Fault detection of networked control system subject to random packet dropout and multi-rate sample data

In this paper, the problem of fault detection is addressed for networked control system (NCS) subject to both random packet dropout and multi-rate sampled data, which is usually unavoidable. Since the networked control system suffers heavy resources competition and network congestion under the limited bandwidth, resulting in multi-rate sampling and data packet dropout. Observer based residual generation and evaluation approach is proposed. First, by lifting the multi-rate multi-sensor model to a single-rate one, a residual generator with causality constraint is obtained. Then, the parameters of the residual generator are achieved by optimizing a performance index, which is robust to the unknown disturbances and packet dropout without loss of the sensitivity to the faults. Finally, an appropriate threshold of residual evaluation is calculated by making full use of the stochastic properties of the disturbances. Simulation results are given to illustrate the effectiveness of the proposed method.