CN-122018306-A - AI Agent-driven distributed rule engine self-evolution system, method, computer equipment and medium

Abstract

The invention discloses an AI Agent-driven distributed rule engine self-evolution system, an AI Agent-driven distributed rule engine self-evolution method, computer equipment and a medium. The industry AI Agent service is an Agent driven by one or more industry big models and used for receiving and understanding business targets, the rule generator is used for converting decision information output by the industry AI Agent service into a rule flow, the rule distribution service is used for pushing the rule flow to equipment appointed by an end side, and the rule execution engine of the end side is used for carrying out local closed-loop execution on the rule flow. The invention can change the complex operation and maintenance mode from manual writing and maintenance rules to automatic AI learning and optimization, greatly liberates manpower and can achieve the optimization effect of far exceeding human expert.

Inventors

• XIA YAN

Assignees

• 深圳开鸿数字产业发展有限公司

Dates

Publication Date

20260512

Application Date

20251210

Claims (20)

1. 1. A distributed rule engine self-evolution system based on AI Agent driving is characterized by comprising a rule evolution platform at a cloud side and a rule execution engine at an end side; Wherein, the rule evolution platform of cloud side includes: industry AI Agent service, which is an Agent driven by one or more industry big models, for receiving and understanding business objectives; The digital twin and data lake is used for collecting real-time state data, historical data and rule execution logs reported by all end-side equipment and constructing a digital twin model which is synchronous with the physical world in real time; the rule generator is used for converting decision information output by the industry AI Agent service into a rule flow executable by equipment at the end side; a rule distribution service for pushing all rule flows generated by the rule generator to the equipment designated by the end side; and the rule execution engine at the end side is used for performing local closed-loop execution on the rule flow issued by the cloud side.
2. 2. The AI Agent-driven based distributed rules engine self-evolution system of claim 1, wherein the end-side rules execution engine further comprises: And the Agent calling node is used for packaging the context data to the cloud-side industry AI Agent service and initiating decision consultation to the industry AI Agent service.
3. 3. The AI Agent-driven based distributed rules engine self-evolution system of claim 2, wherein the end-side rules execution engine further comprises: And the data reporting service is used for reporting the real-time state data, the historical data and the rule execution log of the equipment at the end side to the digital twin and the data lake at the cloud side.
4. 4. The AI Agent-driven based distributed rule engine self-evolution system of claim 1, wherein the cloud-side rule evolution platform further comprises: And the rule optimizer is used for converting the decision information into a structured description file conforming to a rule engine format, wherein the description file is a regenerated new rule flow or a rule flow obtained by carrying out parameter adjustment on the existing rule flow.
5. 5. The AI Agent-driven based distributed rules engine self-evolution system of claim 4, wherein the cloud-side rules evolution platform further comprises: And the expert auditing node is used for pushing the rule flow to a designated operation and maintenance expert for auditing after the rule flow is generated by the rule generator, and transferring the auditing result of the operation and maintenance expert to the rule optimizer.
6. 6. The AI Agent-driven based distributed rules engine self-evolution system of claim 5, wherein the expert audit node includes an intelligent pre-audit mode and a manual final audit mode.
7. 7. The AI Agent-driven based distributed rules engine self-evolution system of claim 1, wherein the rule distribution service includes an on-demand push mode, a batch push mode, and an incremental update mode.
8. 8. The AI Agent-driven distributed rule engine self-evolution system of claim 1, wherein the end-side is packaged as a lightweight micro-Agent to execute rule flow, and wherein the micro-Agent is locally self-decision-making and dynamic tuning.
9. 9. The AI Agent-driven distributed rules engine self-evolution system of claim 1, wherein the industry AI Agent service is further deployable at an edge computing node.
10. 10. An AI Agent-driven based self-evolution method of a distributed rule engine, wherein the AI Agent-driven based self-evolution method of a distributed rule engine is applied to the AI Agent-driven based self-evolution system of any one of claims 1 to 9, and the method comprises: when the cloud-side industry AI Agent service receives a business target, acquiring real-time state data, historical data and rule execution logs; based on analysis and reinforcement learning of the digital twin model, generating a rule flow through a rule generator; And issuing the rule flow to an end side, and performing local closed-loop execution on the rule flow based on a rule execution engine of the end side.
11. 11. The AI Agent-driven self-evolution method of a distributed rule engine of claim 10, wherein when the cloud-side industry AI Agent service receives the business objective, obtaining real-time status data, historical data, and rule execution log comprises: Acquiring an input business target in a natural language form; disassembling the core requirements of the business targets through a semantic understanding module, extracting key indexes, and associating business knowledge maps of corresponding industries; And the receiving end side uploads the real-time state data, the historical data and the rule execution log.
12. 12. The AI Agent-driven based distributed rules engine self-evolution method of claim 10, further comprising: Yun Cejie receives the new rule execution log or the new rule generation condition, generates a new rule flow based on the new rule execution log or the new rule generation condition through the rule optimizer, and issues the new rule flow to the end side.
13. 13. The AI Agent-driven based distributed rules engine self-evolution method of claim 10, further comprising: in the process of executing the rule flow, the terminal side packages the context data to the industry AI Agent service of the cloud side when executing the rule flow to the Agent calling node; and after receiving the context data, the cloud-side industry AI Agent service makes global decisions based on analysis and reinforcement learning of the digital twin model, and generates decision results.
14. 14. The AI Agent-driven based distributed rules engine self-evolution method of claim 10, further comprising: After the cloud side generates a rule flow based on the rule generator, pushing the rule flow to an expert auditing node, auditing based on a designated operation and maintenance expert, and transferring the auditing result of the operation and maintenance expert to a rule optimizer.
15. 15. The AI Agent-driven distributed rule engine self-evolution method of claim 14, wherein the cloud side pushes a rule flow to an expert auditing node after generating the rule flow based on a rule generator, and auditing based on a specified operation and maintenance expert comprises: Compliance verification is carried out on the rule flow, and the rule flow passing the verification is pushed to an expert auditing node; starting an intelligent pre-auditing mode of the expert auditing node, and pre-auditing rule flows passing verification based on the intelligent pre-auditing mode; And starting a manual final auditing mode of the expert auditing node, and pushing the rule flow passing the pre-auditing to a designated operation and maintenance expert for auditing.
16. 16. The AI Agent-driven distributed rule engine self-evolution method of claim 10, wherein issuing the rule flow to an end side comprises: Acquiring label information of the rule flow, wherein the label information is set when a rule generator generates the rule flow; and determining a pushing mode based on the label information, and pushing the rule flow to an end side based on the pushing mode, wherein the pushing mode comprises an on-demand pushing mode, a batch pushing mode and an incremental updating mode.
17. 17. The AI Agent-driven based distributed rules engine self-evolution method of claim 10, further comprising: And packaging the end side into a lightweight micro-agent to execute the rule flow, wherein the micro-agent can perform autonomous decision making and dynamic adjustment locally.
18. 18. The AI Agent-driven based distributed rules engine self-evolution method of claim 10, further comprising: The receiving end side analyzes the execution result of the rule flow and judges whether the execution result meets the expected requirement or not; if the execution result does not meet the expected requirement, the industry AI Agent service receives the execution log of the rule flow, and generates a new rule flow again based on a rule generator and sends the new rule flow to the end side.
19. 19. A computer device comprising a memory, a processor and an AI Agent-driven based distributed rules engine self-evolution program stored in the memory and executable on the processor, the processor implementing the AI Agent-driven based distributed rules engine self-evolution method steps of any one of claims 10-18 when executing the AI Agent-driven based distributed rules engine self-evolution program.
20. 20. A computer readable storage medium, wherein the computer readable storage medium has stored thereon an AI Agent-driven based distributed rule engine self-evolution program, the AI Agent-driven based distributed rule engine self-evolution program implementing the AI Agent-driven based distributed rule engine self-evolution method steps of any one of claims 10-18 on the computer readable storage medium.

Description

AI Agent-driven distributed rule engine self-evolution system, method, computer equipment and medium Technical Field The invention relates to the technical field of intersection of artificial intelligence and distributed control, in particular to a distributed rule engine self-evolution system, a method, computer equipment and a medium based on AI Agent driving. Background Current internet of things operating systems provide a visual super rule engine. The developer or operation and maintenance personnel can construct business logic (scene linkage rule) in a low-code or code-free mode by dragging functional nodes (such as sensor triggering, equipment control, delay, logic judgment and the like), so that the application development threshold is greatly reduced. For example, a rule of "when the smoke sensor value exceeds the threshold value, the shower is immediately turned on and the exhaust fan is started" can be easily created. In the prior art, at least the following defects exist: (1) Rules laid out manually are static. Once deployed, its trigger conditions and execution logic are fixed. In the face of complex, dynamically changing real world scenarios (e.g., tidal changes in traffic flow, process adjustments in industrial production lines), these static rules cannot be adaptively adjusted, are prone to becoming inefficient and even erroneous, and require significant manpower to perform continuous maintenance and optimization. (2) The decision making capability of the rule engine is limited by preset simple logic. It cannot make complex, global information-based reasoning and prediction. For example, in intelligent building energy consumption management, the simple rule of 'air conditioning is started when the temperature is higher than 26 degrees' is far less than an expert decision which can comprehensively consider outdoor weather, future electricity prices and personnel distribution in a building to achieve energy conservation and high efficiency. (2) The optimal rules of complex scenes often settle in the mind of a few senior industry specialists. This implicit knowledge is difficult to effectively scale up into the system. When an expert leaves or a scene changes, optimization of rules becomes a difficult problem. Accordingly, there is a need in the art for improvement. Disclosure of Invention Aiming at the defects in the prior art, the invention provides a distributed rule engine self-evolution method, a system, computer equipment and a medium based on AI Agent driving, and the technical scheme adopted by the invention is as follows: In a first aspect, the invention provides an AI Agent-driven distributed rule engine self-evolution system, which comprises a cloud-side rule evolution platform and an end-side rule execution engine; Wherein, the rule evolution platform of cloud side includes: industry AI Agent service, which is an Agent driven by one or more industry big models, for receiving and understanding business objectives; The digital twin and data lake is used for collecting real-time state data, historical data and rule execution logs reported by all end-side equipment and constructing a digital twin model which is synchronous with the physical world in real time; the rule generator is used for converting decision information output by the industry AI Agent service into a rule flow executable by equipment at the end side; a rule distribution service for pushing all rule flows generated by the rule generator to the equipment designated by the end side; and the rule execution engine at the end side is used for performing local closed-loop execution on the rule flow issued by the cloud side. In one implementation, the rule execution engine of the end side further includes: And the Agent calling node is used for packaging the context data to the cloud-side industry AI Agent service and initiating decision consultation to the industry AI Agent service. In one implementation, the rule execution engine of the end side further includes: And the data reporting service is used for reporting the real-time state data, the historical data and the rule execution log of the equipment at the end side to the digital twin and the data lake at the cloud side. In one implementation, the rule evolution platform of the cloud side further includes: the rule optimizer is used for converting the decision information into a structured description file conforming to a rule engine format, wherein the description file is a regenerated new rule flow or a rule flow obtained by carrying out parameter adjustment on the existing rule flow. In one implementation, the rule evolution platform of the cloud side further includes: And the expert auditing node is used for pushing the rule flow to a designated operation and maintenance expert for auditing after the rule flow is generated by the rule generator, and transferring the auditing result of the operation and maintenance expert to the rule optimizer. In one implementation,