Line Fault Detection Strategy for Multi-Microcontroller Systems Based on PSO-GA Optimized SVM

In this paper, we propose an innovative real-time fault detection method for multi-monolithic computer (MCU) systems that combines a lightweight interpretable artificial intelligence (XAI)-enhanced support vector machine (SVM) and an improved hybrid particle swarm optimization (PSO)-genetic algorithm (GA) optimization technique. XAI-SVM achieves efficient and interpretable fault detection by integrating the high accuracy of a traditional SVM with the XAI-SVM achieves efficient and interpretable fault detection by combining the high accuracy of traditional SVM with the interpretability of Shapley Additive Properties (SHAP), while optimized for resource constrained MCU environments. The improved PSO-GA hybrid optimization algorithm effectively balances global exploration and local exploitation by introducing an adaptive weighting mechanism, which significantly improves the optimization efficiency of XAI-SVM hyperparameters. In addition, a distributed fault detection framework is developed in this paper, which improves the reliability and robustness of the overall system by coordinating multiple MCUs and introducing a lightweight consensus mechanism. Test results show that compared with existing methods, the method proposed in this paper achieves higher fault detection accuracy and interpretability in a multi-MCU system while maintaining excellent real-time performance. The method shows significant improvement in detection efficiency, result interpretability and system adaptability, and is particularly suitable for complex industrial application scenarios requiring fast and accurate fault identification and diagnosis.