Fault-Tolerant Control: Analytical and Soft Computing Solutions

Abstract

The complexity of modern systems and industrial installations, along with continuously growing requirements regarding their operation and control quality, is a serious challenge in the development of control theory as well as process and system diagnostics. The dynamic evolution of fault tolerant control theory witnessed in recent years is a partial solution if this problem. In his chapter, selected issues in fault tolerant control system design using analytical and soft computing methods and approaches are presented. Different structures of fault tolerant control systems are considered, including those with virtual sensors and actuators. Fault estimation and compensation are also discussed. In the second part of the chapter, neural predictive control is considered, with the neural model of processes implemented using feedforward networks with delays and recurrent networks. Included is an example of process control in a two tank system, which well illustrates three implemented control strategies: robust control, robust control with fault compensation, and fault predictive fault tolerant control. Another example of process control in a flow tank system shows the possibilities and efficiency of neural modelling and control.