Intelligent Optimization Algorithm for Power Grid Fault Decision Based on Multi-Source Data Fusion

Power grid fault recovery scheduling often faces challenges such as strong data heterogeneity, low fault identification efficiency, and difficulty in coordinating multiple conflicting objectives. To enhance the intelligence and practicality of fault handling, a power grid fault decision-making algorithm is proposed that integrates multi-source data modeling with intelligent optimization. On the basis of unified fusion of structured and unstructured data, the algorithm incorporates power flow calculation analysis and logical reasoning to achieve rapid fault area localization and preliminary judgment, thereby significantly improving diagnostic accuracy and timeliness. In the optimization stage, an improved chimpanzee foraging optimization algorithm is developed, integrating Hammersley sequence initialization, adaptive weighting, and somersault foraging strategies to address multi-objective coordination and accelerate convergence. Experimental results demonstrate the superiority of the proposed method on typical power grid fault datasets. In the preliminary decision-making phase, the fault identification accuracy reaches 98.7%, with an area under the curve of 0.967, indicating high classification stability. In the optimization scheduling phase, the power restoration ratio peaks at 98.7%, switching operation cost is reduced to 0.15, and fault response time is shortened to 1.1 s, showcasing strong global optimization and scheduling efficiency. The above results show that the algorithm has obvious advantages in fault diagnosis accuracy, dispatch globality and optimization stability. It not only provides a deployable and scalable intelligent optimization solution for power grid fault recovery in complex scenarios, but also provides solid technical support for improving the intelligent decision-making level and operational resilience of power systems. It has important engineering and practical significance.