Neural Networks in Feedback Control Systems

This chapter shows how neural networks (NNs) fulfill the promise of providing model‐free learning controllers for a class of nonlinear systems in the sense that a structural or parameterized model of the system dynamics is not needed. It shows different methods of NN controller design that yield guaranteed performance for systems of different structure and complexity. The control structures discussed in the chapter are multiloop controllers with NNs in some of the loops and an outer tracking unity‐gain feedback loop. Throughout, there are repeatable design algorithms and guarantees of system performance including both small tracking errors and bounded NN weights. Recent results show that approximate dynamic programming (ADP) with critic and actor neural networks, allows the design of adaptive learning controllers that converge online to optimal control solutions, while also guaranteeing closed‐loop stability. ADP also offers design methods for differential games that can be implemented online in real time.