Low-code management system and method based on AIot Internet of Things

Abstract

The invention discloses a low-code management system and method based on the AIot Internet of Things, and relates to the technical field of code management, the system comprises a template learning optimization module, a component dynamic adaptation module, a demand conversion module, a code annotation generation module and a code compatibility conversion module; the template learning optimization module is used for automatically extracting optimization points and updating template settings by analyzing historical cases and user data so as to improve the applicability of the template in practical application; according to the invention, through the template learning optimization module, deep analysis of historical cases and user data is realized, optimization points are automatically extracted, template setting is updated, the applicability and development efficiency of the template in practical application are remarkably improved, and through the automatic template updating process, manual intervention is reduced, the development cost is reduced, and the development efficiency is improved. And meanwhile, the stability and compatibility of the system are kept.

Description

Technical Field

[0001] The present invention relates to the field of code management technology, and specifically to a low-code management system and method based on AIot Internet of Things. Background Art

[0002] With the rapid development of Internet of Things (IoT) technology, more and more devices are connected to the Internet, enabling data collection, exchange and analysis. This connection not only includes traditional computers and smartphones, but also extends to various sensors, smart home appliances, industrial equipment, etc. The application scenarios of IoT are becoming increasingly extensive, ranging from smart homes and smart cities to industrial automation and agricultural monitoring. However, with the diversification of IoT devices and the complexity of application scenarios, traditional software development methods face many challenges;

[0003] Driven by the Internet of Things, the diversity of devices and the complexity of application scenarios have challenged traditional software development methods. Frequent updates and diversification of devices require management systems to quickly adapt to new devices, while traditional manual coding methods cannot meet the needs of rapid iteration and deployment.

[0004] In order to solve the above defects, a technical solution is now provided. Summary of the invention

[0005] The purpose of this invention is to solve the problems of management complexity caused by the diversity and frequent updates of IoT devices, the inefficiency and high cost of traditional development methods, and the maintainability and security challenges of IoT system code, and to propose a low-code management system and method based on AIot IoT.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A low-code management system based on AIot Internet of Things, including:

[0008] Template learning and optimization module, which is used to automatically refine optimization points and update template settings by analyzing historical cases and user data to improve the applicability of templates in practical applications;

[0009] The component dynamic adaptation module is used to identify the information of newly connected devices and automatically adjust the component configuration, and adapt the component functions in real time according to the changes of devices and services during application operation;

[0010] The demand conversion module uses natural language processing technology to parse user demand text and convert it into low-code development operation instructions and logical framework for rapid development;

[0011] The code comment generation module is used to automatically add comments to the generated code according to the parsed requirements, improve the code readability and maintainability, and facilitate subsequent development and maintenance work;

[0012] The code compatibility conversion module is used to identify low-code files in different formats and convert them into system-compatible formats, while also performing security vulnerability repairs and compatibility optimization processing.

[0013] Furthermore, the execution process of the template learning optimization module is as follows:

[0014] Data collection phase: Collect IoT application case data that has been successfully developed and run from the system database. At the same time, collect user operation records during the use of the system, including template selection preferences, component usage frequency, and operation information for modifying templates.

[0015] Classify and organize the collected data according to application type, functional module, and device type;

[0016] Analysis and extraction stage: Use data analysis algorithms to analyze the functional structure of various cases, identify the composition of core functional modules in different applications and the logical relationship between them;

[0017] By mining user operation records, we can analyze the patterns of users adjusting templates when facing different business needs.

[0018] Combined with the functional structure analysis and user behavior pattern mining results, optimization points to improve the applicability of the template are extracted, including the adjustment of the template's default functions, the optimization of the interface layout, and the improvement of the initial parameter settings;

[0019] Template update phase: Based on the extracted optimization points, several template update plans are formulated. The plans include the specific content and steps for adjusting various parts of the template to ensure that the updated template meets the functional optimization requirements while maintaining the stability and compatibility of the system;

[0020] Test several template update plans, determine the priority of the template update plans, and select the template update plan with the highest priority;

[0021] The template is updated according to the highest priority template update plan. During the update process, the template is tested and verified to ensure that the new template can meet expectations in terms of functionality and will not introduce new errors or problems. After the update is completed, the new template is stored in the template library for users to use in subsequent development.

[0022] Furthermore, the specific operation steps of the template learning optimization module to determine the priority of the template update scheme are as follows:

[0023] Comprehensive analysis is performed by collecting evaluation parameters, wherein the evaluation parameters include: function prompt amplitude, degree of improvement in user experience and development cost, and the improvement value, body modification ratio and segmentation value are respectively analyzed through the evaluation parameters, and the obtained improvement value, body modification ratio and segmentation value are calibrated as tc, gt and cj, and after normalization, the following formula is entered: ZPA=tc×ο1+gt×ο2+cj×ο3 to obtain the comprehensive judgment value ZPA, wherein ο1, ο2, ο3 are the preset weight coefficients of the improvement value, body modification ratio and segmentation value respectively;

[0024] The different comprehensive judgment values ​​obtained from the formulated several template update schemes are sorted according to size, and the template update scheme with the largest comprehensive judgment value is selected as the template update scheme with the highest priority.

[0025] Furthermore, the template learning optimization module analyzes the evaluation parameters to obtain the improvement value, the body improvement ratio and the segmentation value in the following specific operation steps:

[0026] Function prompt range: including the number of new functions added, the degree of business process optimization and the improvement of data processing efficiency. The number of new functions added is calculated by counting the number of new functional modules added in each update plan; the degree of business process optimization is quantified by evaluating the degree of optimization of the update plan on the existing business process, specifically by calculating the process simplification ratio and recording it as the simplification ratio; the improvement of data processing efficiency is determined by measuring the improvement of data processing speed after the template is updated, specifically by comparing the time required to process the same amount of data before and after the update, and calculating the time difference; the number of new functions added, the simplification ratio and the time difference are normalized, and the number of new functions added, the simplification ratio and the time difference are used as the sides of the triangle to establish the bottom triangle, and then the correction coefficient is used as the height to establish a triangular prism model, and the surface area of ​​the triangular prism model is calculated and recorded as the improvement value, and this improvement value is used as the standard for measuring the degree of function improvement;

[0027] Improvement in user experience and ease of operation: Through user testing or simulated operations, the average task time of the operation steps required to complete a standard task is counted and recorded as the completion time; Improvement in user friendliness: By inviting users to rate the updated template interface, the updated score difference is calculated; Accuracy of error prompts: The error ratio is calculated by combining the number of errors encountered by users during the operation after the update and all the operation steps; After normalizing the obtained completion time, score difference and number of errors, the ratio of the score difference to the sum of the completion time and the number of errors is calculated and recorded as the system improvement ratio, and this system improvement ratio is used to measure the improvement in user experience;

[0028] Development cost, development time reduction ratio: estimate the reduction ratio of the time required to develop new applications after the update plan is implemented compared with the old template; labor cost savings: calculate the labor cost saved in the development and maintenance process after the update plan is implemented, and quantify it by comparing the reduction in the number of developers and working hours before and after the update. The obtained reduction ratio, reduction in the number of developers and reduction in working hours are normalized and summed to obtain the cost value, which is used as the standard for measuring development cost savings.

[0029] Furthermore, the component dynamic adaptation module operates as follows:

[0030] When a new device attempts to access the system, a connection is established with the device through the communication protocol of the device access layer to obtain basic information about the device, including device type, manufacturer information, device model, and communication interface parameters. At the same time, a detection command is initiated to obtain the functional parameters of the device, including data collection accuracy, supported instruction sets, and data transmission frequency range information;

[0031] Perform feature analysis on the collected device information and match it with the existing device feature models in the component library. If similar models exist, perform adaptation planning based on the existing adaptation solutions and combined with the parameters of the current device.

[0032] According to the adaptation planning results, select appropriate components from the component library or adjust the parameter configuration of existing components, including: for data acquisition components, adjust the data receiving buffer size and parsing algorithm parameters according to the data format and transmission frequency of the device; for display components, adjust the interface layout, font size and chart style according to the screen resolution and display characteristics of the device;

[0033] During the application operation, the device status changes, business rule adjustments and data traffic fluctuations are continuously monitored. Specifically, by setting monitoring points in the system, the device operation data, business logic execution status and network transmission data volume information can be obtained in real time;

[0034] When changes are detected, analyze the impact of the changes on component functions; if new data processing logic is added to the business rules, determine whether new components need to be added or the connection relationship between existing components needs to be adjusted;

[0035] Based on the adaptation decision results, components are adjusted in real time, including dynamically loading new components, modifying component operating parameters, and re-establishing communication connections between components to ensure that the system can still operate stably and provide accurate functional services when equipment and business changes.

[0036] Furthermore, the determination process of similar models in the component dynamic adaptation module is as follows:

[0037] When analyzing the characteristics of the collected device information and matching it with the existing device feature models in the component library, the device feature vector The feature vector corresponding to each device feature model in the component library The cosine similarity is calculated using the following formula: According to the preset similarity threshold τ, when the cosine similarity cos(θ) is greater than or equal to the preset similarity threshold τ, it is determined that a similar model exists;

[0038] When no similar model is found, the new device adaptation rule generation mechanism is started, and the device information is classified and summarized using the machine learning algorithm to generate a preliminary adaptation strategy; specifically, the device information is classified through a classification algorithm based on a neural network, including a multi-layer perceptron, to determine which type of device feature model in the component library is most matched. The number of input layer nodes is preset to n, the number of hidden layer nodes is h, and the number of output layer nodes is m. For the input device information feature vector The calculation after the hidden layer is: in is the weight matrix from the input layer to the hidden layer, is the bias vector of the hidden layer, f is the activation function, and the calculation of the output layer is In the formula is the weight matrix from the hidden layer to the output layer, is the bias vector of the output layer, g is the activation function of the output layer; according to the result of the output layer The category with the highest probability is selected as the matching device feature model category.

[0039] Furthermore, the execution steps of the demand conversion module include:

[0040] B1. Natural language preprocessing:

[0041] Perform word segmentation on the natural language text input by the user, split the sentence into individual words or phrases, and mark the part of speech for each word, and mark the corresponding part of speech; filter out meaningless stop words in the text, including function words and common words irrelevant to the needs;

[0042] B2. Semantic understanding and analysis:

[0043] Identify key entities in the text, including device type, operation object, and action. Use pre-trained entity recognition models, combined with domain dictionaries and context information, to extract entities and their attribute information. Analyze the semantic relationship between entities to determine the logical structure of requirements.

[0044] B3. Operation instruction generation:

[0045] Based on the results of semantic understanding, search for matching functional components in the component library; generate operational instructions for calling these components, and build the connection and interaction logic framework between components according to the demand logic, determine the data flow and trigger conditions, and form preliminary low-code development operational instructions and logic framework for subsequent development.

[0046] Furthermore, the execution steps of the code comment generation module are as follows:

[0047] Obtain the parsed structured demand data from the demand conversion module and extract key information, including business functions, operation processes, data sources and flows, and the equipment and components involved; classify and organize this information for subsequent association with different parts of the code;

[0048] According to the extracted key information of the requirements, plan the structure and content of the comments, and determine the types of comments that need to be added at the beginning of the code, function definition, key logic nodes, etc. For example, add an overall function overview comment at the beginning of the code; add function input and output parameters and function description comments at the function definition; add judgment condition meaning comments at the conditional judgment statement;

[0049] Generate accurate and clear comment text according to the planned comment content, combined with the grammatical structure and format specifications of the code. Use the code parsing tool to locate the code line or code block to which comments need to be added, and insert the comment text accurately. For variable definitions, the comments explain the purpose and value range of the variables; for code logic blocks, the comments explain the business logic and purpose they implement;

[0050] Optimize the format of the generated comments to make them conform to the unified comment style specifications of the team or project, check the completeness and accuracy of the comments, ensure the consistency between the comments and the code, and avoid the situation where the comments are inconsistent with the actual code functions. At the same time, unify and standardize the terms and expressions in the comments to improve the readability of the comments. The comment consistency checking process is as follows:

[0051] Suppose the set of code statements is S = {s1, s2, ..., s p}, the corresponding annotation set is A = {a1, a2, ..., a p}; p is the number of code statements, that is, the number of elements in the code statement set S; for each code statement s1 and its corresponding annotation a1, calculate the consistency index C(s i , a i ), using the semantic similarity calculation method, the code statements and comment texts are converted into semantic vectors respectively. Let the code statement s i The semantic vector of Note a i The semantic vector of Where i is the index variable;

[0052] The consistency index in

[0053] Where q is the vector dimension; j is another index variable;

[0054] If the consistency index C(s i , a i ) is lower than the set threshold, the code statement and comment are marked as inconsistent and need to be corrected.

[0055] Furthermore, the execution steps of the code compatibility conversion module are as follows:

[0056] Read the low-code file to be converted, analyze the file's header information, file structure characteristics, and code syntax characteristics, and identify the file's original format type by comparing it with various known low-code format standards;

[0057] For the identified format, start the corresponding syntax parser, decompose the original code into syntax units, establish a mapping relationship from the original syntax unit to the target syntax unit according to the syntax rules and data structure of this system, and perform preliminary code conversion;

[0058] During the conversion process, use security scanning tools to perform static analysis on the converted code to check whether there are common security vulnerabilities in the code. Any vulnerabilities found will be marked and recorded.

[0059] According to the vulnerability tag information, combined with the security vulnerability repair library and best practices, the code vulnerabilities are repaired, and compatibility adjustments are made for the parts of the code that are incompatible with this system, including modifying the calling parameters of the components in the original code according to the specifications of this system;

[0060] Perform functional and compatibility tests on the repaired and optimized code, simulate different operating scenarios, ensure that the converted code can run correctly in this system and work with other modules and components of the system. If problems are found during testing, return to the previous steps to correct them until they pass the test verification.

[0061] A low-code management method based on AIot Internet of Things, including the following steps:

[0062] S1. Template learning and optimization module: First, data is collected, covering past IoT application cases and user operation records. After classification and organization, it enters the analysis and refinement stage. In this stage, the case structure and user behavior are analyzed through algorithms to extract optimization points that can improve the applicability of the template. Then, multiple update plans are formulated based on these optimization points, and the priority of each plan is comprehensively evaluated. Finally, the best plan is selected to update the template to ensure that the template meets the system requirements in terms of function and compatibility;

[0063] S2. Dynamic adaptation of components: When a new device is connected, the device information is first obtained and adapted to the component library model to determine the initial configuration of the component. During the operation of the system, the changes in the device and business are continuously monitored. Once a change is detected, its impact on the component function is immediately analyzed, and the parameters and connection relationships of the component are dynamically adjusted to ensure the stable operation and functional accuracy of the system;

[0064] S3. Requirements conversion: For the requirements text input by the user, natural language preprocessing is first performed, including word segmentation, part-of-speech tagging and stop word filtering, followed by the semantic understanding and analysis stage to identify key entities and semantic relationships, build a logical framework for the requirements, and finally search for matching components in the component library based on the analysis results to generate corresponding operation instructions and logical frameworks;

[0065] S4. Code comment generation: After obtaining information from step S3, extract the key points of the requirements and classify them to plan the structure and content of the comments. Then generate the comment text according to the plan and insert it into the corresponding code position. After inserting the comments, optimize and check the format of the comments to ensure the completeness, accuracy and consistency of the comments with the code.

[0066] S5. Code compatibility conversion: initially read the low-code file, identify the original format by analyzing the file's header, structure, and syntax features, and then start the syntax parser for the format to perform code conversion. During the conversion process, use security scanning tools to check and mark vulnerabilities. After fixing the vulnerabilities and adjusting compatibility based on the marked information, perform functionality and compatibility tests on the converted code, simulate various operating scenarios, and if any problems occur, return to the corresponding steps to make corrections until the code passes the test.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] (1) The present invention, through the template learning optimization module, realizes in-depth analysis of historical cases and user data, automatically extracts optimization points and updates template settings, significantly improving the applicability and development efficiency of templates in practical applications. Through the automated template update process, manual intervention is reduced, development costs are reduced, and the stability and compatibility of the system are maintained.

[0069] (2) The present invention uses a component dynamic adaptation module to intelligently identify new access device information and automatically adjust component configuration to adapt to changes in devices and services. This real-time component adaptation function ensures that the system can still run stably and provide accurate functional services when devices and services change, thereby enhancing the flexibility and scalability of the system.

[0070] (3) In the present invention, the demand conversion module uses natural language processing technology to convert user demand text into low-code development operation instructions and logical frameworks, which greatly speeds up the development speed. The code comment generation module improves the readability and maintainability of the code, and the code compatibility conversion module ensures that low-code files of different formats can run seamlessly in this system, enhancing the compatibility and security of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0072] Figure 1 This is the overall system block diagram of the present invention. DETAILED DESCRIPTION

[0073] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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.

[0074] It should be understood that the terms "include" and "comprising" used in the specification and claims of the present disclosure indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0075] It should also be understood that the terms used in this disclosure are only for the purpose of describing specific embodiments and are not intended to limit the disclosure. As used in this disclosure and claims, the singular forms of "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in this disclosure and claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations.

[0076] like Figure 1As shown, a low-code management system based on AIot Internet of Things includes a template learning optimization module, a component dynamic adaptation module, a demand conversion module, a code comment generation module and a code compatibility conversion module;

[0077] The template learning optimization module is used to automatically refine optimization points and update template settings by analyzing historical cases and user data to improve the applicability of templates in practical applications;

[0078] Data collection phase: Collect IoT application case data that have been successfully developed and run in the past from the system database. These data cover IoT projects involved in various industry fields such as smart home, industrial manufacturing, and agricultural monitoring. At the same time, collect user operation records during the use of the system, including template selection preferences, component usage frequency, and operation details for modifying templates; classify and organize the collected data according to application type, functional module, and device type. For example, classify all application cases involving temperature sensor management into one category, and classify cases related to smart lighting control into another category, so as to facilitate subsequent targeted analysis;

[0079] Analysis and refinement stage: Use data analysis algorithms to analyze the functional structure of various cases, identify the composition of core functional modules in different applications and the logical relationship between them. For example, in smart home management applications, analyze how functional modules such as device control and scene linkage work together, and find out the functional combination patterns that appear frequently and have good effects; through in-depth mining of user operation records, find out the rules for users to adjust templates when facing different business needs. For example, count what template settings users usually adjust first when facing multi-device management scenarios, whether it is device grouping rules or data display methods; combine the results of functional structure analysis and user behavior pattern mining to extract optimization points that can improve the applicability of templates. These optimization points include adjustments to template default functions, optimization of interface layout, and improvement of initial parameter settings. For example, if it is found that most users always need to manually add specific data analysis charts when using a certain type of device monitoring template, then you can consider adding the chart as the default component of the template;

[0080] Template update phase: Based on the extracted optimization points, several template update plans are formulated. The plans include the specific content and steps for adjusting various parts of the template to ensure that the updated template meets the functional optimization requirements and maintains the stability and compatibility of the system; the formulated template update plans are tested to determine the priority of the template update plans, and comprehensive analysis is performed by collecting evaluation parameters, where the evaluation parameters include:

[0081] Function prompt range: including the number of new functions added, the degree of business process optimization and the improvement of data processing efficiency. The number of new functions added is calculated by counting the number of new functional modules added in each update plan; the degree of business process optimization is quantified by evaluating the degree of optimization of the update plan for the existing business process, specifically by calculating the process simplification ratio and recording it as the simplification ratio; the improvement of data processing efficiency is determined by measuring the improvement in data processing speed after the template is updated, specifically by comparing the time required to process the same amount of data before and after the update, and calculating the time difference; the number of new functions added, the simplification ratio and the time difference are normalized, and the number of new functions added, the simplification ratio and the time difference are used as the sides of the triangle to establish the bottom triangle, and the correction coefficient is used as the height to establish a triangular prism model, and the surface area of ​​the triangular prism model is calculated and recorded as the improvement value, which is used as the standard for measuring the degree of function improvement; the degree of improvement in user experience and the improvement in ease of operation: through user testing or simulated operations, the number of operating steps required to complete standard tasks is counted , and record it as the completion time; interface friendliness improvement: by inviting users to rate the updated template interface, calculate the updated score difference; error prompt accuracy: by the number of errors encountered by users during the operation after the update, combined with all operation steps, calculate the error ratio; after normalizing the obtained completion time, score difference and number of errors, calculate the ratio of the score difference to the sum of the completion time and the number of errors, record it as the system change ratio, and use this system change ratio as a measure of the degree of improvement in user experience; development cost, development time reduction ratio: estimate the reduction ratio of the time required to develop a new application compared with the old template after the implementation of the update plan; labor cost savings: calculate the labor cost saved in the development and maintenance process after the implementation of the update plan, specifically by comparing the reduction in the number of developers and working time before and after the update to quantify it, normalize the obtained reduction ratio, the reduction in the number of developers and the reduction in working time, and sum them to obtain the node value, and use this node value as a standard for measuring development cost savings;

[0082] The obtained improvement value, body modification ratio and node value are calibrated as tc, gt and cj respectively, and after normalization, they are entered into the following formula: ZPA=tc×ο1+gt×ο2+cj×ο3 to obtain the comprehensive judgment value ZPA, where ο1, ο2, ο3 are the preset weight coefficients of the improvement value, body modification ratio and node value respectively; the different comprehensive judgment values ​​ZPA obtained from the formulated several template update schemes are sorted according to size, and the template update scheme with the largest comprehensive judgment value is selected as the template update scheme with the highest priority.

[0083] The template is updated according to the highest priority template update plan. During the update process, the template is tested and verified to ensure that the new template can meet expectations in terms of functionality and will not introduce new errors or problems. After the update is completed, the new template is stored in the template library for users to use in subsequent development.

[0084] The component dynamic adaptation module is used to identify the information of newly connected devices and automatically adjust the component configuration, and adapt the component functions in real time according to the changes of devices and services during application operation;

[0085] When a new device attempts to access the system, a connection is established with the device through the communication protocol of the device access layer to obtain basic information about the device, including device type, manufacturer information, device model, and communication interface parameters. At the same time, a deep detection command is initiated to obtain the functional parameters of the device, including data collection accuracy, supported instruction sets, and data transmission frequency range information. The collected device information is analyzed for features and matched with the existing device feature models in the component library. If a similar model exists, the existing adaptation plan is referenced and combined with the unique parameters of the current device for adaptation planning. The judgment process of similar models is as follows:

[0086] When analyzing the characteristics of the collected device information and matching it with the existing device feature models in the component library, the device feature vector (including device type, function parameters, etc.) and the feature vector corresponding to each device feature model in the component library The cosine similarity is calculated using the following formula: According to the preset similarity threshold τ, when the cosine similarity cos(θ) is greater than or equal to the preset similarity threshold τ, it is determined that a similar model exists;

[0087] When no similar model is found, a new device adaptation rule generation mechanism is started, and a machine learning algorithm is used to classify and summarize the device information to generate a preliminary adaptation strategy. Specifically, a classification algorithm based on a neural network, including a multi-layer perceptron, is used to classify the device information to determine which type of device feature model in the component library is the most matched. The number of input layer nodes is preset to be n (corresponding to the feature dimension of the device information), the number of hidden layer nodes is h, and the number of output layer nodes is m (corresponding to the number of categories of the device feature model in the component library). For the input device information feature vector The calculation after the hidden layer is: in is the weight matrix from the input layer to the hidden layer, is the bias vector of the hidden layer, f is the activation function, and the calculation of the output layer is In the formula is the weight matrix from the hidden layer to the output layer, is the bias vector of the output layer, and g is the activation function of the output layer; According to the result of the output layer Select the category with the highest probability as the matching device feature model category;

[0088] According to the adaptation planning result, select appropriate components from the component library or adjust the parameter configuration of the existing components. For example, for the data acquisition component, adjust parameters such as the data reception buffer size and parsing algorithm according to the data format and transmission frequency of the device; for the display component, adjust the interface layout, font size, and chart style according to the screen resolution and display characteristics of the device; during the operation of the application, continuously monitor the changes in the device status, business rule adjustments, and data traffic fluctuations. Specifically, by setting monitoring points in the system, real-time obtain the device operation data, business logic execution status, and network transmission data volume information;

[0089] When a change is detected, analyze the impact of the change on the component function. For example, when the device data transmission frequency increases, determine whether it is necessary to adjust the data caching strategy; if a new data processing logic is added to the business rule, determine whether it is necessary to add new components or adjust the connection relationship between the existing components; according to the adaptation decision result, perform real-time adjustment on the components, including dynamically loading new components, modifying the operating parameters of the components, and re-establishing the communication connection operation between the components, to ensure that the system can still operate stably and provide accurate functional services under the changes of the device and business.

[0090] The requirement conversion module uses natural language processing technology to parse the user requirement text and convert it into low-code development operation instructions and logical frameworks for rapid development; The process is as follows:

[0091] Natural language preprocessing: Perform word segmentation on the natural language text input by the user, split the sentence into individual words or phrases, and label the part of speech for each word. For example, for the sentence "Monitor the temperature in the factory workshop and alarm", after word segmentation, we get "Monitor", "factory workshop", "temperature", "and", "alarm", and at the same time label their corresponding parts of speech. This helps to accurately understand the grammar structure and semantics of the text in the following; Filter out the stop words without actual meaning in the text, such as common function words like "of", "is", "in" and some general words irrelevant to the requirements. After this step, the text is more refined and highlights the key information;

[0092] Semantic understanding and analysis: Identify key entities in the text, such as equipment type ("temperature sensor"), operation object ("workshop"), and action ("monitoring", "alarm"). Use pre-trained entity recognition models, combined with domain dictionaries and context information, to accurately extract entities and their attribute information. Analyze the semantic relationship between entities and determine the logical structure of the requirements. For example, determine the conditional relationship between "monitoring temperature" and "alarm", that is, trigger an alarm action when the temperature meets certain conditions. Clearly represent the logic of the requirements by constructing a semantic relationship graph or similar data structure.

[0093] Operation instruction generation: Based on the results of semantic understanding, search for matching functional components in the component library. For example, for the "temperature monitoring" requirement, match the temperature data acquisition component, threshold judgment component, etc.; for the "alarm" requirement, match the message notification component, etc.; generate operation instructions for calling these components, and build the connection and interaction logic framework between components according to the demand logic, determine the data flow and trigger conditions, and form preliminary low-code development operation instructions and logic framework for subsequent development.

[0094] The code comment generation module is used to automatically add comments to the generated code according to the parsed requirements, improve the code readability and maintainability, and facilitate subsequent development and maintenance work;

[0095] Obtain the parsed structured demand data from the demand conversion module and extract key information, including business functions, operation processes, data sources and flows, and the equipment and components involved; classify and organize this information so that it can be accurately associated with different parts of the code later; plan the structure and content of the comments based on the extracted key demand information, and determine the types of comments that need to be added at the beginning of the code, function definition, key logic nodes, etc. For example, add an overall function overview comment at the beginning of the code; add function input and output parameters and function description comments at the function definition; add judgment condition meaning comments at the conditional judgment statement;

[0096] Generate accurate and clear comment text according to the planned comment content, combined with the grammatical structure and format specifications of the code. Use the code parsing tool to locate the code line or code block to which comments need to be added, and insert the comment text accurately. For variable definitions, the comments explain the purpose and value range of the variables; for code logic blocks, the comments explain the business logic and purpose of their implementation; optimize the format of the generated comments to make them conform to the unified comment style specifications of the team or project, check the completeness and accuracy of the comments, ensure the consistency of the comments with the code, and avoid the situation where the comments are inconsistent with the actual code functions. At the same time, unify and standardize the terms and expressions in the comments to improve the readability of the comments; the comment consistency checking process is as follows:

[0097] Suppose the set of code statements is S = {s1, s2, ..., s n}, the corresponding annotation set is A = {a1, a2, ..., a p}; p is the number of code statements, that is, the number of elements in the code statement set S; for each code statement s1 and its corresponding annotation a1, calculate the consistency index C(s i , a i ), using the semantic similarity calculation method, the code statements and comment texts are converted into semantic vectors (using pre-trained language models or rule-based semantic representation methods), and the code statement s i The semantic vector of Note a i The semantic vector of Where i is the index variable;

[0098] The consistency index in

[0099] q is the vector dimension; j is an index variable used to traverse the vector dimension when calculating the dot product of the semantic vector. It starts from 1 and ends at the vector dimension q, indicating that when calculating the semantic similarity, the elements of the jth dimension in the vector are operated; if the consistency index C(s i , a i ) is lower than the set threshold, the code statement and comment are marked as inconsistent and need to be corrected.

[0100] The code compatibility conversion module is used to identify low-code files of different formats and convert them into system-compatible formats, while also repairing security vulnerabilities and optimizing compatibility;

[0101] Read the low-code file to be converted, analyze the file's header information, file structure features, and code syntax characteristics, and accurately identify the file's original format type by comparing it with various known low-code format standards; for the identified format, start the corresponding syntax parser, decompose the original code into syntax units, such as variable declarations, function definitions, and logical control statements, and establish a mapping relationship from the original syntax unit to the target syntax unit based on the syntax rules and data structure of this system, and perform preliminary code conversion;

[0102] During the conversion process, use security scanning tools to perform static analysis on the converted code to check whether there are common security vulnerabilities in the code, such as buffer overflow, SQL injection vulnerabilities, improper permission management, etc., mark the vulnerabilities found and record relevant information; according to the vulnerability marking information, refer to the security vulnerability repair library and best practices to repair the code vulnerabilities. At the same time, make compatibility adjustments for the parts of the code that are incompatible with this system, such as component calling methods, data type differences, etc. For example, modify the calling parameters of specific components in the original code according to the specifications of this system; perform functional testing and compatibility testing on the repaired and optimized code, simulate various operating scenarios, and ensure that the converted code can run correctly in this system and can work with other modules and components of the system. If problems are found during the test, return to the previous step to make corrections until they pass the test verification.

[0103] A low-code management method based on AIot Internet of Things, including the following steps:

[0104] Template learning and optimization module: First, data is collected, covering past IoT application cases and user operation records. After classification and organization, it enters the analysis and refinement stage. In this stage, the case structure and user behavior are analyzed through algorithms to extract optimization points that can improve the applicability of the template. Then, multiple update plans are formulated based on these optimization points, and the priority of each plan is comprehensively evaluated. Finally, the best plan is selected to update the template to ensure that the template meets system requirements in terms of function and compatibility.

[0105] Dynamic component adaptation: When a new device is connected, the device information is first obtained and adapted to the component library model to determine the initial configuration of the component. During the operation of the system, the changes in the device and business are continuously monitored. Once a change is detected, its impact on the component function is immediately analyzed, and the parameters and connection relationships of the component are dynamically adjusted to ensure the stable operation and functional accuracy of the system.

[0106] Demand conversion: For the demand text input by the user, natural language preprocessing is first performed, including word segmentation, part-of-speech tagging, and stop word filtering, in order to better understand the text semantics. Then, the semantic understanding and analysis phase is entered to identify key entities and semantic relationships, and to build a logical framework for the demand. Finally, matching components are searched in the component library based on the analysis results, and corresponding operation instructions and logical frameworks are generated to provide a basis for subsequent development.

[0107] Code comment generation: After obtaining information from the requirements conversion step, extract key requirements points and classify them to plan the structure and content of the comments. Then generate accurate comment text according to the plan and insert it into the corresponding code location. After inserting the comments, optimize and check the format of the comments to ensure the completeness, accuracy and consistency of the comments with the code, thereby improving the readability and maintainability of the code.

[0108] Code compatibility conversion: Initially, the low-code file is read, and its original format is identified by analyzing the file's header, structure, and syntax features. Then, the syntax parser is started for the format to perform code conversion. During the conversion process, security scanning tools are used to check and mark vulnerabilities. After fixing the vulnerabilities and adjusting the compatibility based on the marked information, the converted code is tested for functionality and compatibility, simulating various operating scenarios. If any problems occur, return to the corresponding steps to make corrections until the code passes the test to ensure that it can run correctly in the system and work with other modules.

[0109] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A low-code management system based on AIot Internet of Things, characterized by: include: Template learning and optimization module, which is used to automatically refine optimization points and update template settings by analyzing historical cases and user data to improve the applicability of templates in practical applications; The component dynamic adaptation module is used to identify the information of newly connected devices and automatically adjust the component configuration, and adapt the component functions in real time according to the changes of devices and services during application operation; The demand conversion module uses natural language processing technology to parse user demand text and convert it into low-code development operation instructions and logical framework for rapid development; The code comment generation module is used to automatically add comments to the generated code according to the parsed requirements, improve the code readability and maintainability, and facilitate subsequent development and maintenance work; The code compatibility conversion module is used to identify low-code files in different formats and convert them into system-compatible formats, while also performing security vulnerability repairs and compatibility optimization processing.

2. According to a low-code management system based on AIot Internet of Things according to claim 1, it is characterized in that: The execution process of the template learning optimization module is as follows: Data collection phase: Collect IoT application case data that has been successfully developed and run from the system database. At the same time, collect user operation records during the use of the system, including template selection preferences, component usage frequency, and operation information for modifying templates. Classify and organize the collected data according to application type, functional module, and device type; Analysis and extraction stage: Use data analysis algorithms to analyze the functional structure of various cases, identify the composition of core functional modules in different applications and the logical relationship between them; By mining user operation records, we can analyze the patterns of users adjusting templates when facing different business needs. Combined with the functional structure analysis and user behavior pattern mining results, optimization points to improve the applicability of the template are extracted, including the adjustment of the template's default functions, the optimization of the interface layout, and the improvement of the initial parameter settings; Template update phase: Based on the extracted optimization points, several template update plans are formulated. The plans include the specific content and steps for adjusting various parts of the template to ensure that the updated template meets the functional optimization requirements while maintaining the stability and compatibility of the system; Test several template update plans, determine the priority of the template update plans, and select the template update plan with the highest priority; The template is updated according to the highest priority template update plan. During the update process, the template is tested and verified to ensure that the new template can meet expectations in terms of functionality and will not introduce new errors or problems. After the update is completed, the new template is stored in the template library for users to use in subsequent development.

3. According to claim 2, a low-code management system based on AIot Internet of Things is characterized in that: The specific operation steps of the template learning optimization module to determine the priority of the template update scheme are as follows: Comprehensive analysis is performed by collecting evaluation parameters, wherein the evaluation parameters include: function prompt amplitude, degree of improvement in user experience and development cost, and the improvement value, body modification ratio and segmentation value are respectively analyzed through the evaluation parameters, and the obtained improvement value, body modification ratio and segmentation value are calibrated as tc, gt and cj, and after normalization, the following formula is entered: ZPA=tc×ο1+gt×ο2+cj×ο3 to obtain the comprehensive judgment value ZPA, wherein ο1, ο2, ο3 are the preset weight coefficients of the improvement value, body modification ratio and segmentation value respectively; The different comprehensive judgment values ​​obtained from the formulated several template update schemes are sorted according to size, and the template update scheme with the largest comprehensive judgment value is selected as the template update scheme with the highest priority.

4. According to claim 3, a low-code management system based on AIot Internet of Things is characterized in that: The specific operation steps of the template learning optimization module to analyze the evaluation parameters to obtain the improvement value, the body improvement ratio and the segmentation value are as follows: Function prompt range: including the number of new functions added, the degree of business process optimization and the improvement of data processing efficiency. The number of new functions added is calculated by counting the number of new functional modules added in each update plan; The degree of business process optimization is quantified by evaluating the degree of optimization of the update plan for the existing business process, specifically by calculating the process simplification ratio and recording it as the simplification ratio; the improvement of data processing efficiency is determined by measuring the improvement of data processing speed after the template is updated, specifically by comparing the time required to process the same amount of data before and after the update, and calculating the time difference; the number of new functions added, the simplification ratio and the time difference are normalized, and the number of new functions added, the simplification ratio and the time difference are used as the side lengths of the triangle to establish the bottom triangle, and then the correction coefficient is used as the height to establish a triangular prism model, and the surface area of ​​the triangular prism model is calculated and recorded as the improvement value, which is used as the standard for measuring the degree of function improvement; Improvement in user experience and ease of operation: Through user testing or simulated operations, the average task time required to complete the operation steps of a standard task is calculated and recorded as the completion time; Improvement in user friendliness: By inviting users to rate the updated template interface, the difference in the updated rating is calculated; Error prompt accuracy: The error ratio is calculated by combining the number of errors encountered by users during the operation after the update with all the operation steps. The completion time, score difference and number of errors are normalized, and the ratio of the score difference to the sum of the completion time and the number of errors is calculated and recorded as the system improvement ratio, which is used to measure the degree of improvement in user experience. Development cost, development time reduction ratio: estimate the reduction ratio of the time required to develop new applications after the update plan is implemented compared with the old template; labor cost savings: calculate the labor cost saved in the development and maintenance process after the update plan is implemented, and quantify it by comparing the reduction in the number of developers and working hours before and after the update. The obtained reduction ratio, reduction in the number of developers and reduction in working hours are normalized and summed to obtain the cost value, which is used as the standard for measuring development cost savings.

5. According to a low-code management system based on AIot Internet of Things according to claim 1, it is characterized in that: The component dynamic adaptation module runs as follows: When a new device attempts to access the system, a connection is established with the device through the communication protocol of the device access layer to obtain basic information about the device, including device type, manufacturer information, device model, and communication interface parameters. At the same time, a detection command is initiated to obtain the functional parameters of the device, including data collection accuracy, supported instruction sets, and data transmission frequency range information; Perform feature analysis on the collected device information and match it with the existing device feature models in the component library. If similar models exist, perform adaptation planning based on the existing adaptation solutions and combined with the parameters of the current device. According to the adaptation planning results, select the corresponding components from the component library or adjust the parameter configuration of the existing components, including: for the data acquisition component, adjust the data receiving buffer size and parsing algorithm parameters according to the data format and transmission frequency of the device; For display components, adjust the interface layout, font size, and chart style according to the device's screen resolution and display characteristics; During the application operation, the device status changes, business rule adjustments and data traffic fluctuations are continuously monitored. Specifically, by setting monitoring points in the system, the device operation data, business logic execution status and network transmission data volume information can be obtained in real time; When changes are detected, analyze the impact of the changes on component functions; if new data processing logic is added to the business rules, determine whether new components need to be added or the connection relationship between existing components needs to be adjusted; Based on the adaptation decision results, components are adjusted in real time, including dynamically loading new components, modifying component operating parameters, and re-establishing communication connections between components to ensure that the system can still operate stably and provide accurate functional services when equipment and business changes.

6. According to claim 5, a low-code management system based on AIot Internet of Things is characterized in that: The judgment process of similar models in the component dynamic adaptation module is as follows: When analyzing the characteristics of the collected device information and matching it with the existing device feature models in the component library, the device feature vector The feature vector corresponding to each device feature model in the component library The cosine similarity is calculated using the following formula: According to the preset similarity threshold τ, when the cosine similarity cos(θ) is greater than or equal to the preset similarity threshold τ, it is determined that a similar model exists; When no similar model is found, the new device adaptation rule generation mechanism is started, and the device information is classified and summarized using the machine learning algorithm to generate a preliminary adaptation strategy; specifically, the device information is classified through a classification algorithm based on a neural network, including a multi-layer perceptron, to determine which type of device feature model in the component library is most matched. The number of input layer nodes is preset to n, the number of hidden layer nodes is h, and the number of output layer nodes is m. For the input device information feature vector The calculation after the hidden layer is: in is the weight matrix from the input layer to the hidden layer, is the bias vector of the hidden layer, f is the activation function, and the calculation of the output layer is In the formula is the weight matrix from the hidden layer to the output layer, is the bias vector of the output layer, g is the activation function of the output layer; according to the result of the output layer The category with the highest probability is selected as the matching device feature model category.

7. According to claim 1, a low-code management system based on AIot Internet of Things is characterized in that: The execution steps of the demand conversion module include: B1. Natural language preprocessing: Perform word segmentation on the natural language text input by the user, split the sentence into individual words or phrases, and mark the part of speech for each word, and mark the corresponding part of speech; filter out meaningless stop words in the text, including function words and common words irrelevant to the needs; B2. Semantic understanding and analysis: Identify key entities in the text, including device type, operation object, and action. Use pre-trained entity recognition models, combined with domain dictionaries and context information, to extract entities and their attribute information. Analyze the semantic relationship between entities to determine the logical structure of requirements. B3. Operation instruction generation: Based on the results of semantic understanding, search for matching functional components in the component library; generate operational instructions for calling these components, and build the connection and interaction logic framework between components according to the demand logic, determine the data flow and trigger conditions, and form preliminary low-code development operational instructions and logic framework for subsequent development.

8. According to claim 1, a low-code management system based on AIot Internet of Things is characterized in that: The execution steps of the code comment generation module are as follows: Obtain the parsed structured demand data from the demand conversion module and extract key information, including business functions, operation processes, data sources and flows, and the equipment and components involved; Categorize and organize this information so that it can be later associated with different parts of the code; According to the extracted key information of the requirements, plan the structure and content of the comments, and determine the types of comments that need to be added at the beginning of the code, at the function definition, and at the logical node position, including adding an overall function overview comment at the beginning of the code; Add function input and output parameters and function description comments in the function definition; add judgment condition meaning comments in the conditional judgment statement; Generate accurate and clear comment text according to the planned comment content, combined with the grammatical structure and format specifications of the code. Use the code parsing tool to locate the code line or code block to which comments need to be added, and insert the comment text accurately. For variable definitions, the comments explain the purpose and value range of the variables; for code logic blocks, the comments explain the business logic and purpose they implement; Optimize the format of the generated comments to make them conform to the unified comment style specifications of the team or project, check the completeness and accuracy of the comments, ensure the consistency between the comments and the code, and avoid the situation where the comments are inconsistent with the actual code functions. At the same time, unify and standardize the terms and expressions in the comments to improve the readability of the comments. The comment consistency checking process is as follows: Suppose the set of code statements is S = {s1, s2, ..., s p }, the corresponding annotation set is A = {a1, a2, ..., a p }; p is the number of code statements, that is, the number of elements in the code statement set S; for each code statement s1 and its corresponding annotation a1, calculate the consistency index C(s i , a i ), using the semantic similarity calculation method, the code statements and comment texts are converted into semantic vectors respectively. Let the code statement s i The semantic vector of Note a i The semantic vector of Where i is the index variable; The consistency index in Where q is the vector dimension; j is another index variable; If the consistency index C(s i , a i ) is lower than the set threshold, the code statement and comment are marked as inconsistent and need to be corrected.

9. According to a low-code management system based on AIot Internet of Things according to claim 1, it is characterized in that: The execution steps of the code compatibility conversion module are as follows: Read the low-code file to be converted, analyze the file's header information, file structure characteristics, and code syntax characteristics, and identify the file's original format type by comparing it with various known low-code format standards; For the identified format, start the corresponding syntax parser, decompose the original code into syntax units, establish a mapping relationship from the original syntax unit to the target syntax unit according to the syntax rules and data structure of this system, and perform preliminary code conversion; During the conversion process, use security scanning tools to perform static analysis on the converted code to check whether there are common security vulnerabilities in the code. Any vulnerabilities found will be marked and recorded. According to the vulnerability tag information, combined with the security vulnerability repair library and best practices, the code vulnerabilities are repaired, and compatibility adjustments are made for the parts of the code that are incompatible with this system, including modifying the calling parameters of the components in the original code according to the specifications of this system; Perform functional and compatibility tests on the repaired and optimized code, simulate different operating scenarios, ensure that the converted code can run correctly in this system and work with other modules and components of the system. If problems are found during testing, return to the previous steps to correct them until they pass the test verification.

10. A low-code management method based on AIot Internet of Things, characterized in that: The application includes a low-code management system based on AIot Internet of Things as described in claim 9, comprising the following steps: S1. Template learning and optimization module: First, data is collected, covering past IoT application cases and user operation records. After classification and organization, it enters the analysis and refinement stage. In this stage, the case structure and user behavior are analyzed through algorithms to extract optimization points that can improve the applicability of the template. Then, multiple update plans are formulated based on these optimization points, and the priority of each plan is comprehensively evaluated. Finally, the best plan is selected to update the template to ensure that the template meets the system requirements in terms of function and compatibility; S2. Dynamic adaptation of components: When a new device is connected, the device information is first obtained and adapted to the component library model to determine the initial configuration of the component. During the operation of the system, the changes in the device and business are continuously monitored. Once a change is detected, its impact on the component function is immediately analyzed, and the parameters and connection relationships of the component are dynamically adjusted to ensure the stable operation and functional accuracy of the system; S3. Requirements conversion: For the requirements text input by the user, natural language preprocessing is first performed, including word segmentation, part-of-speech tagging and stop word filtering, followed by the semantic understanding and analysis stage to identify key entities and semantic relationships, build a logical framework for the requirements, and finally search for matching components in the component library based on the analysis results to generate corresponding operation instructions and logical frameworks; S4. Code comment generation: After obtaining information from step S3, extract the key points of the requirements and classify them to plan the structure and content of the comments. Then generate the comment text according to the plan and insert it into the corresponding code position. After inserting the comments, optimize and check the format of the comments to ensure the completeness, accuracy and consistency of the comments with the code. S5. Code compatibility conversion: initially read the low-code file, identify the original format by analyzing the file's header, structure, and syntax features, and then start the syntax parser for the format to perform code conversion. During the conversion process, use security scanning tools to check and mark vulnerabilities. After fixing the vulnerabilities and adjusting compatibility based on the marked information, perform functionality and compatibility tests on the converted code, simulate various operating scenarios, and if any problems occur, return to the corresponding steps to make corrections until the code passes the test.

Images