Fault diagnosis of coal-mine-gas charging sensor networks using iterative learning-control algorithm

Abstract

To detect and estimate the faults of discrete linear time-varying uncertain systems, a discrete learning strategy is applied to fault diagnosis, and a new fault-detection and estimation algorithm is proposed. The algorithm adopts the threshold-limit technology. In the selected optimal time domain, a residual signal is used to perform iterative learning correction for the introduced virtual faults so that the virtual faults in an actual system approach the actual faults. The same method is repeated in the remaining optimal time domain to achieve the objective of fault diagnosis. The algorithm not only completes the fault detection and estimation of a discrete linear time-varying uncertain system but also improves the reliability of fault detection and reduces the false alarm rate. Finally, the simulation results verify the effectiveness of the proposed algorithm.