System function modeling system and method

Abstract

The invention relates to the technical field of system function modeling, and relates to a system function modeling system and method, and the method comprises the steps: obtaining modeling process data of a target system; processing the modeling process data to obtain attribute vectors corresponding to the modeling process data, and respectively inputting the attribute vectors into a vector classification model to obtain attribute vector categories corresponding to the attribute vectors; acquiring a software system of the target system; performing formalized language description on the software system of the target system by using a unified form modeling language to obtain function description; representing the function description as a function description graph in a graph representation form; inputting the function description graph into a pre-constructed modular model to obtain a module name, a module definition and a module characteristic corresponding to the function description graph; a module matching mechanism is constructed through combination of multi-dimensional data and a machine learning model, and efficient modeling and module positioning of a complex system are realized through accurate mapping of a module number and a system module entry.

Description

Technical Field

[0001] The present invention relates to the technical field of system function modeling, and more specifically, to a system function modeling system and method. Background Art

[0002] As software systems become increasingly large and complex, traditional system function modeling and module management methods have gradually exposed problems such as low efficiency, poor accuracy, and insufficient scalability. Formal modeling languages ​​(such as Z language) have broad application prospects in software system analysis and modeling. Through the formal description of set theory and predicate logic, they can accurately express the state, operation and constraints of the system, providing strong theoretical support for the modular management of complex systems. However, in the prior art, the application of formal languages ​​still has some limitations. There is an urgent need for an intelligent solution that combines data acquisition, formal language modeling and machine learning technology to improve module positioning efficiency, enhance system scalability, and provide universal support for complex system management in different industries.

[0003] For example, a Chinese patent application with publication number CN103744788A discloses a feature location method based on multi-source software data analysis, which includes: step 1: searching the current software system through information retrieval technology: querying the source code of the current software system for program code related to the current modification request, and recording the program code as feature information a; step 2: mining historical evolution information through data mining technology: querying the evolution history library for historical modification requests related to the current modification request, performing a union operation on the modification elements in the related historical modification requests, and recording the output program code as feature information b; step 3: analyzing the execution trajectory through dynamic analysis technology, the execution trajectory includes marked execution information and complete execution information, performing a subtraction operation on the complete execution information and the marked execution information, and recording the output result as pending execution information; then performing static analysis on the pending execution information to obtain an information set related to the current modification request in the pending execution information, performing a union operation on the information set and the marked execution information, and recording the output program code as feature information c; step 4: performing an intersection calculation on the three types of feature information a, b, and c, and outputting a feature location result.

[0004] Although the above method can meet most scenarios, research and practical application of the above method and existing technology have found that the above method and existing technology have at least the following defects: Existing technologies focus on source code, historical evolution, and execution trajectories, without fully considering higher-level information such as requirement documents, design documents, or system function descriptions, resulting in the omission of some key contextual information. The correlation between the various feature information is not reflected through a unified standardized or formalized modeling method, resulting in information conflicts or inaccurate matching. Secondly, existing technologies rely on specific software systems and historical evolution data, which limits the application of this solution in distributed systems or heterogeneous environments. Furthermore, it still requires a lot of manual participation, the accuracy of feature positioning is not high, and it cannot adapt to the efficient modeling and module positioning of complex systems. Summary of the invention

[0005] In order to overcome the above defects of the prior art and to achieve the above objectives, the present invention provides the following technical solutions: a system function modeling method, comprising: Acquire the modeling process data of the target system, wherein the modeling process data includes basic acquisition data of the modeling object, monitoring acquisition data, analysis method and report generation content; Processing the modeling process data to obtain attribute vectors corresponding to the modeling process data, inputting the attribute vectors into the vector classification model respectively, and obtaining attribute vector categories corresponding to the attribute vectors; Obtain the software system of the target system; Using a unified formal modeling language, a software system of the target system is described in a formal language to obtain a functional description; the functional description is represented graphically as a functional description diagram; Input the function description diagram into the pre-built modular model to obtain the module name, module definition and module characteristics corresponding to the function description diagram; the module characteristics include input interface and output interface; The attribute vector category, module name, module definition and module characteristics are input into the pre-built module matching model to obtain the module matching number, and the module matching number is matched with the pre-built module number-system module entry matching table to obtain the system module entry corresponding to the current module matching number.

[0006] Furthermore, the method for obtaining the attribute vector corresponding to the modeling process data includes: S100: converting modeling process data into a modeling process data object; the modeling process data object includes a process data entity class, and the process data entity class includes process data attributes; S101: Obtain the number of entity classes in the modeling process data object, recorded as STSL; let the initial value of i be 1, and the value range of i be 1 to STSL; S102: Get the i-th entity class; record the number of attributes in the i-th entity class as SXSL; set the initial value of j to 1, and the value range of j is 1 to SXSL; S103: Obtain the jth attribute in the i-th entity class, and convert the jth attribute into an attribute vector; S104: Let j=j+1. If j is less than or equal to SXSL, continue to execute S103. If j is greater than SXSL, let i=i+1. If i is less than or equal to STSL, continue to execute S102. If i is greater than STSL, end the current process.

[0007] Furthermore, the training method of the vector classification model includes: A vector classification data set is collected in advance, wherein the vector classification data set includes G groups of attribute vectors and attribute vector categories corresponding to the G groups of attribute vectors, where G is a positive integer greater than 0; the vector classification data set is divided into a training set and a test set, the attribute vectors in the training set are used as the input of a vector classification model, and the attribute vector categories in the training set are used as the output of the vector classification model, with minimizing the sum of the prediction accuracies of all predicted attribute vector categories as the training goal; training is stopped when the sum of the prediction accuracies converges; the vector classification model is a random forest model or a support vector machine model.

[0008] Furthermore, the method for obtaining the function description includes: Before modeling, the functional requirements and design documents of the software system are collected in advance. The functional requirements include the core interaction relationship of the system, the operation process between functional modules, the constraints during system operation, performance indicators, security requirements and boundary conditions, etc.; the design documents include the interface definition, calling relationship, dependency relationship, data flow and control flow between modules, etc. Use Z language to formally describe functional requirements and design documents, define the status, operation, and constraints of the software system in mathematical language, and obtain a functional description in Z language form.

[0009] Furthermore, the training method of the modular model includes: Q groups of modular training data are collected in advance, where Q is an integer greater than 1, and the modular training data include a function description diagram, and a module name, module definition, and module characteristics corresponding to the function description diagram; each group of modular training data is converted into a feature vector, and all feature vectors are used as inputs of a modular model, the modular model uses the module name, module definition, and module characteristics corresponding to each group of modular training data as outputs, and uses the actual module name, module definition, and module characteristics corresponding to each group of modular training data as prediction targets, and uses minimizing the sum of prediction accuracies of all predicted module names, module definitions, and module characteristics as a training target; training is stopped when the sum of prediction accuracies converges; the modular model is a CNN model.

[0010] Furthermore, the training method of the module matching model includes: A module matching data set is collected in advance, wherein the module matching data set includes K groups of module matching data and module matching numbers corresponding to the K groups of module matching data, K is a positive integer greater than 0, and the module matching data includes attribute vector categories, module names, module definitions and module characteristics; the module matching data set is divided into a training set and a test set, the module matching data in the training set is used as the input of a module matching model, and the module matching numbers in the training set are used as the output of the module matching model, so as to minimize the sum of the prediction accuracies of all predicted module matching numbers as a training goal; the training is stopped when the sum of the prediction accuracies converges; the module matching model is a naive Bayes model or a support vector machine model.

[0011] Furthermore, the method of expressing the function description in a graphical representation as a function description graph includes: Import the function description into Enterprise Architect, LaTeX + TikZ or Graphviz tools to obtain the function description diagram corresponding to the function description.

[0012] Furthermore, the system function modeling method further includes: The metamodel and intermediate queue are used to automatically modularize the system functional description to obtain module definition.

[0013] Furthermore, the method for automatically modularizing the system function description using the metamodel and the intermediate queue includes: S111: Define the system functional metamodel; S112: Mapping the system function description to a function diagram, extracting each function block in the system function description as a function block object in the system function metamodel, and extracting dependencies between function blocks as relationship objects; S113: converting the functional graph into an intermediate queue; S114: extracting module seeds from the intermediate queue; S115: Recursively cluster functional blocks into module definitions based on the module seeds.

[0014] A system function modeling system, implementing the system function modeling method, the system comprising: The first acquisition module is used to obtain the modeling process data of the target system, wherein the modeling process data includes basic acquisition data of the modeling object, monitoring acquisition data, analysis method and report generation content; The first processing module is used to process the modeling process data to obtain the attribute vectors corresponding to the modeling process data, and input the attribute vectors into the vector classification model to obtain the attribute vector categories corresponding to the attribute vectors; The second acquisition module is used to obtain the software system of the target system; The second processing module uses a unified formal modeling language to describe the software system of the target system in a formal language to obtain a functional description; the functional description is represented in a graphical form as a functional description diagram; The third processing module is used to input the function description diagram into the pre-built modular model to obtain the module name, module definition and module characteristics corresponding to the function description diagram; the module characteristics include input interface and output interface; The module matching module is used to input the attribute vector category, module name, module definition and module characteristics into the pre-built module matching model to obtain the module matching number, match the module matching number with the pre-built module number-system module entry matching table, and obtain the system module entry corresponding to the current module matching number.

[0015] The technical effects and advantages of a system function modeling system and method provided by the present invention are as follows: By analyzing and processing the data of the modeling process, the corresponding attribute vector category is obtained, that is, through standardized data processing and vectorized description, the management of system modules is more efficient and organized, providing comprehensive data support for module management and function expansion. Combined with the machine learning model, the module matching data is classified and predicted, and the module matching number is automatically generated and accurately matched with the system module entry without manual intervention, greatly shortening the system function module positioning time.

[0016] Through the combination of data-driven and machine learning technologies, this solution has achieved intelligent and automated processes from data collection, processing to module matching. Its significant advantages are that it improves the efficiency of module positioning and enhances the system's scalability, while significantly reducing the cost and risk of manual analysis; in addition, through the precise mapping of module numbers and system module entrances, efficient modeling and module positioning of complex systems are achieved, providing a comprehensive and reliable solution for the efficient management of complex systems. This solution is versatile and suitable for modular modeling and management scenarios of various complex systems. The flexibility and scalability of the solution enable it to meet the needs of different industries and fields, and has a high value for promotion and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a system function modeling system in Example 1; Figure 2 A flow chart of a system function modeling method in Example 1; Figure 3 A flow chart of a method for obtaining attribute vectors corresponding to modeling process data; Figure 4 Model a system connection diagram for a system function. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Example 1

[0020] See also Figure 1 As shown, a system function modeling system described in this embodiment includes a first acquisition module, a first processing module, a second acquisition module, a second processing module, a third processing module and a module matching module, and each module realizes data transmission through wired and / or wireless connections.

[0021] The first acquisition module is used to obtain the modeling process data of the target system, and the modeling process data includes basic acquisition data of the modeling object, monitoring acquisition data, analysis method and report generation content.

[0022] It should be noted that the modeling process data is the starting point of the modeling process. Taking the capacitor quality assessment system as an example, the basic collected data includes the capacitor model, batch number and performance parameters. The performance parameters include rated voltage and nominal capacity.

[0023] The monitoring and collection data include test conditions and monitoring data corresponding to the test conditions. The test conditions are specified test frequency, ESR, voltage, ambient temperature, cumulative duration, etc.; the monitoring data include measured capacity, ESR, leakage current, etc.

[0024] It should be noted that ESR is the equivalent series resistance of a capacitor, which refers to the effective resistance value of a capacitor in an AC circuit.

[0025] The test conditions are as follows: 1kHz, room temperature 25℃, ESR: 0.08Ω, rated voltage 100V; the monitoring data are as follows: 100 hours: Measured capacitance: 9.91µF, ESR: 0.09Ω, leakage current: 6µA.

[0026] 200 hours: Measured capacitance: 9.85µF, ESR: 0.10Ω, leakage current: 7µA.

[0027] 300 hours: Measured capacitance: 9.80µF, ESR: 0.12Ω, leakage current: 10µA.

[0028] 400 hours: Measured capacitance: 9.71µF, ESR: 0.14Ω, leakage current: 15µA.

[0029] 500 hours: Measured capacitance: 9.64µF, ESR: 0.18Ω, leakage current: 20µA.

[0030] The analysis method is machine learning, that is, machine learning is used to generate capacitor quality scores for basic collected data corresponding to monitoring collected data; the report generated includes the capacitor quality scores.

[0031] The first processing module is used to process the modeling process data to obtain the attribute vectors corresponding to the modeling process data, and input the attribute vectors into the vector classification model to obtain the attribute vector categories corresponding to the attribute vectors.

[0032] like Figure 3 As shown, the method for obtaining the attribute vector corresponding to the modeling process data includes: S100: converting modeling process data into a modeling process data object; the modeling process data object includes a process data entity class, and the process data entity class includes process data attributes; S101: Obtain the number of entity classes in the modeling process data object, recorded as STSL; let the initial value of i be 1, and the value range of i be 1 to STSL; S102: Get the i-th entity class; record the number of attributes in the i-th entity class as SXSL; set the initial value of j to 1, and the value range of j is 1 to SXSL; S103: Obtain the jth attribute in the i-th entity class, and convert the jth attribute into an attribute vector; It should be noted that attributes can be converted into attribute vectors by using word embedding (such as Word2Vec, BERT, etc.). By converting attributes into attribute vectors, efficient semantic expression, automatic matching and intelligent processing can be achieved, providing strong technical support for functional modeling and module management of complex systems. This method not only improves the efficiency of demand analysis and module matching, but also enhances the system's ability to adapt to changing requirements and significantly reduces the cost of manual intervention.

[0033] S104: Let j=j+1. If j is less than or equal to SXSL, continue to execute S103. If j is greater than SXSL, let i=i+1. If i is less than or equal to STSL, continue to execute S102. If i is greater than STSL, end the current process.

[0034] For example, the modeling process data includes basic collection data, monitoring collection data, analysis method and report generation content. The basic collection data includes capacitor model, batch number, rated voltage and nominal capacity; the monitoring collection data includes the specified test frequency, ESR, voltage, ambient temperature, cumulative duration and leakage current; the analysis method is machine learning; and the report generation content includes capacitor quality score.

[0035] The basic collected data, monitoring collected data, analysis method and report generation content are converted into a modeling process data object as a whole. Among them, the basic collected data, monitoring collected data, analysis method and report generation content are all entity classes; the capacitor model, batch number, rated voltage and nominal capacity are the attributes of the basic collected data; the specified test frequency, ESR, voltage, ambient temperature, cumulative duration and leakage current are the attributes of the monitoring collected data; machine learning is the attribute of the analysis method; the capacitor quality score is the attribute of the report generation content.

[0036] The training method of the vector classification model includes: A vector classification data set is collected in advance, wherein the vector classification data set includes G groups of attribute vectors and attribute vector categories corresponding to the G groups of attribute vectors, where G is a positive integer greater than 0; the vector classification data set is divided into a training set and a test set, the attribute vectors in the training set are used as the input of a vector classification model, and the attribute vector categories in the training set are used as the output of the vector classification model, with minimizing the sum of the prediction accuracies of all predicted attribute vector categories as the training goal; training is stopped when the sum of the prediction accuracies converges; the vector classification model is a random forest model or a support vector machine model.

[0037] It should be noted that vectorizing data and classifying it through machine learning models can effectively improve the efficiency, accuracy and automation level of data processing and classification. Data vectorization and machine learning classification can automate the entire process without manually defining rules, reducing manual workload. Through the vector classification results, the system can automatically determine the attribute category of the input data and provide intelligent support for module positioning and function optimization. At the same time, this method can capture complex features, dynamically adapt to data changes, support multi-scenario and multi-task applications, and is an important technical support for data processing and decision-making in intelligent systems.

[0038] The second acquisition module is used to acquire the software system of the target system.

[0039] The second processing module uses a unified formal modeling language to formally describe the software system of the target system to obtain a functional description; and represents the functional description in a graphical form as a functional description diagram.

[0040] The method for obtaining the function description includes: Before modeling, the functional requirements and design documents of the software system are collected in advance. The functional requirements include the core interaction relationship of the system, the operation process between functional modules, the constraints during system operation, performance indicators, security requirements and boundary conditions, etc.; the design documents include the interface definition, calling relationship, dependency relationship, data flow and control flow between modules, etc. Use Z language to formally describe functional requirements and design documents, define the status, operation, and constraints of the software system in mathematical language, and obtain a functional description in Z language form.

[0041] It should be noted that the modeling language includes languages ​​such as Z language and B method. In this embodiment, the modeling language is Z language as an example. This process can be obtained by the technician through analyzing the documents and requirements, or by inputting the sorted functional description into the intelligent code model library and using the automated tools of the model library to generate the initial description code. The initial description code not only provides the basic structure of the functional module, but can also be further expanded into detailed module functions and interaction logic, laying a solid foundation for efficient modeling and functional optimization of complex systems. In addition, through the unified description of the formal language, the logical consistency of the system requirements and the design documents can be enhanced, the interfaces and dependencies between modules can be clarified, and the uncertainty in the module division and function realization process can be reduced. This systematic description method provides key support for subsequent modular modeling, functional verification and intelligent module matching.

[0042] Methods for representing the functional description in a graphical representation as a functional description diagram include: Import the function description into Enterprise Architect (EA), LaTeX + TikZ or Graphviz tools to obtain the function description diagram corresponding to the function description.

[0043] It should be noted that Enterprise Architect is a tool for software and system modeling. It can be used for system architecture design. In system function modeling, it can display function descriptions in an intuitive graphical way. LaTeX is a typesetting system based on TeX. TikZ is a drawing macro package of LaTeX. It can draw accurate and beautiful graphics by writing codes. In system function modeling, it can flexibly customize the style and layout of function description diagrams. Graphviz is a tool for drawing graph structures. In system function modeling, it can quickly convert function descriptions into graphical function description diagrams.

[0044] The third processing module is used to input the function description diagram into the pre-built modular model to obtain the module name, module definition and module characteristics corresponding to the function description diagram; the module characteristics include input interface and output interface.

[0045] The training method of the modular model includes: Q groups of modular training data are collected in advance, where Q is an integer greater than 1, and the modular training data include a function description diagram, and a module name, module definition, and module characteristics corresponding to the function description diagram; each group of modular training data is converted into a feature vector, and all feature vectors are used as inputs of a modular model, the modular model uses the module name, module definition, and module characteristics corresponding to each group of modular training data as outputs, and uses the actual module name, module definition, and module characteristics corresponding to each group of modular training data as prediction targets, and uses minimizing the sum of prediction accuracies of all predicted module names, module definitions, and module characteristics as a training target; training is stopped when the sum of prediction accuracies converges; the modular model is a CNN model.

[0046] The module matching module is used to input the attribute vector category, module name, module definition and module characteristics into the pre-built module matching model to obtain the module matching number, match the module matching number with the pre-built module number-system module entry matching table, and obtain the system module entry corresponding to the current module matching number.

[0047] The training method of the module matching model includes: A module matching data set is collected in advance, wherein the module matching data set includes K groups of module matching data and module matching numbers corresponding to the K groups of module matching data, K is a positive integer greater than 0, and the module matching data includes attribute vector categories, module names, module definitions and module characteristics; the module matching data set is divided into a training set and a test set, the module matching data in the training set is used as the input of a module matching model, and the module matching numbers in the training set are used as the output of the module matching model, so as to minimize the sum of the prediction accuracies of all predicted module matching numbers as a training goal; the training is stopped when the sum of the prediction accuracies converges; the module matching model is a naive Bayes model or a support vector machine model.

[0048] The module number-system module entry matching table (capacitor quality evaluation system) is shown in Table 1: Table 1 Module number-system module entry matching table

[0049] It should be noted that, for example, if the module matching number obtained through the machine learning model is 6, the system module entry is " / src / monitoring / trend-analysis", and the corresponding function description is "performance change trend analysis module"; for example, if the module matching number obtained through the machine learning model is 7, the system module entry is " / src / system / logging / error-handler", and the corresponding function description is "error logging module"; for example, if the module matching number obtained through the machine learning model is 8, the system module entry is " / src / ui / report / display", and the corresponding function description is "interface display module for quality assessment results".

[0050] Through the precise mapping of module numbers and system module entries, efficient modeling and module positioning of complex systems are achieved, providing a comprehensive and reliable solution for the efficient management of complex systems. It has high versatility and is suitable for modular modeling and management scenarios of various complex systems.

[0051] Example 2 This embodiment further improves the design on the basis of the embodiment 1. The difference is that in the actual operation of the embodiment 1, the coupling degree between the modular model and the module definition is relatively high, and the modular abstraction level needs to be improved so that the model can adapt to different definitions. If the conversion between the system function description and the module is too complicated, manual intervention may be required. Automation should be considered, and the metamodel and queue conversion can be used to improve the process consistency. Based on this, a system function modeling system also includes: Automatically define modules, which are used to automatically modularize the system function description using metamodels and intermediate queues to obtain module definitions; The module definition output by the modular model in Example 1 is directly used as a control group to verify the accuracy of automatic modularization, but it is no longer used in system function modeling based on formal language.

[0052] The method of automatically modularizing the system function description using the metamodel and the intermediate queue to obtain the module definition includes: S111: Define the system functional metamodel; The system functional metamodel includes: Function block: name, input and output properties; Relationship: source function block, target function block and relationship type; Function block expression ; Represents each functional block in the functional diagram; Relational Expressions ; In the formula, is the relation source function block; It is the relation target function block; is the relationship type; S112: Mapping the system function description to a function diagram; Extract each function block in the system function description as a function block object in the system function metamodel; Extract dependencies between functional blocks as relationship objects; S113: converting the functional graph into an intermediate queue; For example: Function i: ; relation:<Relj.source.name> ,<Relj.target.name> ,<Relj.type> ; e functions constitute a functional block part; the functional block part is represented as .

[0053] It should be noted that Indicates the name of the i-th function block in the function diagram, which is used to uniquely identify the function block and facilitate the distinction and management of each function block in the process of system function modeling.<Relj.source.name> Indicates the source function block name of the jth relation,<Relj.target.name> Indicates the target function block name of the jth relation,<Relj.type> Indicates the type of the jth relationship, which is used to describe the nature of the association between the source function block and the target function block.

[0054] S114: extracting module seeds from the intermediate queue; Find the function block whose input and output are both 0 as the seed; That is, M={function FB|in=out=0}; S115: recursively clustering functional blocks into module definitions according to the module seeds; like Relationship with seed Mi, Add to Mi and remove the relationship, repeat this process to expand the scope of Mi;

[0055] Example 3 like Figure 4 As shown, a schematic diagram of system function modeling system connection, the system includes: Receiving unit: receiving system software code; System description unit: establish a unified formal modeling language system library, describe the system in a formal language, and obtain the system function description; express the system function description in a graphical form as a system function description diagram; Module differentiation unit: inputs the system function description diagram into the pre-built modular model and outputs the system module; Module definition unit: Use metamodel and intermediate queue to automatically modularize the system function description and obtain module definition; Independent modeling unit: input module definition into the formal modeling language system library, match the formal language corresponding to the module definition, use the formal language to describe the module, and obtain the module description information; Integration unit: integrates the description information of the modules to obtain an integrated module; System modeling unit: Use visual modeling tools to model the integrated modules and obtain the system function model; Verification unit: Perform formal verification on the system functional model and generate a verification report; Optimization backtracking unit: Perform deployment tests on the cloud platform, collect operation data, input operation data and verification reports into the pre-built implicit optimization model, output analysis results, and optimize and adjust the modular model based on the analysis results.

[0056] Example 4 like Figure 2 As shown, the present application also provides a system function modeling method, the method comprising: Acquire the modeling process data of the target system, wherein the modeling process data includes basic acquisition data of the modeling object, monitoring acquisition data, analysis method and report generation content; Processing the modeling process data to obtain attribute vectors corresponding to the modeling process data, inputting the attribute vectors into the vector classification model respectively, and obtaining attribute vector categories corresponding to the attribute vectors; Obtain the software system of the target system; Using a unified formal modeling language, a software system of the target system is described in a formal language to obtain a functional description; the functional description is represented graphically as a functional description diagram; Input the function description diagram into the pre-built modular model to obtain the module name, module definition and module characteristics corresponding to the function description diagram; the module characteristics include input interface and output interface; The attribute vector category, module name, module definition and module characteristics are input into the pre-built module matching model to obtain the module matching number, and the module matching number is matched with the pre-built module number-system module entry matching table to obtain the system module entry corresponding to the current module matching number.

[0057] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0058] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A system function modeling method, characterized in that: Methods include: Acquire the modeling process data of the target system, wherein the modeling process data includes basic acquisition data of the modeling object, monitoring acquisition data, analysis method and report generation content; Processing the modeling process data to obtain attribute vectors corresponding to the modeling process data, inputting the attribute vectors into the vector classification model respectively, and obtaining attribute vector categories corresponding to the attribute vectors; Obtain the software system of the target system; Using a unified formal modeling language, a software system of the target system is described in a formal language to obtain a functional description; the functional description is represented in a graphical form as a functional description diagram; Input the function description diagram into the pre-built modular model to obtain the module name, module definition and module characteristics corresponding to the function description diagram; the module characteristics include input interface and output interface; The attribute vector category, module name, module definition and module characteristics are input into the pre-built module matching model to obtain the module matching number, and the module matching number is matched with the pre-built module number-system module entry matching table to obtain the system module entry corresponding to the current module matching number.

2. A system function modeling method according to claim 1, characterized in that: The method for obtaining the attribute vector corresponding to the modeling process data includes: S100: converting modeling process data into a modeling process data object; the modeling process data object includes a process data entity class, and the process data entity class includes process data attributes; S101: Obtain the number of entity classes in the modeling process data object, recorded as STSL; let the initial value of i be 1, and the value range of i be 1 to STSL; S102: Get the i-th entity class; record the number of attributes in the i-th entity class as SXSL; set the initial value of j to 1, and the value range of j is 1 to SXSL; S103: Obtain the jth attribute in the i-th entity class, and convert the jth attribute into an attribute vector; S104: Let j=j+1. If j is less than or equal to SXSL, continue to execute S103. If j is greater than SXSL, let i=i+1. If i is less than or equal to STSL, continue to execute S102. If i is greater than STSL, end the current process.

3. A system function modeling method according to claim 2, characterized in that: The training method of the vector classification model includes: A vector classification data set is collected in advance, wherein the vector classification data set includes G groups of attribute vectors and attribute vector categories corresponding to the G groups of attribute vectors, where G is a positive integer greater than 0; the vector classification data set is divided into a training set and a test set, the attribute vectors in the training set are used as the input of a vector classification model, and the attribute vector categories in the training set are used as the output of the vector classification model, with minimizing the sum of the prediction accuracies of all predicted attribute vector categories as the training goal; training is stopped when the sum of the prediction accuracies converges; the vector classification model is a random forest model or a support vector machine model.

4. A system function modeling method according to claim 1, characterized in that: The method for obtaining the function description includes: Before modeling, the functional requirements and design documents of the software system are collected in advance. The functional requirements include the core interaction relationship of the system, the operation process between functional modules, the constraints during system operation, performance indicators, security requirements and boundary conditions; the design documents include the interface definition, calling relationship, dependency relationship, data flow and control flow between modules; Use Z language to formally describe functional requirements and design documents, define the status, operation, and constraints of the software system in mathematical language, and obtain a functional description in Z language form.

5. A system function modeling method according to claim 1, characterized in that: The training method of the modular model includes: Q groups of modular training data are collected in advance, where Q is an integer greater than 1, and the modular training data include a function description diagram, and a module name, module definition, and module characteristics corresponding to the function description diagram; each group of modular training data is converted into a feature vector, and all feature vectors are used as inputs of a modular model, the modular model uses the module name, module definition, and module characteristics corresponding to each group of modular training data as outputs, and uses the actual module name, module definition, and module characteristics corresponding to each group of modular training data as prediction targets, and uses minimizing the sum of prediction accuracies of all predicted module names, module definitions, and module characteristics as a training target; training is stopped when the sum of prediction accuracies converges; the modular model is a CNN model.

6. A system function modeling method according to claim 1, characterized in that: The training method of the module matching model includes: A module matching data set is collected in advance, wherein the module matching data set includes K groups of module matching data and module matching numbers corresponding to the K groups of module matching data, K is a positive integer greater than 0, and the module matching data includes attribute vector categories, module names, module definitions and module characteristics; the module matching data set is divided into a training set and a test set, the module matching data in the training set is used as the input of a module matching model, and the module matching numbers in the training set are used as the output of the module matching model, so as to minimize the sum of the prediction accuracies of all predicted module matching numbers as a training goal; the training is stopped when the sum of the prediction accuracies converges; the module matching model is a naive Bayes model or a support vector machine model.

7. A system function modeling method according to claim 4, characterized in that: Methods for representing the functional description in a graphical representation as a functional description diagram include: Import the function description into Enterprise Architect, LaTeX + TikZ or Graphviz tools to obtain the function description diagram corresponding to the function description.

8. A system function modeling method according to claim 1, characterized in that: Also includes: The metamodel and intermediate queue are used to automatically modularize the system functional description to obtain module definition.

9. A system function modeling method according to claim 8, characterized in that: The method for automatically modularizing system function description using a metamodel and an intermediate queue comprises: S111: Define the system functional metamodel; S112: Mapping the system function description to a function diagram, extracting each function block in the system function description as a function block object in the system function metamodel, and extracting dependencies between function blocks as relationship objects; S113: converting the functional graph into an intermediate queue; S114: extracting module seeds from the intermediate queue; S115: Recursively cluster functional blocks into module definitions based on the module seeds.

10. A system function modeling system, implementing a system function modeling method according to any one of claims 1 to 9, characterized in that: The system includes: The first acquisition module is used to obtain the modeling process data of the target system, wherein the modeling process data includes basic acquisition data of the modeling object, monitoring acquisition data, analysis method and report generation content; The first processing module is used to process the modeling process data to obtain the attribute vectors corresponding to the modeling process data, and input the attribute vectors into the vector classification model to obtain the attribute vector categories corresponding to the attribute vectors; The second acquisition module is used to obtain the software system of the target system; The second processing module uses a unified formal modeling language to describe the software system of the target system in a formal language to obtain a functional description; the functional description is represented in a graphical form as a functional description diagram; The third processing module is used to input the function description diagram into the pre-built modular model to obtain the module name, module definition and module characteristics corresponding to the function description diagram; the module characteristics include input interface and output interface; The module matching module is used to input the attribute vector category, module name, module definition and module characteristics into the pre-built module matching model to obtain the module matching number, match the module matching number with the pre-built module number-system module entry matching table, and obtain the system module entry corresponding to the current module matching number.

Images