摘要: 针对分布式电力系统 “节点多、地域广、接入碎片化” 的特征及现有安全防护技术存在的风险执行方式单一、分级缺失、业务连续性保障不足、响应滞后等问题，提出一种多维度协同的高可靠性可自愈风险决策引擎。该引擎整合了决策矩阵构建、策略生成与冲突检测、指令封装与发送三大核心模块，并辅以闭环反馈优化机制。通过设备历史行为评分、安全规则匹配程度等五维输入向量构建动态决策矩阵，实现观察、限速、强化、隔离四级风险分级响应。实验结果表明，该引擎策略匹配准确率提升40% 以上，业务中断风险降低75%，运维效率提升90%，可满足分布式电力系统秒级风险响应需求，为智能电网安全稳定运行提供技术支撑。

Abstract: Addressing the characteristics of distributed power systems (DPS)–characterized by numerous nodes, wide geographical reach, and fragmented access–and the shortcomings of existing security technologies such as single risk execution methods, lack of hierarchical classification, insufficient business continuity assurance, and delayed response, this paper proposes a multi-dimensional, collaborative, highly reliable, and self-healing risk decision engine. This engine integrates three core modules: decision matrix construction, strategy generation and conflict detection, and instruction encapsulation and transmission, supplemented by a closed-loop feedback optimization mechanism. A dynamic decision matrix is constructed using five-dimensional input vectors, including device historical behavior scores and security rule matching degrees, to achieve four levels of risk-level response: observation, rate limiting, reinforcement, and isolation. Experimental results show that the engine improves strategy matching accuracy by over 40%, reduces business interruption risk by 75%, and improves operation and maintenance efficiency by 90%, meeting the second-level risk response requirements of distributed power systems and providing technical support for the safe and stable operation of smart grids.