Low-code flexible complex process customization method for intelligent fault management
Patent Information
- Application Number
- CN202310694125.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-06-13
- Publication Date
- 2026-09-22
- Estimated Expiration
- 2043-06-13
AI Technical Summary
[0010]本发明的目的是提供一种面向智能故障管理的低代码柔性化复杂流程定制方法，以解决当前智能故障管理平台普适性不高、流程设计时柔性化程度低的问题，解决当前智能故障管理流程设计耗时长、迭代速度慢、无法快速落地部署的问题，提高智能能故障管理流程运行时柔性化程度，提高智能故障管理流程运行时的鲁棒性
[0050](1)本发明使用低代码技术，通过智能推荐机制实现智能故障管理流程的半自动设计，能够快速简便地进行复杂流程的个性化设计，仅在节点的设置时涉及简单的配置，如一些选项的选择、简单条件的编写，不需要进行源代码的编写修改，从而大大降低的设计人员的代码能力要求，降低设计人员的培训成本，减少流程设计的时间，从而提高智能故障管理流程的迭代速度，可以大幅缩减智能故障管理流程的落地部署时间，同时也提高智能故障管理平台的普适性，实现针对不同场景定制化设计流程。
Smart Images
Figure CN116804928B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of fault management technology, and in particular relates to a low-code, flexible, and complex process customization method for intelligent fault management. Background Technology
[0002] Fault management encompasses the entire lifecycle of a fault, including fault discovery and standardized handling, fault isolation, and fault storage and retrieval. Fault discovery and standardized handling refers to identifying faults through fault detection and classifying them according to their severity and level. Fault isolation involves diagnosing and comprehensively analyzing faults to determine their location and cause. Fault storage and retrieval involve storing relevant fault information in a database and providing query and statistical information, including faulty equipment, fault occurrence time, fault symptoms, and possible causes.
[0003] Low-code technology is a software development technique derived from advanced software development techniques. It enables users to build and deliver application software more quickly and with less coding through a visual approach, thereby comprehensively reducing the costs of software development, configuration, deployment, and training.
[0004] Currently, there are relatively few fault management platforms on the market, with fault prediction platforms being more prevalent. These fault prediction platforms generally lack full lifecycle management of faults, and the few existing fault management platforms are mostly limited to a single domain, with relatively fixed processes and low flexibility. Low-code technology is primarily used in OA (Office Automation) systems.
[0005] In the field of intelligent fault management, the current common practice is to set up fixed automated processes for a single device or field. This approach has the following drawbacks:
[0006] (1) Due to the differences in the causes and phenomena of faults in different fields, the intelligent fault management process needs to be customized at some nodes. Therefore, conventional intelligent fault management methods are often not universally applicable and are difficult or even impossible to apply to other fields.
[0007] (2) Since the fault management process is set up for a single device or field, the process is often not visualized and there is no function to edit the process. The process needs to be manually coded during the design. Therefore, the conventional intelligent fault management method takes a long time to design the process, has a slow iteration speed, and cannot achieve rapid deployment.
[0008] (3) Conventional intelligent fault management methods are basically set up with fixed processes, and the processes will not be changed during operation. Therefore, their flexibility during operation is not high, and they cannot fully improve the intelligence of the entire process.
[0009] (4) When a conventional intelligent fault management process encounters an operational anomaly, it can only stop the process to repair it. This means that fault management cannot be performed while repairing the anomaly, and the overall robustness of the process is not high. Summary of the Invention
[0010] The purpose of this invention is to provide a low-code, flexible, and complex process customization method for intelligent fault management, in order to solve the problems of low universality and low flexibility in process design of current intelligent fault management platforms, as well as the problems of long design time, slow iteration speed, and inability to be quickly deployed in current intelligent fault management processes, thereby improving the flexibility and robustness of intelligent fault management processes during operation.
[0011] This invention provides a low-code, flexible, and complex process customization method for intelligent fault management, including:
[0012] Phase 1, Process Design Phase:
[0013] (1) Analyze the requirements of the intelligent fault management process, and make the selection of the process features to be designed on the provided interface based on the requirements analysis results, including application scenarios, fault phenomena, and fault causes.
[0014] (2) Based on the intelligent fault management process design reference example recommendation function, intelligently recommend process design reference examples to designers according to requirements, and correct the recommendation results according to the selection rate of designers and the degree of modification of the selected process reference examples, including personalized modification of process links and configuration of task nodes.
[0015] (3) Deployment process;
[0016] (4) Simulate and run the model to verify its feasibility. If the verification result is infeasible, return to step (2) to re-modify the process links and configure task nodes. If the verification result is feasible, execute stage two:
[0017] Phase Two, Process Execution Phase:
[0018] 1) Dynamically add or remove task nodes according to rules, including:
[0019] The performance of a series of interconnected task nodes is evaluated using metrics including runtime, signal-to-noise ratio, harmonic mean of accuracy and recall after data processing, true positive rate, and precision. The evaluation results are recorded as a vector, resulting in a vector representing the performance of each task node, e. i The execution effect vector of a series of task nodes is
[0020] The expected results to be achieved by each task node are recorded using vectors. These vectors are obtained by vectorizing the text description of the algorithm corresponding to each task node, including:
[0021] When the algorithm corresponding to the task node is designed, it uses text descriptions of running time and running quality. Vectorizing the text descriptions yields the expected running effect vector of the task node.
[0022] The expected overall performance of the task node is recorded as a vector P, and the distance between the two vectors E and P is used as the judgment criterion:
[0023]
[0024] Where ||·||2 is the 2-norm of the vector.
[0025] When d(E, P) is less than the set threshold d t When d(E, P) is greater than the set threshold d, the operation process remains unchanged; t During this process, task nodes in the execution flow are dynamically added or removed; the rule for addition and removal is: obtain the execution effect vector after the task node jumps. When the minimum value of d(E′, P) is less than the set threshold d t When the process is in progress, a task node jump operation is performed, and the jump node is the task node that minimizes d(E′, P); when the minimum value of d(E′, P) is greater than a set threshold d t When adding a task node, the process is to add a task node; when adding a task node, the node must satisfy d(E″, P) < d. t The task node is E″, which is the prediction vector of the running effect after adding the task node. When there are multiple task nodes that meet the conditions, the task node that minimizes d(E″, P) is added.
[0026] 2) Handling abnormal nodes in the intelligent fault management process for service call chain failures, including:
[0027] Monitor the service call chain status of the task nodes in the process. When a task node is found to be running abnormally due to a service call chain failure, pause the intelligent fault management process, change the configuration of the abnormal node to generate a new process template, and migrate the currently abnormal process running instance to the new process template. At the same time, capture the service call chain abnormality and notify the technical personnel to repair the abnormal node.
[0028] Phase Three, Process Closure Phase:
[0029] Data related to the process, including process execution time, node operation status, and process design status, is saved to the database module for later use.
[0030] Furthermore, the recommended function calculation method for the intelligent fault management process design reference example described in step (2) of phase one is as follows:
[0031] Generate a user vector u = (x1, x2, ..., y) based on the designer's selection. n ), where x i These are the feature values related to the process to be designed; the process reference examples in the process case library have their own process vectors c = (y1, y2, ..., y...). n ), where y i The feature values of the process reference example;
[0032] Find the cosine values of the user vector u and the process vector c, and obtain the similarity function between the user vector and the process vector of the intelligent fault management process reference example:
[0033]
[0034] When process reference examples are recommended to designers, the recommendations are adjusted based on the designers' selection rate and the extent of modifications made to the selected process reference examples. The designers' selection rate... s represents the number of times the process reference example was selected, n represents the number of times the process reference example was recommended, and the degree of modification of the process reference example. r i N represents the number of single-step modifications to the selected process reference example, where N is the total number of links and nodes in the selected process reference example; a threshold S is set for the designer's selection rate. T The threshold R for the degree of modification to the process reference example. T When S < S T Or R > R T When the weighting function reaches 0, the weighting function is:
[0035]
[0036] Finally, the recommended function for the intelligent fault management process design reference example is obtained:
[0037] F = θsim(u, c).
[0038] Furthermore, the steps for designers to design an intelligent fault management process based on the recommended function in the intelligent fault management process design reference example are as follows:
[0039] ① Designers select the feature information of the process to be designed on the provided page;
[0040] ② Generate user vectors based on the selection;
[0041] ③ Use the recommendation function of the intelligent fault management process design reference example to calculate the recommended value of the process reference example in the process case library, and recommend the process design reference example based on the calculation result;
[0042] ④ Designers select appropriate process design reference examples and design processes based on them;
[0043] ⑤ Based on the design results provided by the designers, revise the weighted function value of the process design reference example.
[0044] Furthermore, step 2) in phase two also includes:
[0045] If the designer has not designed the reconfiguration of task nodes, the internal matching mechanism will perform node configuration matching changes, including:
[0046] Set feature values for the relevant attributes configured for the nodes. These attributes include the type of input data required for the corresponding service, the service runtime, and the format of the service output. Based on the set feature values, obtain the configuration feature vector A = (a1, a2, ..., a...). n ), where a i Configure characteristic values for relevant attributes for abnormal task nodes; different task node configurations available for reconfiguration in the same node configuration library have their own characteristic vectors B = (b1, b2, ..., b...). n ), where b i Configure the feature values of relevant attributes for other task nodes in the configuration library for nodes of the same type, and calculate the matching function results:
[0047]
[0048] Where r(A, B) represents the correlation between feature vector A and feature vector B. For a i The average value, For b i The average value is used to determine the node configuration closest to 1, which is then used as the replacement configuration for the abnormal node.
[0049] By employing the above approach, and through a low-code, flexible, and complex process customization method for intelligent fault management, the following technical effects are achieved:
[0050] (1) This invention uses low-code technology and achieves semi-automatic design of intelligent fault management process through intelligent recommendation mechanism. It can quickly and easily carry out personalized design of complex process. Only simple configuration is involved when setting nodes, such as the selection of some options and the writing of simple conditions. There is no need to write and modify the source code, which greatly reduces the coding ability requirements of designers, reduces the training cost of designers, and reduces the time of process design. This improves the iteration speed of intelligent fault management process, can greatly reduce the deployment time of intelligent fault management process, and also improves the universality of intelligent fault management platform, realizing customized design process for different scenarios.
[0051] (2) The present invention can dynamically add task nodes during the operation of the intelligent fault management process, thereby improving the flexibility and intelligence of the process operation.
[0052] (3) When a task node experiences an operational anomaly, the present invention can change the configuration of the abnormal node and continue the operation of the intelligent fault management process while repairing the abnormal node, thereby improving the overall efficiency of the process and realizing fault management when repairing process anomalies.
[0053] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description
[0054] Figure 1 This is a schematic diagram of the functional modules in the low-code flexible complex process customization method for intelligent fault management of the present invention;
[0055] Figure 2 This is a flowchart of the process design, process operation, and process termination stages in one embodiment of the present invention;
[0056] Figure 3 This is a diagram illustrating the implementation steps of the recommendation mechanism in a reference example of intelligent fault management process design in one embodiment of the present invention.
[0057] Figure 4 This is a schematic diagram illustrating the dynamic addition and removal rules of task nodes during the operation of the intelligent fault management process in one embodiment of the present invention;
[0058] Figure 5 This is a schematic diagram of node reconfiguration in one embodiment of the present invention;
[0059] Figure 6 This is a schematic diagram illustrating the node reconfiguration matching principle in one embodiment of the present invention. Detailed Implementation
[0060] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.
[0061] like Figure 1 As shown, the technical architecture of the low-code flexible complex process customization method for intelligent fault management in this embodiment can be divided into four modules: a process design module, a process execution module, a data storage module, and a database module. The process design module mainly includes intelligent recommendation functions for process design references, simulation execution functions, and other functions for process design, such as node configuration. The process execution module includes functions for dynamically adding or removing task nodes, handling abnormal nodes, and other functions specific to intelligent fault management process execution instances. The data storage module is mainly used for saving information related to the intelligent fault management process after its execution. The database module includes tables related to process resource storage, process execution, organization and user information, process history records, and general process data; these tables provide data support for the implementation of the functions of this invention.
[0062] like Figure 2 As shown in the figure, this embodiment provides a low-code, flexible, and complex process customization method for intelligent fault management. The method consists of three stages: process design stage, process execution stage, and process termination stage, which are described in detail below:
[0063] Phase 1: Process Design Phase
[0064] In this stage, designers complete the personalized customization of the intelligent fault management process and verify the feasibility of the designed intelligent fault management process.
[0065] During the design process, designers can first conduct a requirements analysis of the intelligent fault management process. Based on the results of the requirements analysis, they can select the characteristics of the process to be designed on the provided interface, including application scenarios, fault phenomena, fault causes, and other relevant information. Based on the designer's selections, a user vector u = (x1, x2, ..., x...) can be generated. n ), where x i These are the feature values related to the process to be designed. The process reference examples in the process case library also have their own process vector c = (y1, y2, ..., y...). n ), where y i The feature values are the information related to the process reference example. The similarity between the two vectors can be determined by calculating their cosine values. Therefore, the similarity function between the user vector and the process vector of the intelligent fault management process reference example can be obtained:
[0066]
[0067] This represents the ideal recommendation effect. In reality, feedback needs to be incorporated to adjust the recommendation effect. Therefore, a weighted function for user feedback needs to be designed. When a process reference example is recommended to designers, the recommendation result can be adjusted based on the designers' selection rate and the degree of modification made to the selected process reference example. Designer selection rate... Where s represents the number of times the process reference example is selected, and n represents the number of times the process reference example is recommended. (Process reference example modification level) Where, r i Let N be the number of links and nodes modified in a single process reference example, and N be the total number of links and nodes in the selected process reference example. Considering practical considerations, a threshold S for the designer's selection rate should be set. T The threshold R for the degree of modification to the process reference example. T When S < S T Or R > R T When , the value of the weighting function is 0. Therefore, the weighting function is:
[0068]
[0069] Finally, the recommended function for the intelligent fault management process design reference example is obtained:
[0070] F = θsim(u, c)
[0071] The above is a recommended reference example mechanism for the intelligent fault management process design of this invention. For example... Figure 3 As shown, the steps for designers to use this mechanism to design intelligent fault management processes are as follows: ① Designers select the feature information of the process to be designed on the provided page; ② A user vector is generated based on the selection; ③ The recommended value of the process reference examples in the process case library is calculated using the recommendation function of the intelligent fault management process design reference examples, and a process design reference example is recommended based on the calculation result; ④ Designers select a suitable process design reference example and design the process based on it; ⑤ The weighted function value of the process design reference example is adjusted based on the designer's design results.
[0072] Phase Two: Process Execution Phase
[0073] During the intelligent fault management process operation phase, users can activate relevant functions for specific process execution instances as needed, such as pausing, activating, and canceling the process. Simultaneously, the system will also intelligently activate certain functions for special circumstances.
[0074] (1) Construction of dynamic addition and deletion rules for task nodes during the intelligent fault management process
[0075] When unforeseen circumstances lead to redundancy in some nodes or a shortage of current nodes, this invention will dynamically add or remove task nodes according to the task node dynamic addition and removal rules during the operation of the intelligent fault management process to ensure the normal operation of the process. The node dynamic addition and removal rules during the operation of the intelligent fault management process are constructed as follows:
[0076] The performance of a series of interconnected task nodes can be evaluated using metrics such as runtime, signal-to-noise ratio, harmonic mean of accuracy and recall after data processing, true positive rate, and precision. The evaluation results are recorded using a vector. Therefore, the performance vector of each task node can be denoted as e. i The running effect of a series of task nodes is The expected results of these task nodes can also be recorded using vectors. These vectors are obtained by vectorizing the text descriptions of the algorithms corresponding to these task nodes. When the algorithm for each task node is designed, it can be described in text terms such as runtime and performance quality. Vectorizing these text descriptions yields the expected runtime effect vector for the task node. The overall expected runtime effect of the task node is recorded as vector P. The distance between two vectors is used as a criterion for judgment.
[0077]
[0078] Where ||·||2 is the 2-norm of the vector.
[0079] When d(E, P) is less than the set threshold d t When d(E, P) is greater than the set threshold d, the operation process remains unchanged; t During execution, task nodes are dynamically added or removed. The rule for addition or removal is: obtain the execution effect vector after the task node jump. When the minimum value of d(E′, P) is less than the set threshold d t When the process is in progress, a task node jump operation is performed, and the jump node is the task node that minimizes d(E′, P); when the minimum value of d(E′, P) is greater than a set threshold d t During this process, the workflow performs a task node addition operation. When adding a task node, a node satisfying d(E″, P) < d will be added. t The task nodes are defined as follows: E″ is the predicted vector of the running effect after adding the task node. When there are multiple task nodes that meet the conditions, the task node that minimizes d(E″, P) is added. A specific illustration is shown below. Figure 4 As shown.
[0080] (2) Setting up intelligent fault management process and abnormal node handling methods for service call link failures.
[0081] Inevitably, during the operation of the intelligent fault management process, situations may arise where task nodes malfunction due to service link call failures. To address this issue, this invention proposes a method for handling abnormal nodes configured in the intelligent fault management process task nodes. The specific implementation of this method is as follows:
[0082] During the operation of the intelligent fault management process, the low-code flexible complex process customization method proposed in this invention monitors the service call chain status of process task nodes. When a task node is detected to be running abnormally due to a service call chain failure, the abnormal node handling method is initiated. At this time, the system first pauses the intelligent fault management process, modifies the configuration of the abnormal node to generate a new process template, and migrates the currently abnormal process execution instance to the new process template. Simultaneously, it also captures service call chain anomalies and notifies technical personnel to repair the abnormal node. The entire replacement process is as follows: Figure 5 As shown. In this way, the repair and handling of abnormal task nodes that experience service call chain failures can be carried out without interrupting the intelligent fault management process, thereby improving the robustness of the process.
[0083] Generally, task node reconfiguration is determined by the designers during the process design phase; that is, the designers decide which node configuration to replace the abnormal node configuration. However, if the designers did not design task node reconfiguration, an internal matching mechanism can be used to match and change the node configuration. The specific principle is as follows: Feature values are set for the relevant attributes of the node configuration. These attributes include the type of input data required by the corresponding service, the service runtime, the format of the service output results, etc., thus obtaining the configuration feature vector A = (a1, a2, ..., a...). n ), where a i Configure characteristic values for relevant attributes of abnormal task nodes. Different task node configurations available for reconfiguration in the same node configuration library also have their own characteristic vectors B = (b1, b2, ..., b...). n ), where b i Configure feature values for relevant attributes for other task nodes in the configuration library for nodes of the same type. Calculate the matching function result:
[0084]
[0085] Where r(A, B) represents the correlation between feature vector A and feature vector B. For a i The average value, For b i The average value is calculated. The node configuration with the result closest to 1 is used as the replacement configuration for the abnormal node. A schematic diagram of the entire process is shown below. Figure 6 As shown.
[0086] Phase Three: Process Closure Phase
[0087] This stage primarily involves saving process-related information, including process execution time, node status, and process design details. This data will be stored in the database module for later use.
[0088] This low-code, flexible, and complex process customization method for intelligent fault management has the following technical advantages:
[0089] 1) This invention utilizes low-code technology to enable flexible customization of complex intelligent fault management processes in different fields. Based on the current need for intelligent fault management to achieve flexible and rapid deployment according to different fields and different focuses in the same field, it can quickly and autonomously design personalized complex processes for different scenarios, realize flexibility in process design, improve the universality of intelligent fault management platform, and solve the problems of low universality and low flexibility in process design of current intelligent fault management platforms.
[0090] 2) This invention, by incorporating an intelligent recommendation mechanism, enables semi-automatic design of intelligent fault management processes, reducing design time, increasing iteration speed, and facilitating rapid deployment. This solves the problems of long design times, slow iteration speeds, and inability to quickly deploy current intelligent fault management processes. Currently, most common low-code platforms require manual design of the entire process, necessitating users to manually design all process links and configure all task nodes. This design pattern is unsuitable for intelligent fault management process design, as these processes are often complex, characterized by long process links and numerous node configurations. To address this, this invention designs an intelligent fault management process reference example recommendation mechanism that intelligently recommends process design reference examples to users. Users can modify these reference examples to design their own intelligent fault management processes, significantly reducing manual operations and greatly improving design speed.
[0091] 3) This invention achieves real-time construction of intelligent fault management process templates by adding dynamic rules for adding and removing process nodes, thereby improving the intelligence and flexibility of the process during operation. In the intelligent fault management process, some consecutive nodes are related, such as fault normalization processing, and their purpose is to prepare for fault classification. However, in the actual operation of the process, due to some unforeseen circumstances, some nodes may become redundant or the current node may not meet the requirements. To address this scenario, this invention constructs a dynamic addition and removal rule for task nodes that takes effect during the operation of the intelligent fault management process. The operating effect of task nodes is used as the judgment criterion to dynamically jump to and add task nodes, thereby realizing the dynamic construction of process operation instances and improving the intelligence and flexibility of the intelligent fault management process.
[0092] 4) This invention, through its method for handling abnormal nodes during service call link failures, enables the intelligent fault management process to operate normally while repairing service call link failures at abnormal nodes. This achieves fault management during anomaly repair, improves the overall operational efficiency of the process, and enhances the robustness of the intelligent fault management process. During the operation of the intelligent fault management process, due to various reasons, it is inevitable that task nodes will malfunction due to service call link failures. Currently, the common approach is to pause the process and wait for technicians to repair the service call of the abnormal node before restarting it. This method is not very efficient. Therefore, to improve the overall efficiency of the intelligent fault management process, this invention adopts a method for handling abnormal nodes in the intelligent fault management process for service call link failures. According to this method, when a service call link failure causes a task node to malfunction, the configuration of that node can be replaced, and the currently malfunctioning process instance can be migrated to the intelligent fault management process template with the replaced abnormal task node configuration. Simultaneously, the service call anomaly will be captured, and technicians will be notified to repair the call link of the abnormal node, thereby ensuring the normal operation of the intelligent fault management process during anomaly repair and improving the overall robustness of the process.
[0093] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A low-code, flexible, and complex process customization method for intelligent fault management, characterized in that: include: Phase 1, Process Design Phase: (1) Analyze the requirements of the intelligent fault management process, and make the selection of the process features to be designed on the provided interface based on the requirements analysis results, including application scenarios, fault phenomena, and fault causes. (2) Based on the intelligent fault management process design reference example recommendation function, intelligently recommend process design reference examples to designers according to requirements, and correct the recommendation results according to the selection rate of designers and the degree of modification of the selected process reference examples, including personalized modification of process links and configuration of task nodes. (3) Deployment process; (4) Simulate and run the model to verify its feasibility. If the verification result is infeasible, return to step (2) to re-modify the process links and configure task nodes. If the verification result is feasible, execute stage two: Phase Two, Process Execution Phase: 1) Dynamically add or remove task nodes according to rules, including: The performance of a series of interconnected task nodes is evaluated using metrics including runtime, signal-to-noise ratio, harmonic mean of accuracy and recall after data processing, true positive rate, and precision. The evaluation results are recorded as a vector, resulting in a vector representing the performance of each task node, e. i The execution effect vector of a series of task nodes is The expected results to be achieved by each task node are recorded using vectors. These vectors are obtained by vectorizing the text description of the algorithm corresponding to each task node, including: When the algorithm corresponding to the task node is designed, it uses text descriptions of running time and running quality. Vectorizing the text descriptions yields the expected running effect vector of the task node. The expected overall performance of the task node is recorded as a vector P, and the distance between the two vectors E and P is used as the judgment criterion: Where, ||·||2 is the 2-norm of the vector; When d(E,P) is less than the set threshold d t When d(E,P) is greater than the set threshold d, the operation process remains unchanged; t During this process, task nodes in the execution flow are dynamically added or removed; the rule for addition and removal is: obtain the execution effect vector after the task node jumps. When the minimum value of d(E′,P) is less than the set threshold d t When the process is in progress, a task node jump operation is performed, and the jump node is the task node that minimizes d(E′,P); when the minimum value of d(E′,P) is greater than a set threshold d t When adding a task node, the process adds nodes that satisfy d(E″,P). <d t The task node is E″, which is the prediction vector of the running effect after adding the task node. When there are multiple task nodes that meet the conditions, the task node that minimizes d(E″,P) is added. 2) Handling abnormal nodes in the intelligent fault management process for service call chain failures, including: Monitor the service call chain status of the task nodes in the process. When a task node is found to be running abnormally due to a service call chain failure, pause the intelligent fault management process, change the configuration of the abnormal node to generate a new process template, and migrate the currently abnormal process running instance to the new process template. At the same time, capture the service call chain abnormality and notify the technical personnel to repair the abnormal node. Phase Three, Process Closure Phase: Data related to the process, including process execution time, node operation status, and process design status, is saved to the database module for later use.
2. The low-code, flexible, complex process customization method for intelligent fault management according to claim 1, characterized in that, The recommended function calculation method for the intelligent fault management process design reference example described in step (2) of phase one is as follows: Generate a user vector u = (x1, x2, ..., x) based on the designer's selection. n ), where x i These are the feature values related to the process to be designed; the process reference examples in the process case library have their own process vectors c = (y1, y2, ..., y...). n ), where y i The feature values of the process reference example; Find the cosine values of the user vector u and the process vector c, and obtain the similarity function between the user vector and the process vector of the intelligent fault management process reference example: When process reference examples are recommended to designers, the recommendations are adjusted based on the designers' selection rate and the extent of modifications made to the selected process reference examples. The designers' selection rate... s represents the number of times the process reference example was selected, n represents the number of times the process reference example was recommended, and the degree of modification of the process reference example. r i N represents the number of single-step modifications to the selected process reference example, where N is the total number of links and nodes in the selected process reference example; a threshold S is set for the designer's selection rate. T The threshold R for the degree of modification to the process reference example. T When S T Or R>R T When the weighting function reaches 0, the weighting function is: Finally, the recommended function for the intelligent fault management process design reference example is obtained: F = θsim(u,c).
3. The low-code, flexible, complex process customization method for intelligent fault management according to claim 2, characterized in that, The steps for designers to design an intelligent fault management process based on the recommended function in the intelligent fault management process design reference example are as follows: ① Designers select the feature information of the process to be designed on the provided page; ② Generate user vectors based on the selection; ③ Use the recommendation function of the intelligent fault management process design reference example to calculate the recommended value of the process reference example in the process case library, and recommend the process design reference example based on the calculation result; ④ Designers select appropriate process design reference examples and design processes based on them; ⑤ Based on the design results provided by the designers, revise the weighted function value of the process design reference example.
4. The low-code, flexible, complex process customization method for intelligent fault management according to claim 1, characterized in that, Step 2) in Phase Two also includes: If the designer has not designed the reconfiguration of task nodes, the internal matching mechanism will perform node configuration matching changes, including: Set feature values for the relevant attributes configured for the nodes. These attributes include the type of input data required for the corresponding service, the service runtime, and the format of the service output. Based on the set feature values, obtain the configuration feature vector A = (a1, a2, ..., a...). n ), where a i Configure characteristic values of relevant attributes for abnormal task nodes; different task node configurations available for reconfiguration in the same type of node configuration library have their own characteristic vectors B = (b1, b2, ..., b n ), where b i Configure the feature values of relevant attributes for other task nodes in the configuration library for nodes of the same type, and calculate the matching function results: Where r(A,B) represents the degree of correlation between feature vector A and feature vector B. For a i The average value, For b i The average value is used to determine the node configuration closest to 1, which is then used as the replacement configuration for the abnormal node.