Artificial Intelligence-Based Fault Detection Using Kalman Filter Innovation Sequence

This paper proposes a novel approach for fault detection in dynamic systems by integrating artificial intelligence (AI) techniques with the Kalman Filter innovation sequence. Leveraging AI’s pattern recognition capabilities and the Kalman Filter’s robust state estimation, the methodology aims to enhance fault detection accuracy and adaptability. By analyzing the innovation sequence, deviations from expected system behavior indicative of faults or anomalies are identified. The study explores the application of machine learning algorithms, facilitated by automated machine learning frameworks in Microsoft .NET framework, to interpret the innovation sequence and differentiate between normal and faulty system behavior. Comparison with statistical analysis of the innovation sequence developed in this research further confirms the suggested fault detection process. 19.700 simulation scenarios using a model developed in MATLAB Simulink demonstrated the efficacy of the proposed approach in detecting various fault scenarios. Results indicate distinct fault signatures in the innovation sequence, enabling accurate fault detection even in complex, dynamic systems. This research contributes to advancing fault detection techniques, offering a promising solution applicable across diverse industries.