences is calculated. The larger the interval and the more balanced the distribution, the stronger the discreteness. The above methods can be flexibly selected according to actual application scenarios, or combined with the original method to form a more robust discreteness evaluation system.

[0086] On the other hand, the embodiment also provides a software testing system based on an autonomous controllable super-fusion industrial control platform, which comprises:

[0087] A collection unit acquires software testing information of the industrial control platform, takes each functional module in the software as an agent, and records trigger conditions in each agent.

[0088] A test unit performs functional testing on a single agent according to the trigger conditions.

[0089] An analysis unit analyzes the trigger conditions of each agent, generates a relationship network between the agents, and generates a hybrid test signal according to the functional test results of a single agent and the characteristic relationship of the relationship network.

[0090] After completing the test of each hybrid test signal, a new test signal is generated based on the test results of the last time, and the test is ended when the maximum test number is reached or the termination condition is reached.

[0091] If the above functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or parts of the present application that make contributions to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0092] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, which can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch instructions from an instruction execution system, apparatus or device and execute the instructions, or in conjunction with such instruction execution system, apparatus or device. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transport a program for use by or in connection with an instruction execution system, apparatus or device or in conjunction with such instruction execution system, apparatus or device.

[0093] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection having one or more wires (electrical devices), a portable computer diskette (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CD ROM). In addition, the computer-readable medium can even be paper or other suitable medium on which the program can be printed, as the program can be electronically obtained, for example, by optical scanning of the paper or other medium, followed by editing, interpreting or otherwise processing the program as necessary, and then storing it in a computer memory.

[0094] It should be understood that portions of the present application can be implemented in hardware, software, firmware, or combinations thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, implementation can be with any or a combination of the following technologies, which are all well known in the art: a discrete logic circuit having logic gates for implementing logic functions upon an application of data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0095] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and all of them should be covered in the scope of the claims of the present application.

Claims

1. A software testing method based on an autonomously controllable hyper-converged industrial control platform, characterized in that, The method comprises the following steps: acquiring software test information of an industrial control platform, taking each functional module in the software as an intelligent agent, and recording a trigger condition in each intelligent agent; performing functional test on a single intelligent agent according to the trigger condition; analyzing the trigger condition of each intelligent agent to generate a relationship network among the intelligent agents; generating a hybrid test signal according to the functional test result of the single intelligent agent and the characteristic relationship of the relationship network; generating a new test signal based on the test result of the last test after completing the test of each hybrid test signal, and ending the test when the maximum test number is reached or the termination condition is reached; the relationship network comprises taking each intelligent agent as a node and constructing an edge between any two nodes; if there is an intersection between the response parameters and the value range of the two nodes, the response parameters and the value range of the intersection are taken as the edge; if there is no intersection between the response parameters and the value range of the two nodes, the edge between the two nodes is disconnected; the generation process of the hybrid test signal comprises: obtaining a second relationship network for hybrid test signal generation by removing the intelligent agents that do not respond during the functional test of the single intelligent agent from the relationship network; constructing an attention matrix of the intelligent agents based on the second relationship network; transforming the attention weight in the attention matrix into a selection probability, and randomly selecting the intelligent agents according to the selection probability; in the selected intelligent agents, if there is an edge between two intelligent agents, a parameter value is randomly selected in the edge corresponding response parameter and value range, and a parameter value is randomly selected in the response parameter and value range that belongs to the trigger condition of the intelligent agent but is outside the edge, as the hybrid test signal; in the generation process of the hybrid test signal, the constraints comprise that there is no conflict between the test values of the response parameters of each intelligent agent in the hybrid test signal; one hybrid test signal is generated only once; and the number of intelligent agents corresponding to each hybrid test signal is not less than 2; each element in the attention matrix represents the attention degree of an intelligent agent: the use frequency of the intelligent agent in the historical record is multiplied by m, and then normalized to obtain; the new test signal comprises: if the last test is passed, the attention weight of the intelligent agent selected in the last test is evaporated at a rate of θ%, and the attention weight of the intelligent agent not selected in the last test is enhanced at a rate of θ%; the adjusted attention is transformed into a selection probability, and the intelligent agents are randomly selected according to the selection probability; After each test signal generation, the attention is redistributed among the response parameters in each agent: Let the interval between parameters i , contain w steps . If a parameter value t in the interval is selected by the mixed test signal, then the attention coefficient is increased from 0 towards both ends, centered at , and is d x at intervals of d from ; after the selection of the intelligent agents is completed, the response parameters of all intelligent agents and the value range of each response parameter that meets the second relationship network are obtained, the attention weight of each parameter value is obtained by normalizing the attention coefficient in the value range of each response parameter, the attention is transformed into a probability, and the parameter value is randomly selected according to the probability; if the last test is not passed, the abnormal intelligent agent combination in the test result is screened: the test result after excluding x intelligent agents one by one is traversed, x is measured from 2, x is increased by 1 each time the traversal is completed and the test is not passed, until the test is passed. Output an abnormal agent combination after testing; wherein, denotes the minimum step size for each change of the parameter i, denotes the attention coefficient for one unit of the parameter i.

2. The software testing method based on an autonomously controllable hyperconverged industrial control platform as claimed in claim 1, characterized in that: The software test information includes all functional modules in the industrial control platform, and corresponding response parameters and value ranges of each functional module when triggering a response; The functional modules include components for implementing various functions in the platform; The response parameters include data types of each functional module for triggering a response, and value ranges for defining the behavior boundary and function triggering condition of the functional module.

3. The software testing method based on the autonomically controllable hyperconverged industrial control platform according to claim 2, characterized in that: The function test on a single agent includes adjusting the corresponding response parameters of each functional module to the value range triggering a response, and selecting boundary values and any n x m intermediate values of the value range for each response parameter for testing; If the agent responds in the test process, the test is passed; if the agent does not respond in the test process, the parameter interval of the non-response parameter is determined by the bisection method, and the agent, the non-response parameter and its parameter interval, or the parameter interval of the parameter combination are output; Wherein, n represents the preset number of selected values, and m represents the interval length of the value range divided by the minimum change step of the corresponding response parameter.

4. The software testing method based on the autonomically controllable hyperconverged industrial control platform according to claim 3, characterized in that: The termination condition includes the following conditions being met at the same time: Each agent is selected by the mixed test signal at least twice; Each agent, in the respective parameter interval, is selected by the mixed test signal with discrete parameter values; The test number is greater than the minimum number.

5. A software test system based on an autonomously controllable super-fusion industrial control platform using the method of any one of claims 1-4, characterized in that: An acquisition unit obtains software test information of the industrial control platform, takes each functional module in the software as an agent, and records a triggering condition in each agent; A test unit performs a function test on a single agent according to the triggering condition; An analysis unit analyzes the triggering condition of each agent to generate a relationship network between the agents; A mixed test signal is generated according to the function test result of a single agent and the characteristic relationship of the relationship network; After completing the test of each mixed test signal, a new test signal is generated based on the test result of the last test, and the test is ended when the maximum test number is reached or the termination condition is reached.

6. A computer device comprising: A memory and a processor; The memory stores a computer program, and the processor executes the computer program to implement the steps of the method of any one of claims 1-4.

7. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to implement the steps of the method of any one of claims 1-4.

Images