Strategy-driven intelligent converged terminal automatic configuration deployment method and system

Abstract

The application discloses a kind of based on strategy driven intelligent fusion terminal automation configuration deployment method and system, it is related to the field of fusion terminal automation configuration technology, this method includes: receiving and parsing user service intention, is compiled into structured policy model;Continuously collect and fuse multi-source data to form context information;Context is monitored and is automatically triggered policy model when meeting condition, and the optimal resource allocation scheme is obtained by multi-objective optimization algorithm in combination with context;The scheme is converted into specific configuration instruction sequence and is executed down;The running performance data after continuous collection execution is compared and analyzed with the evaluation index in policy model, forms closed loop feedback, to evaluate policy efficiency and trigger strategy adjustment or reconfiguration process.The application realizes the whole process automation and dynamic self-adaptation from service intention to intelligent fusion terminal equipment configuration, solves the problems of static configuration rigidity, resource management isolation, policy and intention disconnection and lack of continuous optimization.

Description

Technical Field

[0001] This application relates to the field of automated configuration technology for converged terminals, and in particular to a policy-driven automated configuration and deployment method and system for intelligent converged terminals. Background Technology

[0002] With the deep integration of the Internet of Things (IoT) and edge computing, intelligent converged terminals (such as edge gateways and industrial controllers that integrate computing, storage, networking, and sensing functions) are being deployed extensively in smart manufacturing, smart cities, and other scenarios. However, the large-scale, distributed deployment and dynamic operation and maintenance of intelligent converged terminals pose a significant challenge to traditional configuration management methods. Currently, the mainstream configuration deployment methods include:

[0003] Template- and script-based automated configuration tools, such as Ansible, SaltStack, and Puppet, require administrators to pre-write configuration templates or scripts for different terminal models and roles. These templates are then distributed and executed in batches to the target terminals via a central server to modify configuration files and start or stop services. This method heavily relies on predefined static templates and cannot adaptively adjust to real-time terminal resource status, network conditions, or dynamic business needs.

[0004] Policy management methods based on simple rule engines introduce rule engines into certain gateways or IoT platforms, where administrators configure a series of "condition-action" rules to achieve automated terminal configuration. However, these rules are reactive, discrete, and have a low logical hierarchy; they are difficult to express complex, multi-objective business intents; conflicts easily arise between rules, and they lack the ability to proactively optimize and coordinate from a global perspective.

[0005] In summary, the configuration strategies generated by existing methods are essentially static or semi-static, and cannot be adaptively adjusted after deployment based on dynamic factors such as real-time terminal load, network jitter, and changes in business priorities. This leads to a rapid mismatch between the configuration and the actual operating environment, resulting in decreased performance. Furthermore, there is a lack of collaborative configuration and unified scheduling mechanisms for the computing, storage, network, and application functions within and between intelligent converged terminals. Administrators must manually translate high-level business objectives into a large number of low-level, technical configuration commands or rules, a process that is tedious, error-prone, and costly to adjust when business intentions change.

[0006] Therefore, there is an urgent need for an automated configuration and deployment method for intelligent converged terminals that can directly understand business intent, make dynamic intelligent decisions based on global real-time context, and achieve self-optimization through closed-loop feedback. Summary of the Invention

[0007] To address the technical problems of the prior art, this application provides a strategy-driven automated configuration and deployment method and system for intelligent converged terminals. Through intent understanding, context awareness, and dynamic decision-making, business intents are automatically transformed into refined configuration instructions adapted to different terminals and scenarios. Through closed-loop feedback, continuous evaluation of configuration effects and self-optimization of strategies are achieved, thereby realizing zero-contact deployment, adaptive configuration, and intelligent operation and maintenance of intelligent converged terminals, improving operation and maintenance efficiency, resource utilization, and the stability of business service quality.

[0008] This application provides a policy-driven automated configuration and deployment method for intelligent converged terminals, including: Step S10: Receive and parse the user's intent description oriented towards business objectives, and compile it into a structured strategy model containing operation objects, business objectives, constraints, effective conditions, and evaluation indicators. Step S20: Continuously collect real-time status data from network infrastructure, intelligent converged terminals and upper-layer applications, and process them into contextual information that reflects the overall operating status. Step S30: Monitor context information. When the terminal state change is detected to meet the effective conditions defined in the strategy model, the corresponding strategy model is automatically triggered. Based on the triggered strategy model and its business objectives and constraints, combined with the current context information, the optimal resource configuration scheme adapted to the current real-time environment is generated through a multi-objective optimization decision algorithm. Step S40: Convert the optimal resource configuration scheme into a configuration instruction sequence for the target intelligent converged terminal and send it to the target intelligent converged terminal for execution to complete the automated deployment and activation of the configuration. After the configuration is executed, continuously collect the operating performance data of the target intelligent converged terminal. Step S50 involves comparing and analyzing the collected operational performance data with the predefined evaluation indicators in the strategy model to form a closed-loop feedback, evaluate the effectiveness of the strategy, and trigger the strategy adjustment process.

[0009] Furthermore, in step S10, the detailed steps for compiling and generating the structured strategy model include: Step S11: Perform semantic analysis on the intentions described in the natural language and domain-specific language input by the user, and identify and extract the core elements including the operation object, business objectives and constraints. Step S12: Map the extracted core elements to predefined strategy atoms, where the strategy atom is a basic strategy unit that describes a single constraint and business objective. Step S13: Based on the business logic, combine the relevant strategy atoms and inject the core elements parsed from the intent description, add the time, event conditions and evaluation indicators for the strategy to take effect, and combine them into a complete strategy model that can be loaded and reasoned.

[0010] Furthermore, the real-time status data mentioned in step S20 includes at least network topology information, link bandwidth and latency, packet loss rate, memory utilization of the intelligent fusion terminal, process status, sensor data, service quality requirements defined by the upper-layer application, and service flow priority information.

[0011] Furthermore, in step S30, the operation process of the multi-objective optimization decision algorithm includes: Step S31: Construct an objective function based on the business objectives in the strategy model, and formally define the optimization problem under multiple constraints using the constraints in the strategy model and the real-time available resources in the context information as constraints. Step S32: Solve the optimization problem and output multiple candidate resource configuration schemes; Step S33: Check whether there are any business objectives and resource conflicts between each candidate resource configuration scheme and other policies that are currently in effect in the system, and resolve them according to predefined priority rules; Step S34: From the candidate resource configuration schemes, select the scheme with the highest adaptability to the current system state, the lowest execution risk, and the lowest change cost as the optimal resource configuration scheme to be executed.

[0012] Furthermore, in step S50, the automatic triggering strategy adjustment process includes the following detailed steps: Step S51: Calculate the deviation between the running performance data and the evaluation index. If the deviation continues to exceed the preset first threshold but is lower than the preset second threshold, it is determined that the strategy parameters of the strategy model need to be adjusted. Step S52: When it is determined that adjustment is needed, the deviation between the current running performance data and the evaluation indicators, as well as the latest context information, are fed back to the decision engine as input, and step S30 is triggered. In step S53, if the deviation exceeds the preset second threshold and the context information analysis indicates that the preconditions of the business intent have been fundamentally changed, the strategy model is determined to be inapplicable, a model reconstruction suggestion is generated, and step S10 is triggered.

[0013] This application also provides a policy-driven automated configuration and deployment system for intelligent converged terminals, including: Strategy Model Generation Module: This module receives and parses user intent descriptions oriented towards business objectives, compiles them into a structured strategy model containing operation objects, business objectives, constraints, effective conditions, and evaluation metrics. This module consists of an intent translator, a strategy compiler, and a strategy repository. Data perception and fusion module: used to continuously collect real-time status data from network infrastructure, intelligent converged terminals and upper-layer applications, and fuse and process it into contextual information that reflects the overall operating status; Intelligent Decision Engine: Used to monitor context information. When a change in terminal state is detected that meets the effective conditions defined in the strategy model, the corresponding strategy model is automatically triggered. Based on the triggered strategy model and its business objectives and constraints, combined with the current context information, the optimal resource configuration scheme adapted to the current real-time environment is generated through a multi-objective optimization decision algorithm. This module consists of a context monitor, a strategy loader, a strategy trigger, a decision core, and a conflict resolver.

[0014] Configuration Execution and Feedback Module: This module is used to convert the optimal resource configuration scheme into a sequence of configuration instructions for the target intelligent converged terminal and send it to the target intelligent converged terminal for execution, thereby completing the automated deployment and activation of the configuration. After the configuration is executed, it continuously collects the operating performance data of the target intelligent converged terminal. This module consists of a configuration converter and distributor and a device driver library.

[0015] Evaluation and Optimization Control Module: This module compares and analyzes operational performance data with predefined evaluation metrics in the strategy model to form a closed-loop feedback, evaluate strategy effectiveness, and trigger the strategy adjustment process. This module consists of an effectiveness evaluation engine and an optimization arbitrator.

[0016] This application provides a policy-driven automated configuration and deployment method and system for intelligent converged terminals. By constructing a complete autonomous loop from intent understanding and intelligent decision-making to closed-loop optimization, it effectively solves the core challenges faced by existing methods in the automated configuration and deployment of intelligent converged terminals. Its beneficial effects are mainly reflected in the following four aspects: The method proposed in this invention realizes the transformation of operation and maintenance mode from manual response to intent-driven and automatic execution. Users only need to provide business objectives, and the method can automatically complete the entire process from intent translation, strategy generation, intelligent decision-making to accurate distribution. This not only reduces the requirements for the professional skills of operation and maintenance personnel, but also fundamentally avoids human operation errors, and significantly improves the efficiency and accuracy of large-scale, distributed terminal deployment and changes.

[0017] This method uses global real-time context information for continuous perception and dynamic decision-making. The generated configuration strategy can be adaptively adjusted as the environment changes. This enables the intelligent converged terminal configuration to match the dynamically changing business needs and operating environment in real time, solving the fundamental contradiction of mismatch between static configuration and dynamic environment.

[0018] Furthermore, this method can perform unified modeling and joint optimization of multi-dimensional resources such as computing, storage, network, and applications within and across intelligent converged terminals from a global perspective, generating a globally optimal collaborative configuration scheme. This breaks the resource silo problem existing in traditional configuration and deployment methods, achieves an overall improvement in resource utilization efficiency, and effectively avoids resource waste.

[0019] More importantly, this method evaluates the execution effect of configuration strategies in real time through an automated closed loop, and automatically triggers strategy optimization and refactoring when the objectives are not met. This gives the system the ability to proactively discover problems, automatically optimize, and even self-heal, enabling operations and maintenance to shift from passively responding to faults to proactively ensuring and continuously optimizing, significantly improving the system's resilience and long-term operational stability. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating a policy-driven automated configuration and deployment method for intelligent converged terminals, as provided in an embodiment of this application.

[0021] Figure 2 This is a schematic diagram of the structure of a policy-driven intelligent converged terminal automated configuration and deployment system provided in an embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] Example 1: This application provides a policy-driven automated configuration and deployment method for intelligent converged terminals, such as... Figure 1 As shown, the method includes: Step S10: Receive and parse the user's intent description oriented towards business objectives, and compile it into a structured strategy model containing operation objects, business objectives, constraints, effective conditions, and evaluation indicators.

[0024] In this embodiment, the user inputs their business intent through an interactive interface (Web portal, command-line tool, or API), selecting from the following input methods: Natural language description: For example, a user enters in the text box, "Provide the highest level of network service quality assurance for the video streams provided by all AI quality inspection cameras in production line A. It must be ensured that the end-to-end latency is consistently below 100 milliseconds, and the bandwidth used by a single camera stream must not exceed 30% of the available bandwidth on its access port. This policy is effective during production hours from 8:00 AM to 8:00 PM, Monday to Friday." Structured form filling: For example, users can select "Production Line A" and "AI Quality Inspection Video Stream" as target objects through drop-down menus, input boxes and other form elements, select "Highest Network Service Quality Assurance" as the service level, manually enter "Latency <100ms" and "Bandwidth Utilization <30%" as specific indicators, and set the effective time conditions through the time selector; After receiving user input, the system parses the intent description using an intent translator. For the natural language description entered by the user in the text box, the built-in natural language processing model performs word segmentation, entity recognition, and relation extraction. For structured forms, the corresponding field values ​​are read directly. The parsing process includes: The target of the operation was identified as all devices labeled "AI Quality Inspection Camera" on "Production Line A"; The core business objective is identified as "end-to-end latency", with the constraint of "less than 100 milliseconds"; the qualitative description of "highest level of network service quality assurance" is transformed into a specific internal priority value according to a predefined mapping table, and mapped to priority "7"; The resource constraint was identified as "bandwidth utilization is less than 30%"; The effective time condition is identified as "08:00-20:00 every Monday to Friday"; The natural language processing model segments the user-input Chinese sentence into independent lexical units, identifies and classifies key entities in the sentence, including device and terminal entities, resource entities, and numerical and indicator entities, and extracts entity relationships between operation objects, business objectives, and constraints. The model performs neighborhood adaptive fine-tuning based on a pre-trained BERT model. The BERT configuration file is downloaded from the network, and the BERT is fine-tuned using labeled intent description statements related to the business domain, resulting in a dedicated pre-trained natural language processing model that is integrated into the intent translator.

[0025] After parsing, a preliminary structured intent object is output, containing core elements such as the operation object, business objective, constraints, and effective conditions; The system's built-in policy compiler receives the aforementioned structured core elements and compiles an executable structured policy model. Detailed steps include: The compiler accesses the policy atom library and maps each element to a predefined policy atom. The policy atom library is a structured and extensible knowledge base that stores the smallest, indivisible policy logic units that make up complex policies, called policy atoms. A policy atom is an abstract encapsulation of a single, specific, and explicit configuration intention. The compiler combines all policy atoms into a policy execution graph and performs static analysis in the process to check for potential conflicts between atoms. For example, it checks whether there are extreme cases where low latency and bandwidth constraints are difficult to satisfy simultaneously. The compiler automatically generates quantitative evaluation metrics based on business objectives and constraints and sets thresholds for the performance metrics of each policy atom for subsequent closed-loop evaluation. Finally, the compiler generates a complete, directly loadable and interpretable structured strategy model, which includes the operation object, business objectives, constraints, effective conditions and evaluation metrics, represented in JSON format and stored in the strategy repository.

[0026] Step S20: Continuously collect real-time status data from network infrastructure, intelligent converged terminals and upper-layer applications, and process them into contextual information that reflects the overall operating status.

[0027] In this embodiment, the goal of step S20 is to construct a comprehensive system situational overview to provide a data foundation for subsequent intelligent decision-making; the specific implementation is divided into two main stages: multi-source heterogeneous data acquisition and data processing and fusion. Phase 1: Multi-source heterogeneous data acquisition, collecting real-time status data from network infrastructure, intelligent converged terminals, and upper-layer applications, specifically including: Network infrastructure layer data collection: The collection targets are core switches, routers, firewalls, SDN controllers, and wireless access points. The content and process of collection are as follows: Topology and Connection Status: Obtain the connection relationships between devices and port status from the API via LLDP; Traffic and performance data: Collects IP, port, protocol, packet count, and byte count via NetFlow; collects port-level real-time rate, throughput, packet loss rate, and number of error packets via SNMP; Path status: Obtain the latency and jitter of the critical path through the SDN controller.

[0028] Data collection at the intelligent converged terminal layer: The collection targets are various intelligent converged terminals deployed at the edge, and the collected content includes: Resource status: CPU core utilization, memory usage and utilization, remaining storage space, dedicated accelerator utilization and temperature; System and process status: running processes, container list and their resource usage, operating system logs, and health status of critical services; Local network and interfaces: IP address, speed, traffic statistics, and TCP connection status of each network interface on the terminal; Business and sensor data: The specific business status carried by the terminal, the connected industrial protocol data points, and local sensor readings.

[0029] Upper application layer data collection: The collection targets are business applications, microservices, message brokers, and databases running on terminals, edge servers, or in the cloud; application code instrumentation, sidecar proxies, and service mesh collect the following content: Application performance metrics: service response time, transaction processing rate, error rate, call chain trace data; Business logic metrics: video analytics task queue length, AI model inference latency, data batch processing job completion status, key performance indicators (KPIs); Service and Dependency Status: Obtain the dependency health status between services and the traffic distribution status of the API gateway through the service mesh.

[0030] Phase Two: Data Processing and Fusion. This phase involves the centralized processing of the collected raw data, including the following detailed steps: Data from different protocols is parsed and converted into an internally unified standardized data model to identify and filter transient extreme values ​​caused by network jitter and collection anomalies; all bandwidth units are unified to Mbps, latency to ms, and utilization to percentage, and all data is timestamped synchronously. Create a dynamic resource profile for each intelligent converged terminal, linking all its resource metrics, running processes, service lists, and business tags. Combine this with network topology data and flow data to construct a complete logical view from the intelligent converged terminal to the network path and then to the processing node. The processed data is then integrated around the business entity. For example, for the business entity "Production Line A Quality Inspection Video Stream," its context information dynamically includes: Business status: current frame rate, recognition accuracy, queue waiting time; Terminal status: CPU and GPU utilization and temperature of the edge gateway where the camera is located; Network path status: Real-time bandwidth utilization, latency, and packet loss rate of the links through which the video stream passes; Target service status: The health status and processing latency of the AI ​​inference service receiving this stream.

[0031] Ultimately, it generates contextual information that can be efficiently queried, in real time, and with multiple dimensions of relevance during the intelligent decision-making process.

[0032] Step S30: Monitor context information. When the terminal state change is detected to meet the effective conditions defined in the strategy model, the corresponding strategy model is automatically triggered. Based on the triggered strategy model and its business objectives and constraints, combined with the current context information, an optimal resource configuration scheme adapted to the current real-time environment is generated through a multi-objective optimization decision algorithm.

[0033] In this embodiment, step S30 is executed by the intelligent decision engine. At the time of policy triggering, based on a comprehensive understanding of real-time context information, an executable optimal resource allocation scheme is calculated to achieve specific business objectives. The specific implementation is divided into three stages: automatic policy triggering, multi-objective optimization problem construction and solving, and scheme generation and adjudication. Phase 1: Automatic Strategy Triggering, including the following detailed steps: After the intelligent decision engine starts, it performs initialization operations. The context monitor within the engine receives real-time updated context information. At the same time, the strategy loader within the engine loads all active structured strategy models from the strategy repository into memory, forming a set of strategy models. Each model contains its complete definition: operation object, business objective, constraints, effective conditions, and evaluation indicators. The policy triggers within the engine perform cyclic monitoring for each policy model loaded into memory: The logic expression for the activation of each strategy model is read from its activation condition. The latest context information is used as a variable to perform real-time Boolean evaluation on the above logic expression. When the activation condition of a certain strategy model changes from "0" to "1", it signifies the occurrence of a strategy triggering event. When a policy trigger event is identified, the system immediately creates a policy execution instance, a temporary object containing all the information needed for decision-making. Record the specific strategy model that was triggered, and accurately locate the specific target object of the strategy from the context that caused the strategy triggering event. Save the complete context information closely related to the target object at the moment of triggering as a decision snapshot. Put this instance into the decision task queue and wait for the decision core to process it.

[0034] Phase Two: Problem Construction and Solving for Multi-Objective Optimization, including the following detailed steps: Take a policy execution instance from the decision task queue generated in Phase 1 and transform it into a standard multi-objective optimization problem: Analyze business objectives and the current resource environment to identify adjustable variables, including the amount of bandwidth reserved for the target business. The transmission path selected for the service The target edge server for scheduling business computing tasks The number of CPU cores allocated on the target server And the queue priority weights set on the server. ; The business objectives in the strategy model are transformed into mathematical optimization objectives. For example, with minimizing latency and load balancing as joint objectives, a multi-objective optimization objective function is established. The latency minimization objective is transformed into minimizing a latency prediction function based on a theoretical model, and the load balancing objective is transformed into minimizing the variance of CPU utilization of the server cluster. The two objective functions are weighted and added together to obtain the final overall objective function. Transform the constraints in the policy model and the real-time resource limitations in the context snapshot into mathematical constraints: For example, the constraint "bandwidth utilization less than 30%" in the strategy model is transformed into Real-time resource constraints are that the estimated memory usage is less than or equal to the available memory of the target server, and physical constraints are the pose range limitations of the decision variables, for example... It is an integer, and ; The decision core invokes the built-in optimization solver to solve the multi-objective optimization problem constructed above: Based on the nature of the problem (such as whether it is linear or contains integer variables), the appropriate solution algorithm is automatically selected (simplex method for linear programming, branch and bound method for mixed integer programming, or heuristic genetic algorithm). The solver runs within a strictly defined computation time window, searching for the combination of decision variable assignments that minimizes the overall objective function value within the feasible solution space defined by all constraints. After solving, it outputs multiple candidate resource allocation schemes, each scheme being a set of determined decision variable values.

[0035] Phase Three, Solution Generation and Decision-Making, includes the following detailed steps: The intelligent decision engine's built-in conflict resolver assesses the feasibility, compatibility, and risk of the candidate resource configuration schemes output from Phase 2 to generate the final, unique solution. Check whether the resources required by each candidate solution are already occupied by other active policies in the system. If a conflict is detected, make a decision based on the priority attributes defined in the policy model, allowing higher-priority policies to preempt resources from lower-priority policies. If there is no conflict or the conflict has been resolved, comprehensively evaluate the remaining candidate solutions, including the following factors: Performance benefits: Predicted values ​​of each objective function; Execution risks and costs: the complexity of configuration changes and the potential disruption to other services in the system; Stability: The sensitivity of the scheme to small fluctuations in context parameters; Based on preset weights, the comprehensive score of each scheme is calculated by combining the above evaluation factors, and the scheme with the highest score is selected as the final optimal resource allocation scheme. It is then encapsulated into a standardized resource allocation scheme description object that can be understood by downstream modules.

[0036] Step S40: Convert the optimal resource configuration scheme into a sequence of configuration instructions for the target intelligent converged terminal and send it to the target intelligent converged terminal for execution to complete the automated deployment and activation of the configuration. After the configuration is executed, continuously collect the operating performance data of the target intelligent converged terminal.

[0037] In this embodiment, the goal of step S40 is to accurately, reliably, and atomically apply the optimal resource allocation scheme generated by the intelligent decision engine to the target physical device, and immediately initiate effect monitoring to prepare data for closed-loop evaluation. This includes the following detailed steps: The configuration converter and distributor receives the optimal resource configuration scheme description object generated in step S30 from the output channel of the intelligent decision engine, parses the action fields in the scheme, identifies the target device and resource type to be operated, and translates the abstract resource configuration actions into specific, executable configuration instructions that can be recognized by the corresponding device model and operating system according to the built-in device driver library. All instructions are sorted according to device and execution dependency to generate multiple atomic configuration instruction sequences, each sequence corresponding to a target device, and the instructions in the sequence are idempotent.

[0038] Static validation is performed on the generated raw instructions to check for obvious syntax errors and commands that may cause security risks. The validated instruction sequence, along with the corresponding policy execution ID, expected configuration effect, rollback instructions, and other metadata, is packaged into a configuration transaction package.

[0039] The configuration converter and distributor distributes configuration transaction packets to the execution and feedback agents residing on the target devices via secure, authenticated channels (such as SSH / TLS, NETCONF with TLS, or secure message buses such as MQTT over TLS). The agents on the target devices receive the transaction packets, verify their signatures and integrity, and execute them line by line by calling the device's local configuration interfaces (such as system APIs, CLI wrappers, and device SDKs) according to the instruction sequence within the transaction packets. The agent adopts a "execute one instruction at a time, confirm one instruction at a time" mode. After each instruction is successfully executed, the agent immediately reads the device's runtime status for confirmation. If any instruction fails to execute or the confirmation is inconsistent, the agent stops immediately and records the error point.

[0040] The execution and feedback agent collects the operational performance data of the target intelligent converged terminal based on the evaluation indicators defined in the strategy model. For example, for latency indicators, the agent directly parses the stream-level latency from the local network stack and application logs, rather than just the whole machine CPU data. The data perception and fusion module synchronously adjusts its collection strategy to perform targeted and focused monitoring of network paths, server resources, etc., related to this configuration change.

[0041] The collected performance data is labeled with the corresponding policy execution ID and resource configuration scheme ID, and reported through the standardized channel of the data perception and fusion module, providing direct input for the closed-loop evaluation in step S50.

[0042] Step S50 involves comparing and analyzing the collected operational performance data with the predefined evaluation indicators in the strategy model to form a closed-loop feedback, evaluate the effectiveness of the strategy, and trigger the strategy adjustment process.

[0043] In this embodiment, the goal of step S50 is to automatically determine the health status of the strategy by quantitatively comparing the actual operating results with the expected goals, and to intelligently drive the iteration and optimization of the strategy, including the following detailed steps: The performance evaluation engine of the evaluation and optimization control module performs time series alignment and correlation on the marked runtime performance data, associated context information, and evaluation indicators of the corresponding strategy model extracted from the strategy warehouse in step S40, forming a structured dedicated dataset for strategy performance evaluation.

[0044] The performance evaluation engine performs real-time calculations on the evaluation dataset based on the calculation method, sampling period, and expected range of each evaluation indicator defined in the strategy model, and calculates a performance score for each evaluation indicator. The scoring model includes: Achievement score: Full marks are awarded when the indicator is within the expected range; Deviation deduction: When an indicator deviates, a deduction is calculated based on the severity and trend of the deviation. The greater the deviation and the worse the trend, the lower the score.

[0045] Based on the weights of each indicator (defined in the strategy model), a comprehensive strategy effectiveness score (a value between 0 and 1) is calculated using weighted averages. At the same time, an effectiveness diagnostic report is generated, indicating which specific indicator failed to meet the target, the current deviation value, and the associated contextual factors.

[0046] The arbitrator receives a comprehensive performance score and diagnostic report, and automatically determines the level of optimization action required based on preset rules and a state machine. If the overall score is higher than the preset threshold If all indicators meet the standards, the strategy is considered to be running well, and only logs are recorded without triggering any optimization actions. If the overall score is within the preset threshold and If the error occurs between these parameters, it is determined to be a policy execution deviation. The system generates a policy re-decision request, which includes the policy execution ID, current performance diagnosis, and the latest context information. This request will be sent back to the intelligent decision engine, triggering the re-execution of step S30. If the overall score is lower than the preset threshold If the diagnostic report indicates a fundamental or trend-driven change in the context (e.g., the target terminal model has been widely upgraded, leading to a change in the capability model, or the business intent itself has been updated), then the strategy model is determined to be mismatched. The system generates a strategy reconstruction suggestion, which includes: the original strategy model ID, a detailed failure analysis, and a description of the new business intent or the constraints that need to be modified. This suggestion will be sent to the strategy model generation module, triggering step S10.

[0047] Example 2: The policy-driven automated configuration and deployment system for intelligent converged terminals provided in this embodiment of the invention can execute the policy-driven automated configuration and deployment method for intelligent converged terminals provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method, such as... Figure 2 As shown, it has the following modules: Strategy Model Generation Module: This module receives and parses user intent descriptions oriented towards business objectives, compiles them into a structured strategy model containing operation objects, business objectives, constraints, effective conditions, and evaluation metrics. This module consists of an intent translator, a strategy compiler, and a strategy repository. Data perception and fusion module: used to continuously collect real-time status data from network infrastructure, intelligent converged terminals and upper-layer applications, and fuse and process it into contextual information that reflects the overall operating status; Intelligent Decision Engine: Used to monitor context information. When a change in terminal state is detected that meets the effective conditions defined in the strategy model, the corresponding strategy model is automatically triggered. Based on the triggered strategy model and its business objectives and constraints, combined with the current context information, the optimal resource configuration scheme adapted to the current real-time environment is generated through a multi-objective optimization decision algorithm. This module consists of a context monitor, a strategy loader, a strategy trigger, a decision core, and a conflict resolver.

[0048] Configuration Execution and Feedback Module: This module is used to convert the optimal resource configuration scheme into a sequence of configuration instructions for the target intelligent converged terminal and send it to the target intelligent converged terminal for execution, thereby completing the automated deployment and activation of the configuration. After the configuration is executed, it continuously collects the operating performance data of the target intelligent converged terminal. This module consists of a configuration converter and distributor and a device driver library.

[0049] Evaluation and Optimization Control Module: This module compares and analyzes operational performance data with predefined evaluation metrics in the strategy model to form a closed-loop feedback, evaluate strategy effectiveness, and trigger the strategy adjustment process. This module consists of an effectiveness evaluation engine and an optimization arbitrator.

[0050] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0051] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A policy-driven automated configuration and deployment method for intelligent converged terminals, characterized in that, The method includes: Step S10: Receive and parse the user's intent description, which is business-goal oriented, and compile it into a structured strategy model containing the operation object, business objective, constraints, effective conditions, and evaluation indicators. The compilation and generation process of the structured strategy model includes: Semantic analysis is performed on the intent described by the user's input in natural language and domain-specific language to identify and extract core elements including the operation object, business objectives and constraints; The extracted core elements are mapped to predefined strategy atoms, which are basic strategy units that describe single constraints and business objectives. Based on business logic, relevant strategy atoms are combined and injected with the core elements parsed from the intent description, along with the timing, event conditions, and evaluation metrics for strategy activation, to form a complete strategy model that can be loaded and reasoned. Step S20: Continuously collect real-time status data from network infrastructure, intelligent converged terminals, and upper-layer applications, and process and fuse them into contextual information reflecting the overall operational status. The real-time status data includes at least network topology information, link bandwidth and latency, packet loss rate, memory utilization of intelligent fusion terminals, process status, sensor data, service quality requirements defined by upper-layer applications, and service flow priority information. Create a dynamic resource profile for each intelligent converged terminal, linking all its resource metrics, running processes, service lists, and business tags. Combine this with network topology data and flow data to construct a complete logical view from the intelligent converged terminal to the network path and then to the processing node. Then, integrate the processed real-time status data with the business entity as the center, linking the business status, the status of the carrying terminal, the network path status, and the status of the target service. Step S30: Monitor context information. When the terminal state change is detected to meet the effective conditions defined in the strategy model, the corresponding strategy model is automatically triggered. Based on the triggered strategy model and its business objectives and constraints, combined with the current context information, the optimal resource configuration scheme adapted to the current real-time environment is generated through a multi-objective optimization decision algorithm. Step S40: Convert the optimal resource configuration scheme into a configuration instruction sequence for the target intelligent converged terminal and send it to the target intelligent converged terminal for execution to complete the automated deployment and activation of the configuration. After the configuration is executed, continuously collect the operating performance data of the target intelligent converged terminal. Step S50: Compare and analyze the collected operational performance data with the predefined evaluation metrics in the strategy model to form a closed-loop feedback, evaluate the strategy effectiveness, and trigger the strategy adjustment process. The triggering strategy adjustment process includes: Calculate the deviation between the operating performance data and the evaluation indicators. If the deviation continues to exceed the preset first threshold but is lower than the preset second threshold, it is determined that the strategy parameters of the strategy model need to be adjusted. When it is determined that adjustment is needed, the deviation between the current operating performance data and the evaluation indicators, as well as the latest context information, are fed back to the decision engine as input, and step S30 is triggered. If the deviation exceeds the preset second threshold, and the context information analysis indicates that the preconditions of the business intent have fundamentally changed, then the strategy model is determined to be inapplicable, a model reconstruction suggestion is generated, and step S10 is triggered.

2. The strategy-driven automated configuration and deployment method for intelligent converged terminals as described in claim 1, characterized in that, In step S30, the operation of the multi-objective optimization decision algorithm includes: Step S31: Construct an objective function based on the business objectives in the strategy model, and formally define the optimization problem under multiple constraints using the constraints in the strategy model and the real-time available resources in the context information as constraints. Step S32: Solve the optimization problem and output multiple candidate resource configuration schemes; Step S33: Check whether there are any business objectives and resource conflicts between each candidate resource configuration scheme and other policies that are currently in effect in the system, and resolve them according to predefined priority rules; Step S34: From the candidate resource configuration schemes, select the scheme with the highest adaptability to the current system state, the lowest execution risk, and the lowest change cost as the optimal resource configuration scheme to be executed.

3. The strategy-driven automated configuration and deployment method for intelligent converged terminals as described in claim 2, characterized in that, Step S33 includes the following detailed steps: Check whether the resources required by each candidate solution are already occupied by other active policies in the system. If a conflict is detected, make a decision based on the priority attribute defined in the policy model, allowing higher priority policies to preempt the resources of lower priority policies. If there is no conflict or the conflict has been resolved, the remaining candidate solutions are comprehensively evaluated, and the evaluation factors include performance benefits, execution risks and costs, and stability.

4. A policy-driven automated configuration and deployment system for intelligent converged terminals, characterized in that: The system is used to implement the policy-driven automated configuration and deployment method for intelligent converged terminals as described in any one of claims 1-3, and the system comprises: Strategy Model Generation Module: This module receives and parses user intent descriptions oriented towards business objectives, compiles them into a structured strategy model containing operation objects, business objectives, constraints, effective conditions, and evaluation metrics. This module consists of an intent translator, a strategy compiler, and a strategy repository. Data perception and fusion module: used to continuously collect real-time status data from network infrastructure, intelligent converged terminals and upper-layer applications, and fuse and process it into contextual information that reflects the overall operating status; Intelligent Decision Engine: Used to monitor context information. When it detects that the terminal state change meets the effective conditions defined in the strategy model, it automatically triggers the corresponding strategy model. Based on the triggered strategy model and its business objectives and constraints, combined with the current context information, it generates the optimal resource configuration scheme adapted to the current real-time environment through a multi-objective optimization decision algorithm. This module consists of a context monitor, a strategy loader, a strategy trigger, a decision core, and a conflict resolver. Configuration execution and feedback module: This module is used to convert the optimal resource configuration scheme into a sequence of configuration instructions for the target intelligent converged terminal and send it to the target intelligent converged terminal for execution, so as to complete the automated deployment and activation of the configuration. After the configuration is executed, it continuously collects the operating performance data of the target intelligent converged terminal. This module consists of a configuration converter and distributor and a device driver library. Evaluation and Optimization Control Module: This module compares and analyzes operational performance data with predefined evaluation metrics in the strategy model to form a closed-loop feedback, evaluate strategy effectiveness, and trigger the strategy adjustment process. This module consists of an effectiveness evaluation engine and an optimization arbitrator.

Images