In response to the problems of poor detection accuracy and low efficiency in existing insulation state detection methods for power capacitors, based on the analysis of the online monitoring system for power capacitors, this paper proposes an improved adaptive neural fuzzy inference system for detecting the insulation status of power capacitors. The parameters of the adaptive network-based fuzzy inference system are optimized through the improved particle swarm optimization algorithm, improving model detection accuracy and reducing model training time. The performance of the proposed method is analyzed through numerical examples, verifying its effectiveness and superiority. The results indicate that, compared with conventional detection methods, the detection results of the proposed detection method are basically the same as the actual insulation state level, and can accurately detect the insulation state, the detection speed can meet the needs of real-time detection, and the detection accuracy is higher.