A circuit breaker breaking strategy optimization method and system based on reinforcement learning

By using a reinforcement learning-based circuit breaker tripping strategy optimization method, an adaptive tripping control strategy is generated through a multi-physics coupling model and multi-agent collaborative decision-making. This solves the multi-objective conflict and synchronization problems of circuit breakers under complex fault conditions, and improves the safety and stability of the tripping process.

CN122194688APending Publication Date: 2026-06-12CHINA POWER HUARUI TECH CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Applications(China)
Current Assignee / Owner
CHINA POWER HUARUI TECH CO LTD
Filing Date
2026-04-23
Publication Date
2026-06-12

AI Technical Summary

Technical Problem

Existing circuit breaker breaking strategies are ill-suited to the complex characteristics of dynamic coupling of multiple physical fields, and cannot adjust control parameters in real time. This leads to multi-objective conflicts and increased equipment losses. Furthermore, it is difficult to achieve precise synchronization at the microsecond level, and there are risks of current backflow and operational overvoltage.

Method used

By employing a reinforcement learning-based approach, a multi-physics coupling model of the circuit breaker is constructed by collecting multi-physics operating parameters. This model enables state observation and multi-agent collaborative decision-making, generating collaborative control commands. Furthermore, online parameter fine-tuning is performed through real-time monitoring to form an adaptive disconnection control strategy.

Benefits of technology

It achieves global collaborative optimization of arc extinction speed, insulation recovery efficiency and equipment loss, improves the safety, stability and intelligent control level of the breaking process, adapts to complex and variable fault conditions, and reduces the trial and error cost of online learning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122194688A_ABST
    Figure CN122194688A_ABST
Patent Text Reader

Abstract

The application provides a circuit breaker breaking strategy optimization method and system based on reinforcement learning, which comprises the following steps: determining the comprehensive state characteristics of the current breaking stage according to the operating parameter data in the circuit breaker breaking process; performing multi-agent collaborative decision on the comprehensive state characteristics to obtain multiple collaborative control instructions; performing simulation deduction and dynamic punishment on all the collaborative control instructions based on the comprehensive state characteristics to obtain the immediate reward signal of the current control strategy; determining the pre-trained strategy model responding to different fault working conditions through the immediate reward signal; deploying the pre-trained strategy model in the edge control device of the circuit breaker, and performing online parameter fine-tuning on the pre-trained strategy model through the running data stream to obtain the adaptive breaking control strategy matching the actual running working condition of the circuit breaker. By adopting the scheme of the application, an intelligent breaking strategy capable of dynamically sensing the states of multiple physical fields to collaboratively optimize the multi-objective control requirements and adapt to complex and variable fault working conditions can be constructed.
Need to check novelty before this filing date? Find Prior Art