Abstract This paper presents a hybrid artificial intelligence framework designed to improve the diagnosis of error and subsequent optimization of operational efficiency in complex energy systems. By integrating knowledge-based systems with evolutionary algorithms, we address the critical need for fast and accurate fault identification as a prerequisite for maintaining the reliability of the network. Knowledge-based components use expert rules to precisely determine the locations and types of errors, and then dynamically limit genetic algorithms (GAs) to optimize restoration strategies such as load reduction and network reconfiguration. Performance metrics reported in this study were obtained from multiple independent simulation trials conducted under varying fault locations and noise conditions. The proposed framework achieved higher fault-diagnosis accuracy with an average detection latency of less than 55.65 ms by performing comprehensive simulations of the IEEE 30-bus and 118-bus systems, across different test conditions. By intelligently reducing the search space, the hybrid approach reduces the total system recovery time to less than 1.0 s, a significant improvement over the independent evolutionary methods that require more than 5.0 s. In addition, the optimization module successfully limited load loss to less than 5%, surpassing traditional methods by more than 50% in terms of service continuity. Robustness analysis confirmed diagnostic accuracy of more than 98% even below the signal-to-noise ratio of 20 dB, and IEEE 118 bus scale-up test showed a 38% to 47% increase in computational efficiency compared to standalone particle swarm optimization (PSO). These results highlight the performance of the framework and the practical application of critical energy infrastructure management with real-time, resilient management.
Babu et al. (2026) studied this question.