Fault diagnosis of distributed networked control systems with stochastic communication logic

In order to deal with the limited bandwidth of the network, two-level DNCS with stochastic communication logic based on time-dependent Poisson process (TDPP) is presented and central fault diagnosis unit is designed for DNCS with data dropout. The central FDI unit and each subsystem both have identical estimators and for each subsystem, instead of transmitting data to the central FDI unit every sample time, only at certain time which is decided by TDPP based on the difference between the actual and estimated state of subsystem, the actual state is broadcasted to state estimator in the central FDI unit, otherwise this is no data on the network. When the central FDI unit gets the data from subsystem, its estimators is updated by the actual state, otherwise uses estimated state. Considering stochastic communication logic and data dropout at the same time, the switch between the actual and estimated state in the central FDI unit is assumed to obey a homogeneous Markovian chain. Under this special communication pattern, the fault detection filter is proposed and the fault detection problem is converted into an auxiliary H_infin filtering problem. Finally the simulations are presented and discussed.