Parity Space Vector Machine Approach to Robust Fault Detection for Linear Discrete-Time Systems

In this paper, a novel robust fault detection (FD) approach called parity space vector machine (PSVM) is proposed for linear discrete-time systems. Aiming to achieve a tradeoff between false alarm rate (FAR) and FD rate (FDR) simultaneously, we focus our study on an integrated design of parity space-based FD in the context of residual generation and residual evaluation. Without a prior knowledge of the distribution of the unknown inputs, we propose to construct a PSVM model and formulate the underlying FD problem as a distribution-free Bayes optimal classifier, where the FAR and FDR indicate the worst-case classification accuracies of future residuals for the fault free case and faulty case. Then a bank of parity space vectors and corresponding thresholds can be designed integratedly by applying the techniques of the minimum error minimax probability machine and, at the same time, an optimal tradeoff between FAR and FDR is achieved. Finally, the effectiveness of the proposed approach is demonstrated on a longitudinal control system of unmanned aerial vehicle and further comparison with a traditional parity space-based FD is also addressed.

Automated summary

The paper introduces a novel robust fault detection method called parity space vector machine (PSVM) for linear discrete-time systems. By constructing a PSVM model and formulating the fault detection problem as a distribution-free Bayes optimal classifier, the approach achieves a tradeoff between false alarm rate (FAR) and fault detection rate (FDR) effectively.