Distributed rule engine self-evolution system and method based on AI Agent drive, computer equipment and medium

Abstract

The invention discloses a distributed rule engine self-evolution system and method based on AI Agent driving, computer equipment and a medium, the system comprises a rule evolution platform on a cloud side and a rule execution engine on an end side, and the rule evolution platform on the cloud side comprises an industry AI Agent service, a digital twinning and data lake, a rule generator and a rule distribution service. The industry AI Agent service is an intelligent agent driven by one or more industry large models and is used for receiving and understanding a business target; the rule generator is used for converting decision information output by the industry AI Agent service into a rule process; the rule distribution service is used for pushing the rule flow to equipment specified by the end side; and the rule execution engine at the end side is used for performing local closed-loop execution on the rule process. According to the method, a complex operation and maintenance mode can be converted into AI autonomous learning and optimization from manual rule writing and maintenance, manpower is greatly liberated, and the optimization effect far better than that of human experts can be achieved.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of artificial intelligence and distributed control, and in particular to a self-evolving system, method, computer device and medium for a distributed rule engine driven by AI Agent. Background Technology

[0002] Current IoT operating systems offer a visual super rule engine. Developers or operations personnel can drag and drop functional nodes (such as sensor triggers, device control, delays, and logical judgments) to build business logic (scene linkage rules) in a "low-code" or "no-code" manner, greatly reducing the application development threshold. For example, it is easy to create a rule such as "when the smoke sensor value exceeds the threshold, immediately turn on the sprinkler and start the exhaust fan".

[0003] The existing technology has at least the following drawbacks: (1) Rules programmed manually are static. Once deployed, their triggering conditions and execution logic are fixed. In the face of complex and dynamically changing real-world scenarios (such as tidal changes in traffic flow and process adjustments in industrial production lines), these static rules cannot adapt and are prone to inefficiency or even errors, requiring a lot of manpower for continuous maintenance and optimization.

[0004] (2) The decision-making ability of the rule engine is limited by the preset simple logic. It cannot perform complex reasoning and prediction based on global information. For example, in smart building energy management, the simple rule of "turn on the air conditioner when the temperature is above 26 degrees" is far less energy-efficient and effective than an expert decision that can comprehensively consider outdoor weather, future electricity prices, and the distribution of people in the building.

[0005] (2) The optimal rules for complex scenarios are often stored in the minds of a few senior industry experts. This tacit knowledge is difficult to effectively and scalably solidify into the system. When experts leave or the scenario changes, the optimization of the rules becomes a problem.

[0006] Therefore, existing technologies still need improvement. Summary of the Invention

[0007] The technical problem to be solved by this invention is to provide a self-evolving method, system, computer device, and medium for a distributed rule engine driven by AIAgent, addressing the aforementioned deficiencies of the prior art. The technical solution adopted by this invention is as follows: In a first aspect, the present invention provides a distributed rule engine self-evolution system based on AI Agent driven, the system comprising: a rule evolution platform on the cloud side and a rule execution engine on the edge side; The cloud-based rule evolution platform includes: Industry AI Agent services are intelligent agents driven by one or more large industry models, used to receive and understand business objectives; Digital twins and data lakes are used to aggregate real-time status data, historical data, and rule execution logs reported by all edge devices, and to build a digital twin model that is synchronized with the physical world in real time. A rule generator is used to transform the decision information output by the industry AI Agent service into a rule process that can be executed by the edge device. The rule distribution service is used to push all rule processes generated by the rule generator to the specified devices on the edge. The rule execution engine on the endpoint is used to perform local closed-loop execution of the rule process issued from the cloud.

[0008] In one implementation, the endpoint rule execution engine further includes: The Agent call node is used to package context data into the cloud-based industry AI Agent service and initiate decision consultation to the industry AI Agent service.

[0009] In one implementation, the endpoint rule execution engine further includes: The data reporting service is used to report real-time status data, historical data, and rule execution logs of the device on the edge to the digital twin and data lake on the cloud side.

[0010] In one implementation, the cloud-based rule evolution platform further includes: A rule optimizer is used to transform the decision information into a structured description file that conforms to the rule engine format. The description file can be a newly generated rule flow or a rule flow obtained by adjusting the parameters of an existing rule flow.

[0011] In one implementation, the cloud-based rule evolution platform further includes: The expert review node is used to push the rule process to a designated operation and maintenance expert for review after the rule generator generates the rule process, and to transfer the review result of the operation and maintenance expert to the rule optimizer.

[0012] In one implementation, the expert review node includes an intelligent pre-review mode and a manual final review mode.

[0013] In one implementation, the rule distribution service includes an on-demand push mode, a batch push mode, and an incremental update mode. In another implementation, the endpoint is encapsulated as a lightweight micro-agent to execute the rule process, and the micro-agent can make autonomous decisions and make dynamic adjustments locally.

[0014] In one implementation, the industry AI Agent service can also be deployed on edge computing nodes.

[0015] Secondly, embodiments of the present invention also provide a self-evolution method for a distributed rule engine driven by an AI Agent, wherein the self-evolution of the distributed rule engine driven by an AI Agent is applied to the self-evolution system of the distributed rule engine driven by an AI Agent as described in any of the above solutions, and the method includes: Once the cloud-based industry AI Agent service receives the business objective, it acquires real-time status data, historical data, and rule execution logs. Based on analysis and reinforcement learning of digital twin models, a rule generator is used to generate rule processes. The rule process is sent to the endpoint, and the rule process is executed locally in a closed loop based on the endpoint's rule execution engine.

[0016] In one implementation, after the cloud-based industry AI Agent service receives the business objective, it acquires real-time status data, historical data, and rule execution logs, including: Obtain the business objective in the natural language form of the input; The semantic understanding module breaks down the core requirements of the business objectives, extracts key indicators, and associates them with the corresponding industry's business knowledge graph. The receiving end uploads real-time status data, historical data, and rule execution logs.

[0017] In one implementation, the method further includes: When the cloud receives new rule execution logs or new rule generation conditions, the rule optimizer generates new rule processes based on these logs and sends them to the edge.

[0018] In one implementation, the method further includes: During the execution of the rule process on the client side, when the Agent call node is executed, the context data is packaged and sent to the industry AI Agent service on the cloud side. After receiving the context data, the cloud-based industry AI Agent service makes global decisions based on the analysis and reinforcement learning of the digital twin model and generates decision results.

[0019] In one implementation, the method further includes: After generating the rule process based on the rule generator, the cloud side pushes the rule process to the expert review node, reviews it based on the designated operation and maintenance experts, and transfers the review results of the operation and maintenance experts to the rule optimizer.

[0020] In one implementation, after the cloud side generates a rule flow based on the rule generator, it pushes the rule flow to the expert review node for review by a designated operations and maintenance expert, including: The rules and processes are verified for compliance, and those that pass the verification are pushed to the expert review node. The intelligent pre-review mode of the expert review node is activated, and the rule process that has passed the verification is pre-reviewed based on the intelligent pre-review mode; The manual final review mode of the expert review node is initiated, and the rules and processes that have passed the pre-review are pushed to the designated operation and maintenance experts for review.

[0021] In one implementation, the rule process is sent to the endpoint, including: Obtain the tag information of the rule process, which is set by the rule generator when generating the rule process; Based on the tag information, a push mode is determined, and the rule process is pushed to the terminal based on the push mode. The push mode includes: on-demand push mode, batch push mode, and incremental update mode.

[0022] In one implementation, the method further includes: The edge device is encapsulated into a lightweight micro-agent to execute the rule process, and the micro-agent can make autonomous decisions and make dynamic adjustments locally.

[0023] In one implementation, the method further includes: The receiving end analyzes the execution result of the rule process to determine whether the execution result meets the expected requirements. If the execution result does not meet the expected requirements, the industry AI Agent service receives the execution log of the rule process, regenerates a new rule process based on the rule generator, and sends it to the edge.

[0024] Thirdly, embodiments of the present invention also provide a computer device, wherein the computer device includes a memory, a processor, and an AI Agent-driven distributed rule engine self-evolution program stored in the memory and executable on the processor. When the processor executes the AI ​​Agent-driven distributed rule engine self-evolution program, it implements the steps of the AI ​​Agent-driven distributed rule engine self-evolution method of any of the above-mentioned schemes.

[0025] Fourthly, embodiments of the present invention also provide a computer-readable storage medium, wherein the computer-readable storage medium stores a self-evolving program for a distributed rule engine driven by an AI Agent, the self-evolving program for a distributed rule engine driven by an AI Agent implementing the steps of the self-evolving method for a distributed rule engine driven by an AI Agent as described in any of the above schemes on the computer-readable storage medium.

[0026] Beneficial Effects: Compared with existing technologies, this invention provides a self-evolving distributed rule engine system based on AI Agent-driven architecture. The system includes a cloud-side rule evolution platform and an edge-side rule execution engine. The cloud-side rule evolution platform includes an industry AI Agent service, a digital twin and data lake, a rule generator, and a rule distribution service. The industry AI Agent service is an intelligent agent driven by one or more industry-wide models, used to receive and understand business objectives. The digital twin and data lake aggregates real-time status data, historical data, and rule execution logs reported by all edge devices, and constructs a digital twin model synchronized with the physical world in real time. The rule generator transforms the decision information output by the industry AI Agent service into executable rule flows for edge devices. The rule distribution service pushes all rule flows generated by the rule generator to designated edge devices. The edge-side rule execution engine performs local closed-loop execution of the rule flows distributed from the cloud side.

[0027] This invention completely transforms the operation and maintenance mode of complex systems from manually writing and maintaining rules to AI autonomously learning and optimizing, greatly liberating human resources and achieving optimization results far exceeding those of human experts. Furthermore, by loading different industry-specific large-scale models, the AI ​​Agent service can act as a built-in, tireless industry expert (such as energy consumption experts or traffic experts), providing continuous, accurate, and forward-looking decision support. In this invention, the local closed-loop execution of edge devices ensures the absolute reliability of basic functions and emergency response, while the cloud-based AI brain endows the system with powerful intelligence to cope with complex and unknown situations. The combination of these two aspects meets the stringent requirements of serious scenarios such as industrial applications. In addition, this invention provides a standard, reusable framework, enabling advanced AI large-scale model capabilities to be safely, efficiently, and cost-effectively deployed to thousands of edge devices and scenarios, greatly accelerating the industrialization of AI. Attached Figure Description

[0028] Figure 1 This is a principle block diagram of a distributed rule engine self-evolution system based on AI Agent driven by an embodiment of the present invention.

[0029] Figure 2This is a schematic diagram of the principle architecture of a self-evolving distributed rule engine system based on AI Agent driven by an embodiment of the present invention.

[0030] Figure 3 This diagram illustrates the application environment of the AI ​​Agent-driven distributed rule engine self-evolution method provided in this embodiment of the invention.

[0031] Figure 4 This is a flowchart of a preferred embodiment of the AI ​​Agent-driven distributed rule engine self-evolution method provided in this invention.

[0032] Figure 5 This is a flowchart illustrating the process of obtaining relevant data in the AI ​​Agent-driven distributed rule engine self-evolution method provided in this embodiment of the invention.

[0033] Figure 6 The execution flowchart of the Agent calling node in the self-evolution method of the distributed rule engine based on AI Agent driven by the embodiment of the present invention is shown in the figure.

[0034] Figure 7 The execution flowchart of the expert review node in the AI ​​Agent-driven distributed rule engine self-evolution method provided in the embodiments of the present invention.

[0035] Figure 8 This is an execution flowchart of the rule distribution process in the AI ​​Agent-driven distributed rule engine self-evolution method provided in this embodiment of the invention.

[0036] Figure 9 This is a flowchart illustrating the execution result analysis in the AI ​​Agent-driven distributed rule engine self-evolution method provided in this embodiment of the invention.

[0037] Figure 10 This is a closed-loop workflow diagram of the AI ​​Agent-driven distributed rule engine self-evolution method provided in an embodiment of the present invention.

[0038] Figure 11 This is an overall diagram of the AI ​​Agent-driven distributed rule engine self-evolution method provided in an embodiment of the present invention.

[0039] Figure 12 A schematic block diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0041] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content, operations, or steps, nor does it require execution in the described order. For example, some operations or steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0042] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. For example, the first control information and the second control information are only used to distinguish different control information and do not limit their order. Those skilled in the art will understand that the words "first" and "second" do not limit the quantity or the order of execution, and that the words "first" and "second" do not necessarily imply that they are different. It should also be understood that the terms "and / or" as used in this specification and the appended claims refer to any combination of one or more of the associated listed items and all possible combinations, and include such combinations.

[0043] To address the core pain points of existing rule engines mentioned above, such as their static nature, shallow decision-making capabilities, and reliance on human experts, this invention provides a self-evolving distributed rule engine system driven by an AI Agent. This system aims to deeply integrate the deep reasoning and learning capabilities of AI Agent services with the reliable execution capabilities of the rule engine. It achieves a paradigm shift in business rules from "human design" to "AI-driven autonomous generation and optimization," enabling rules to adaptively evolve based on environmental changes and operational performance. Furthermore, it constructs an edge-cloud collaborative intelligent decision-making closed loop, transforming industry expert knowledge into scalable and continuously iterative AI Agent services through large-scale models, ultimately achieving "zero-code" and even "self-driven" intelligent solutions for complex business scenarios.

[0044] The core idea of ​​this invention is to use the AI ​​Agent service as the "brain" and the super rule engine as the "cerebellum and neural network". The AI ​​Agent service is responsible for complex situational awareness, learning and decision-making on the cloud or edge side, continuously generating and optimizing concise and efficient execution rules; while the edge-side rule engine is responsible for highly reliable, low-latency local closed-loop execution of these rules and feeding back execution data, forming a complete self-evolving intelligent closed loop of "perception-cognition-decision-execution-learning".

[0045] In practical applications, such as Figure 1 As shown, the AI ​​Agent-driven distributed rule engine self-evolution system of this embodiment includes a cloud-side / edge-side rule evolution platform and an edge-side rule execution engine. Specifically, the cloud-side rule evolution platform includes: industry AI Agent service, digital twin and data lake, rule generator, and rule distribution service.

[0046] The industry AI Agent service is the core of decision-making. It is an intelligent agent driven by one or more large industry models (such as a tunnel management model or a building energy consumption model) to receive and understand business objectives, such as maintaining carbon monoxide concentration in the tunnel below safe levels and minimizing fan energy consumption. Digital twins and a data lake are used to aggregate real-time status data, historical data (including historical fault data and historical maintenance data), and rule execution logs reported by all edge devices. This aggregated data constructs a digital twin model that is synchronized with the physical world in real time. The digital twin model serves as the data foundation for the industry AI Agent service's learning and inference. In this embodiment, the digital twin model can construct a digital twin of the device based on the reported data using 3D modeling and real-time rendering technology. The twin's parameters (such as temperature and vibration frequency) are synchronized with the physical device at millisecond levels. Simultaneously, machine learning algorithms are used to simulate and analyze the twin, predicting the device's future operating status (such as predicting the temperature value 10 minutes later based on the current temperature change trend), providing forward-looking data support for the generation and optimization of rule processes. When hardware changes or the operating environment of the edge device changes, the twin model adjusts synchronously through an incremental update mechanism, avoiding the resource consumption of a full rebuild. The rule generator acts as the hand of the industry AI Agent service, transforming the decision information output by the industry AI Agent service into executable rule flows for the edge device. The rule distribution service pushes all rule flows generated by the rule generator to the specified edge devices. During the transformation of decision information into rule flows, the complexity of the rules needs to be adjusted based on the hardware capabilities of the edge device (such as computing resources and communication bandwidth). For example, for edge devices with limited computing power, the condition judgment levels of the rules are simplified. The rule flow includes core elements such as trigger nodes, condition judgment nodes, execution nodes, and Agent invocation nodes.

[0047] Furthermore, the cloud-based rule evolution platform described in this embodiment also includes a rule optimizer and an expert review node. The rule optimizer is used to transform the decision information into a structured description file conforming to the rule engine format. This description file can be a newly generated rule flow, or a rule flow obtained by adjusting the parameters of an existing rule flow. For example, a new rule flow can be regenerated by modifying the trigger threshold or adding rule generation conditions.

[0048] The cloud-side rule evolution platform in this embodiment also includes an expert review node. This node pushes the rule process generated by the rule generator to a designated operations expert for review, and then forwards the expert's review results to the rule optimizer. Experts can "like," "reject," or suggest modifications to the rule process. Approved rule processes directly enter the rule distribution queue; rules that fail review (e.g., those with logical conflicts or potential for device malfunctions) are annotated by experts with the problem type (e.g., "threshold settings are too low, leading to false alarms") and fed back to the rule optimizer. The rule optimizer adjusts the structured description file based on expert feedback, re-triggers the rule generation process, and forms a closed loop of "generation-review-optimization." This feedback from human experts serves as a strong signal and is incorporated into the reinforcement learning reward function of the industry AI Agent service, thereby guiding the industry AI Agent service to learn the optimal strategy that conforms to human expectations and industry standards more quickly and safely. This solves the "black box" and "uncontrollable" problems of AI decision-making, making it particularly suitable for safety-critical fields (such as healthcare and gas), ensuring that the evolution of AI is always under the supervision of human experts, and increasing the credibility of the system.

[0049] In one implementation, the expert review node in this embodiment includes an intelligent pre-review mode and a manual final review mode. In practical applications, after the rule flow is generated, the industry AI Agent service first performs compliance verification on the rule flow, such as verifying whether it conforms to industry security standards and whether there are logical conflicts. The rule flow that passes the verification is then pushed to the expert review node. Next, the intelligent pre-review mode of the expert review node is activated, and the rule flow that passes the verification is pre-reviewed based on this mode. Then, the manual final review mode of the expert review node is activated, and the rule flow that passes the pre-review is pushed to the designated operation and maintenance expert for review. Experts can view the rule details, decision basis, and simulated execution review methods through a visual interface, and the review results are synchronized to the rule optimizer in real time. Furthermore, for rule flows of core equipment (such as nuclear power plant reactor control equipment), multi-level expert review can be mandated to ensure zero risk.

[0050] In one implementation, the rule distribution service includes an on-demand push mode, a batch push mode, and an incremental update mode. In a specific application, this embodiment can obtain tag information of the rule flow when issuing the rule flow. This tag information is set by the rule generator when generating the rule flow and reflects the push method and target device of the rule flow. Therefore, based on the tag information, the push mode can be determined, including on-demand push mode, batch push mode, and incremental update mode, and then the rule flow is pushed to the endpoint based on the push mode.

[0051] The edge-side rule execution engine in this embodiment is used for local closed-loop execution of rule flows issued from the cloud side. Even if the connection with the cloud side is lost, the basic emergency rules on the local side can still function normally. The edge-side rule execution engine includes Agent invocation nodes and a data reporting service. The Agent invocation node is a special type of remote AI decision node. When the rule flow reaches this node, it does not perform simple local judgment, but packages the context data (such as real-time data from multiple sensors) to the industry AI Agent service on the cloud side and initiates decision consultation to the industry AI Agent service on the cloud side via a soft bus. The data reporting service is used to report the real-time status data, historical data, and rule execution logs (such as the executed rule flow, execution results, and execution duration) of the edge-side device to the digital twin and data lake on the cloud side.

[0052] On the device side, multi-dimensional data is collected via a data reporting service, including: real-time status data (such as device operating parameters, sensor monitoring data, and network connection status), historical data (such as operating trajectory data for the past 72 hours), and rule execution logs (such as rule trigger time, execution results, and anomaly information). During the collection process, a data cleaning module filters out noisy data (such as abnormal values ​​from sensor false alarms) and standardizes the data (such as unifying data formats and unit conversions). The processed data is reported to the cloud-based digital twin and data lake via a pre-defined communication link. Real-time data is written to a time-series database, supporting rapid queries by time dimension; historical data and log data are written to a distributed file system for long-term archiving and batch analysis. Data reporting includes metadata such as a unique device identifier, collection timestamp, and data reliability score, facilitating data traceability and quality assessment on the cloud side.

[0053] Combination Figure 2 As shown, from Figure 2As can be clearly seen, the system consists of two parts: a "cloud-based rule evolution platform" and an "edge-based rule execution engine." The cloud-based industry AI Agent service is responsible for learning, inference, and rule optimization, aggregating all data through digital twins and a data lake. It is evident that the digital twins and data lake provide data support for the industry AI Agent service, which in turn performs learning and inference based on them. Next, the industry AI Agent service uses a rule generator to transform decision information into executable rule flows, which are then optimized by a rule optimizer. A rule distribution service is responsible for uniformly distributing the rule flows. The edge-based rule execution engine performs local closed-loop execution based on the distributed rule flows and, when necessary, initiates real-time decision consultation with the industry AI Agent service through Agent call nodes, achieving collaborative autonomy between intelligence and rules. Furthermore, during local closed-loop execution, edge-based sensors and actuators generate real-time data, which is reported to the digital twins and data lake, simultaneously providing data support for the cloud-based industry AI Agent service.

[0054] In one implementation, the edge device in this embodiment can be encapsulated as a lightweight micro-agent to execute rule processes. This micro-agent can make autonomous decisions and dynamically adjust locally, and can be deployed across devices using containerization technology (such as Docker). In this case, the cloud no longer issues rule processes in JSON format, but instead issues a compiled and optimized "micro-Agent package" containing objectives and constraints. The edge device's execution engine is no longer a simple process parser, but an Agent runtime. This grants the edge device greater autonomy and flexibility, enabling it to better handle sudden and uncertain local events, representing an intermediate form in the evolution towards a fully autonomous device. Similarly, the micro-agent incorporates Agent call nodes, data reporting services, and a local rule execution module. The data reporting service is configured with both scheduled and triggered reporting mechanisms. The scheduled reporting cycle can be configured according to business needs (e.g., 1 minute / time), and immediate reporting is triggered when the device status experiences abnormal fluctuations. The Agent call node pre-configures the communication interface and authentication information of the cloud-side industry AI Agent service to ensure secure interaction.

[0055] Complex industry AI agent services primarily run on the cloud side with powerful computing capabilities. In other implementations, this embodiment addresses scenarios with extremely high latency requirements and a need for local intelligence (such as high-speed production line quality inspection and vehicle-to-everything (V2X) communication). A lightweight or distilled industry AI agent service can be deployed on edge computing nodes (such as the RK3588 gateway), resulting in an edge agent. In this way, the edge agent calling node prioritizes initiating decision-making consultations with the edge agent within the local area network. The edge agent handles most of the real-time decision-making tasks. Communication with the main industry AI agent service on the cloud side only occurs when global historical data is needed for model retraining or for large-scale complex simulations. This significantly reduces decision latency, decreases reliance on wide area network bandwidth, and enhances system privacy and autonomous intelligence capabilities in offline scenarios.

[0056] This embodiment employs a hybrid communication protocol of "MQTT+HTTPS" to establish a cloud-based interactive channel. Real-time status data and rule execution logs from the edge devices are transmitted via MQTT to ensure efficient transmission under low bandwidth conditions. The Agent calling nodes and the industry AI Agent service's decision-making consultation and rule distribution service's rule push are transmitted via HTTPS to ensure data transmission security. Simultaneously, a gateway service is deployed on the cloud side to perform edge device authentication, data encryption / decryption, and request routing, preventing unauthorized device access.

[0057] In other implementations, this embodiment can also include data security, fault tolerance, and scalability assurance mechanisms. Specifically, the data security mechanism is manifested in: AES encryption is used before data is reported from the device side; data anonymization and access control are implemented in cloud storage to ensure the security of device and business data; a token authentication mechanism is used during Agent invocation to prevent malicious requests. The fault tolerance mechanism is manifested in: the cloud-side rule distribution service has a retry mechanism; when the device is offline, rules are temporarily stored in a local queue and automatically pushed back when the device is online; when the device's micro-agent fails to execute a rule, a backup rule (such as a default safe running rule) is automatically activated to avoid device downtime. The scalability assurance mechanism is manifested in: the cloud side adopts a microservice architecture, supporting the addition of industry AI Agent services and the expansion of data lake storage capacity; the device's micro-agent supports rule plug-in deployment, facilitating the addition of new rule types and execution actions.

[0058] Through the above implementation process, the system has built a distributed rule engine system of "cloud-side evolution, edge-side execution, data-driven, and autonomous optimization". It not only leverages the intelligent decision-making and big data analysis advantages of cloud-side AI Agent, but also relies on edge-side micro-intelligent agents to achieve low-latency execution, effectively improving the response efficiency and adaptability of business scenarios.

[0059] The modules in the aforementioned AI Agent-driven distributed rule engine self-evolution system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the terminal in hardware form or independent of it, or stored in the terminal's memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0060] Based on the above embodiments, the present invention also provides a self-evolution method for a distributed rule engine driven by an AI Agent. This self-evolution method for a distributed rule engine driven by an AI Agent can be applied to, for example... Figure 3 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be placed in the cloud or on other network servers. The AI ​​Agent-driven distributed rule engine self-evolution method can be executed by terminal 102 or server 104, or it can be executed collaboratively by terminal 102 and server 104.

[0061] The terminal 102 can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, IoT device, or portable wearable device. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices, etc.

[0062] Server 104 can be an independent physical server, a cloud server, or a service node in a blockchain system, where the service nodes in the blockchain system form a peer-to-peer network.

[0063] In addition, server 104 can also be a server cluster consisting of multiple physical servers, which can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0064] Terminal 102 and server 104 can be connected via Bluetooth, USB (Universal Serial Bus) or network, etc., and this application does not impose any restrictions.

[0065] In some embodiments, such as Figure 4As shown, a self-evolving method for a distributed rule engine driven by AI Agent is provided. This method can be applied to terminals to control devices on both the cloud and edge sides. The terminal can be an intelligent product such as a computer. Specifically, the method includes the following steps: Step S100: After the cloud-based industry AI Agent service receives the business objective, it acquires real-time status data, historical data, and rule execution logs.

[0066] Step S200: Based on the analysis and reinforcement learning of the digital twin model, generate rule processes through a rule generator; Step S300: The rule flow is sent to the endpoint, and the rule flow is executed locally in a closed loop based on the endpoint's rule execution engine. In one implementation, such as... Figure 5 As shown, step S100 includes the following steps: Step S101: Obtain the business objective in natural language form as input; Step S102: Deconstruct the core requirements of the business objective through the semantic understanding module, extract key indicators, and associate them with the business knowledge graph of the corresponding industry; Step S103: Real-time status data, historical data, and rule execution logs uploaded by the receiving end.

[0067] Specifically, based on the method of this embodiment, after receiving the business objective, the cloud-based industry AI Agent service will understand and learn the business objective and conduct multi-dimensional data correlation analysis. Specifically, operations and maintenance personnel can input business objectives in natural language form (such as "reduce industrial robot downtime" or "optimize the energy consumption data collection frequency of smart meters") through the cloud-based management platform. After receiving the objective, the industry AI Agent service uses a semantic understanding module to break down the core requirements, extract key indicators (such as downtime, collection frequency, and energy consumption data), and associate them with the corresponding industry's business knowledge graph (such as common fault types of industrial robots and communication protocol specifications of meters). The industry AI Agent service calls the historical data interfaces of the digital twin and data lake to associate historical equipment operation data, fault records, and rule execution logs related to the objective (such as changes in equipment failure rate after the implementation of a certain type of rule). Then, the data analysis capabilities of the large model are used to mine the inherent correlation between the business objective and the data indicators. For example, for the objective of "reducing robot downtime," the analysis concludes that "when the robot joint temperature exceeds 60℃ and the vibration frequency is greater than 5Hz, the failure risk increases by 80%."

[0068] For example, in a smart tunnel management scenario, the tunnel contains numerous sensors (such as those for detecting CO concentration, wind speed, and traffic flow) and actuators (such as fans, lights, and traffic signals), managed locally by multiple edge controllers (equipped with an enhanced rule engine). A rule evolution platform is deployed in the cloud, along with a tunnel traffic management AI Agent service. In practical applications, operations personnel set business objectives, such as ensuring air quality within the tunnel meets standards while minimizing total energy consumption. Once the cloud-based industry AI Agent service receives these business objectives, it begins analyzing historical data from the digital twin and data lake. Using the aforementioned smart tunnel management scenario as an example, the industry AI Agent service can analyze the strong correlation between "traffic flow" and "CO concentration." Next, the rule generator automatically outputs rule flow A. For example, rule flow A might be: "Every minute, check all CO sensor values; if any value exceeds 150 ppm, set the corresponding area's fan speed to 80%." This rule flow A is distributed to all edge devices via the rule distribution service, and the edge device's rule execution engine performs local closed-loop execution of rule flow A. Furthermore, the industry AI Agent service can analyze real-time status data and rule execution logs reported by edge devices in the digital twin and data lake, and then make autonomous decisions to determine which edge devices to distribute the rule flow for execution.

[0069] Furthermore, in this embodiment, the cloud side receives new rule execution logs or new rule generation conditions. Based on these new logs or conditions, the rule optimizer generates a new rule flow and distributes it to the edge side. For example, taking the aforementioned smart tunnel management scenario, after one week of operation, the industry AI Agent service analyzes the newly reported execution logs and energy consumption data from the edge side. It discovers that while rule flow A ensures safety, during low traffic hours at midnight, the fans still start at high power due to occasional concentration fluctuations, resulting in energy waste. The industry AI Agent service learns and optimizes, generating rule flow B: "Based on rule A, add a condition: only execute at 80% speed when 'current traffic flow is greater than 20 vehicles / minute'; otherwise, execute at 40% speed." The new rule flow is automatically distributed to the edge side, achieving a significant reduction in energy consumption.

[0070] In one implementation, such as Figure 6 As shown in the figure, the technical solution of this embodiment also includes the following steps: Step S1: During the execution of the rule process on the client side, when the Agent call node is executed, the context data is packaged to the industry AI Agent service on the cloud side. Step S2: After receiving the context data, the cloud-based industry AI Agent service makes a global decision based on the analysis and reinforcement learning of the digital twin model and generates the decision result.

[0071] For example, taking the aforementioned smart tunnel management scenario as an example, one day, a traffic jam occurs in the tunnel with extremely high traffic volume, and multiple CO sensors simultaneously trigger alarms. Rule B on the edge side is frequently triggered, and all fans are running at high power, but the CO concentration continues to rise slowly. At this point, a more advanced emergency plan rule preset by the maintenance personnel is triggered, which includes an Agent calling node. This Agent calling node immediately packages the context data (i.e., real-time data from all sensors in the tunnel) and initiates decision consultation with the industry AI Agent service on the cloud side. After receiving the data, the industry AI Agent service immediately analyzes the digital twin model. It not only sees the CO concentration but also the increasing traffic volume at the entrance, and infers from the model that the congestion will peak in 15 minutes. It makes a globally optimal decision that goes beyond simple threshold judgment: not only should all fans be increased to 100% power, but the traffic control system should also be activated to immediately shut off the traffic lights at the tunnel entrance and push detour information to the navigation application. This complex decision is returned to the edge side, where the rule execution engine continues to execute subsequent actions, such as controlling the traffic lights. Therefore, this embodiment not only achieves a complete intelligent upgrade, from simple threshold alarms to adaptive energy saving based on data learning, and then to AI-assisted global optimal decision-making in emergency situations, but also perfectly achieves the core objective of the invention and solves all the pain points in the background technology.

[0072] In one implementation, after the rule generator generates the rule flow, this implementation can also push the rule flow to the expert review node for review by a designated operations and maintenance expert. Specifically, for example... Figure 7 As shown, the specific steps include the following: Step S11: Perform compliance verification on the rule process and push the rule process that passes the verification to the expert review node; Step S12: Activate the intelligent pre-review mode of the expert review node, and pre-review the rule process that has passed the verification based on the intelligent pre-review mode; Step S13: Activate the manual final review mode of the expert review node, and push the rules and processes that have passed the pre-review to the designated operation and maintenance experts for review.

[0073] Specifically, the cloud-side rule evolution platform in this embodiment also includes an expert review node. This node pushes the rule process generated by the rule generator to a designated operations expert for review, and then forwards the expert's review results to the rule optimizer. Experts can "like," "reject," or suggest modifications to the rule process. Approved rule processes directly enter the rule distribution queue; rules that fail review (e.g., those with logical conflicts or potential for device malfunctions) are annotated by experts with the problem type (e.g., "threshold settings are too low, leading to false alarms") and fed back to the rule optimizer. The rule optimizer adjusts the structured description file based on expert feedback, re-triggering the rule generation process, forming a closed loop of "generation-review-optimization." This feedback from human experts serves as a strong signal, incorporated into the reinforcement learning reward function of the industry AI Agent service, thereby guiding the industry AI Agent service to learn the optimal strategy that aligns with human expectations and industry standards more quickly and safely. This solves the "black box" and "uncontrollable" problems of AI decision-making, making it particularly suitable for safety-critical fields (such as healthcare and gas), ensuring that the evolution of AI is always under the supervision of human experts, and increasing the credibility of the system.

[0074] This embodiment's expert review node includes an intelligent pre-review mode and a manual final review mode. In practical applications, after the rule flow is generated, the industry AI Agent service first performs compliance verification on the rule flow, such as verifying whether it conforms to industry security standards and whether there are logical conflicts. The verified rule flow is then pushed to the expert review node. Next, the intelligent pre-review mode of the expert review node is activated, and the verified rule flow is pre-reviewed based on this mode. Then, the manual final review mode of the expert review node is activated, and the rule flow that has passed the pre-review is pushed to the designated operations and maintenance expert for review. Experts can view the rule details, decision basis, and simulated execution methods through a visual interface, and the review results are synchronized to the rule optimizer in real time. Furthermore, for rule flows of core equipment (such as nuclear power plant reactor control equipment), multi-level expert review can be mandated to ensure zero risk.

[0075] Furthermore, in one implementation, this embodiment, when issuing the rule process, such as... Figure 8 As shown, it includes the following steps: Step S01: Obtain the tag information of the rule process, wherein the tag information is set by the rule generator when generating the rule process; Step S02: Based on the tag information, determine the push mode, and push the rule process to the terminal based on the push mode. The push mode includes: on-demand push mode, batch push mode, and incremental update mode.

[0076] Specifically, in practical applications, this embodiment can obtain the tag information of the rule process when issuing the rule process. The tag information is set by the rule generator when generating the rule process, and the tag information can reflect the push method and target device of the rule process. Therefore, based on the tag information, the push mode can be determined. The push mode includes: on-demand push mode, batch push mode, and incremental update mode, and then the rule process is pushed to the terminal based on the push mode.

[0077] This embodiment's rule distribution service can push rules to corresponding edge devices via message queues based on the applicable scope of the rule process (e.g., specifying robots in a certain workshop or smart meters in a certain area). A differential update mechanism is used during distribution, pushing only the changed parts of the rules (e.g., updating only threshold parameters instead of the entire rule) to reduce network transmission overhead. After receiving the rules, the edge device verifies their completeness and legality through a rule verification module. If verification passes, the old local rules are overwritten; if verification fails, an exception is reported to the cloud. After loading the rule process, the edge's rule execution engine initiates a local closed-loop execution mechanism. During execution, if the rule conditions are met (e.g., both robot temperature and vibration frequency reach thresholds), local execution actions (e.g., initiating a cooling program) are directly triggered without waiting for cloud instructions, achieving millisecond-level response. Simultaneously, execution logs are recorded in real-time to the local cache and reported to the cloud data lake at preset intervals.

[0078] In addition, in other implementation methods, the cloud-based digital twin and data lake periodically perform batch analysis on the rule execution logs and equipment operation data reported by the edge devices to evaluate the execution effectiveness of the rule processes. Analysis metrics include rule trigger accuracy, fault warning timeliness, and percentage improvement in equipment operating efficiency. For example, by comparing robot downtime before and after rule implementation, the actual benefits of the rules can be calculated. Figure 9 As shown, after the rule process is completed on the endpoint, this embodiment further includes the following steps: Step S10: The receiving end analyzes the execution result of the rule process and determines whether the execution result meets the expected requirements. Step S20: If the execution result does not meet the expected requirements, the industry AI Agent service receives the execution log of the rule process, regenerates a new rule process based on the rule generator, and sends it to the end side.

[0079] In this embodiment, the cloud-based industry AI Agent service analyzes whether the execution results meet the expected requirements based on the following three scenarios: 1. The execution effect does not meet expectations (e.g., the early warning accuracy rate is only 85%, lower than the target of 95%); 2. The business scenario changes (e.g., increased equipment load leads to the need to adjust the fault threshold); 3. New business requirements emerge (e.g., a new "robot energy consumption optimization" target is added). If the execution results do not meet the expected requirements, the industry AI Agent service receives the execution logs of the rule process and generates corresponding optimization decision information for different scenarios. The optimization decision information completes the rule update according to the "rule optimization-generation-review-distribution" process, and regenerates a new rule process based on the rule generator, and distributes it to the edge. After the new rule process is pushed to the edge, it overwrites the old rules and starts a new round of execution and data collection process. Through this closed-loop mechanism of "data collection-effect analysis-rule optimization-execution feedback", the system realizes the autonomous evolution of rules and continuously improves its adaptability to business scenarios.

[0080] Reference Figure 10 As shown, Figure 10 The collaborative working mechanism of the two core closed loops of this invention is demonstrated in the paper. Figure 10 The upper part is the "self-evolutionary learning closed loop", which includes the following steps: 1. The edge continuously reports the operation to the digital twin and data lake; 2. The AI ​​Agent regularly analyzes the digital twin model; 3. Discover optimization space or learn new patterns; 4. Generate or optimize business rules; 5. The rules are issued to the edge for execution, thereby achieving continuous learning and self-evolution. Figure 10 The lower half is the real-time Agent invocation decision-making closed loop. When the endpoint encounters a scenario where it cannot make a decision during execution, it includes the following steps: 1. The endpoint rule execution encounters an Agent invocation node; 2. The on-site context data is packaged and reported; 3. The AI ​​Agent performs real-time inference and decision-making; 4. It returns an immediate execution instruction or temporary rule; 5. The endpoint receives and continues to execute the action, achieving a millisecond-level closed-loop response. The two closed loops work together, taking into account both long-term evolution and real-time intelligence. Further, as... Figure 11 As shown, Figure 11 The diagram illustrates the overall overview of the invention, combining an architecture layer (cloud-edge-device) with a collaborative process of dual closed loops (long-term learning and real-time decision-making). The cloud-side industry AI Agent service continuously learns and optimizes rules from digital twins and data lakes; the edge-side rule execution engine efficiently executes rules and can trigger immediate Agent decisions, forming a dual-layer intelligent closed loop of "self-learning + self-adaptation." The system maintains continuous iteration between dynamic data and rules, enabling the overall network or device system to continuously optimize and self-evolve.

[0081] In summary, this invention provides a self-evolving distributed rule engine system driven by AI Agents. It creatively combines the complex cognitive reasoning capabilities of AI Agents (in the cloud / edge) with the simple and reliable execution capabilities of rule engines (on the edge), leveraging their strengths to address the contradiction between high edge computing power requirements and insufficient decision reliability in AI deployments in IoT scenarios. Furthermore, this invention proposes a self-evolving method for distributed rule engines driven by AI Agents. Through a complete closed loop of "data reporting → model learning → rule generation / optimization → cloud-side distribution → edge-side execution," it enables business logic to possess self-iterative and continuous optimization capabilities, achieving a qualitative leap from "low-code" to "adaptive." In addition, this invention creates a novel Agent invocation node mechanism, introducing the ability to seek external decision-making support from cloud-based AI into traditional rule engines. This allows even simple rule processes on the edge to obtain global, in-depth, and professional decision support from AI Agents at critical moments, greatly enhancing the system's intelligence ceiling.

[0082] In some embodiments, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 12 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a self-evolving method for a distributed rule engine based on an AI Agent. The display unit of the computer device is used to form a visually visible image. It can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0083] Those skilled in the art will understand that Figure 12The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0084] In some embodiments, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps: Once the cloud-based industry AI Agent service receives the business objective, it acquires real-time status data, historical data, and rule execution logs. Based on analysis and reinforcement learning of digital twin models, a rule generator is used to generate rule processes. The rule process is sent to the endpoint, and the rule process is executed locally in a closed loop based on the endpoint's rule execution engine.

[0085] In practical applications, after the cloud-based industry AI Agent service receives the business objective, it acquires real-time status data, historical data, and rule execution logs, including: Obtain the business objective in the natural language form of the input; The semantic understanding module breaks down the core requirements of the business objectives, extracts key indicators, and associates them with the corresponding industry's business knowledge graph. The receiving end uploads real-time status data, historical data, and rule execution logs.

[0086] In practical applications, when one or more processors 100 execute computer program 102, the following can also be achieved: When the cloud receives new rule execution logs or new rule generation conditions, the rule optimizer generates new rule processes based on these logs and sends them to the edge.

[0087] In practical applications, when one or more processors 100 execute computer program 102, the following can also be achieved: During the execution of the rule process on the client side, when the Agent call node is executed, the context data is packaged and sent to the industry AI Agent service on the cloud side. After receiving the context data, the cloud-based industry AI Agent service makes global decisions based on the analysis and reinforcement learning of the digital twin model and generates decision results.

[0088] In practical applications, when one or more processors 100 execute computer program 102, the following can also be achieved: After generating the rule process based on the rule generator, the cloud side pushes the rule process to the expert review node, reviews it based on the designated operation and maintenance experts, and transfers the review results of the operation and maintenance experts to the rule optimizer. In practical applications, when one or more processors 100 execute computer program 102, the following can also be achieved: The edge device is encapsulated into a lightweight micro-agent to execute the rule process, and the micro-agent can make autonomous decisions and make dynamic adjustments locally.

[0089] In practical applications, when one or more processors 100 execute computer program 102, the following can also be achieved: The receiving end analyzes the execution result of the rule process to determine whether the execution result meets the expected requirements. If the execution result does not meet the expected requirements, the industry AI Agent service receives the execution log of the rule process, regenerates a new rule process based on the rule generator, and sends it to the edge.

[0090] It should be noted that the user information (including but not limited to user device function information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0091] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0092] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0093] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A self-evolving distributed rule engine system based on AI Agent-driven architecture, characterized in that, The system includes: a cloud-based rule evolution platform and an edge-based rule execution engine; The cloud-based rule evolution platform includes: Industry AI Agent services are intelligent agents driven by one or more large industry models, used to receive and understand business objectives; Digital twins and data lakes are used to aggregate real-time status data, historical data, and rule execution logs reported by all edge devices, and to build a digital twin model that is synchronized with the physical world in real time. A rule generator is used to transform the decision information output by the industry AI Agent service into a rule process that can be executed by the edge device. The rule distribution service is used to push all rule processes generated by the rule generator to the specified devices on the edge. The rule execution engine on the endpoint is used to perform local closed-loop execution of the rule process issued from the cloud.

2. The AI ​​Agent-driven distributed rule engine self-evolution system according to claim 1, characterized in that, The rule execution engine on the endpoint also includes: The Agent call node is used to package context data into the cloud-based industry AI Agent service and initiate decision consultation to the industry AI Agent service.

3. The AI ​​Agent-driven distributed rule engine self-evolution system according to claim 2, characterized in that, The rule execution engine on the endpoint also includes: The data reporting service is used to report real-time status data, historical data, and rule execution logs of the device on the edge to the digital twin and data lake on the cloud side.

4. The AI ​​Agent-driven distributed rule engine self-evolution system according to claim 1, characterized in that, The cloud-based rule evolution platform also includes: A rule optimizer is used to transform the decision information into a structured description file that conforms to the rule engine format. The description file is either a newly generated rule flow or a rule flow obtained by adjusting the parameters of an existing rule flow.

5. The AI ​​Agent-driven distributed rule engine self-evolution system according to claim 4, characterized in that, The cloud-based rule evolution platform also includes: The expert review node is used to push the rule process to a designated operation and maintenance expert for review after the rule generator generates the rule process, and to transfer the review result of the operation and maintenance expert to the rule optimizer.

6. The AI ​​Agent-driven distributed rule engine self-evolution system according to claim 5, characterized in that, The expert review nodes include an intelligent pre-review mode and a manual final review mode.

7. The AI ​​Agent-driven distributed rule engine self-evolution system according to claim 1, characterized in that, The rule distribution service includes on-demand push mode, batch push mode, and incremental update mode.

8. The AI ​​Agent-driven distributed rule engine self-evolution system according to claim 1, characterized in that, The edge device is encapsulated into a lightweight micro-agent to execute rule processes, and the micro-agent can make autonomous decisions and make dynamic adjustments locally.

9. The AI ​​Agent-driven distributed rule engine self-evolution system according to claim 1, characterized in that, The industry AI Agent service can also be deployed on edge computing nodes.

10. A self-evolutionary method for a distributed rule engine driven by AI Agent, characterized in that, The AI ​​Agent-driven distributed rule engine self-evolution method is applied to the AI ​​Agent-driven distributed rule engine self-evolution system according to any one of claims 1-9, and the method includes: Once the cloud-based industry AI Agent service receives the business objective, it acquires real-time status data, historical data, and rule execution logs. Based on analysis and reinforcement learning of digital twin models, a rule generator is used to generate rule processes. The rule process is sent to the endpoint, and the rule process is executed locally in a closed loop based on the endpoint's rule execution engine.

11. The self-evolutionary method for a distributed rule engine driven by AI Agent according to claim 10, characterized in that, Once the cloud-based industry AI Agent service receives the business objective, it acquires real-time status data, historical data, and rule execution logs, including: Obtain the business objective in the natural language form of the input; The semantic understanding module breaks down the core requirements of the business objectives, extracts key indicators, and associates them with the corresponding industry's business knowledge graph. The receiving end uploads real-time status data, historical data, and rule execution logs.

12. The self-evolutionary method for a distributed rule engine driven by AI Agent according to claim 10, characterized in that, The method further includes: When the cloud receives new rule execution logs or new rule generation conditions, the rule optimizer generates new rule processes based on these logs and sends them to the edge.

13. The self-evolutionary method for a distributed rule engine driven by AI Agent according to claim 10, characterized in that, The method further includes: During the execution of the rule process on the client side, when the Agent call node is executed, the context data is packaged and sent to the industry AI Agent service on the cloud side. After receiving the context data, the cloud-based industry AI Agent service makes global decisions based on the analysis and reinforcement learning of the digital twin model and generates decision results.

14. The self-evolutionary method for a distributed rule engine driven by AI Agent according to claim 10, characterized in that, The method further includes: After generating the rule process based on the rule generator, the cloud side pushes the rule process to the expert review node, reviews it based on the designated operation and maintenance experts, and transfers the review results of the operation and maintenance experts to the rule optimizer.

15. The self-evolutionary method for a distributed rule engine driven by AI Agent according to claim 14, characterized in that, After generating a rule flow based on the rule generator, the cloud side pushes the rule flow to the expert review node for review by designated operations and maintenance experts, including: The rules and processes are verified for compliance, and those that pass the verification are pushed to the expert review node. The intelligent pre-review mode of the expert review node is activated, and the rule process that has passed the verification is pre-reviewed based on the intelligent pre-review mode; The manual final review mode of the expert review node is initiated, and the rules and processes that have passed the pre-review are pushed to the designated operation and maintenance experts for review.

16. The self-evolutionary method for a distributed rule engine driven by AI Agent according to claim 10, characterized in that, The rule process is distributed to the endpoint, including: Obtain the tag information of the rule process, which is set by the rule generator when generating the rule process; Based on the tag information, a push mode is determined, and the rule process is pushed to the terminal based on the push mode. The push mode includes: on-demand push mode, batch push mode, and incremental update mode.

17. The self-evolutionary method for a distributed rule engine driven by AI Agent according to claim 10, characterized in that, The method further includes: The edge device is encapsulated into a lightweight micro-agent to execute the rule process, and the micro-agent can make autonomous decisions and make dynamic adjustments locally.

18. The self-evolutionary method for a distributed rule engine driven by AI Agent according to claim 10, characterized in that, The method further includes: The receiving end analyzes the execution result of the rule process to determine whether the execution result meets the expected requirements. If the execution result does not meet the expected requirements, the industry AI Agent service receives the execution log of the rule process, regenerates a new rule process based on the rule generator, and sends it to the edge.

19. A computer device, characterized in that, The computer device includes a memory, a processor, and an AI Agent-driven distributed rule engine self-evolution program stored in the memory and executable on the processor. When the processor executes the AI ​​Agent-driven distributed rule engine self-evolution program, it implements the steps of the AI ​​Agent-driven distributed rule engine self-evolution method as described in any one of claims 10-18.

20. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an AI Agent-driven distributed rule engine self-evolution program, which implements the steps of the AI ​​Agent-driven distributed rule engine self-evolution method as described in any one of claims 10-18 on the computer-readable storage medium.

Images