Elements of learning control systems with applications to industrial processes

In the theory of control systems, it has been noted that cybernetics begins where the possibility of mathematical description of the controlled process terminates. This statement reflects the fact that, for complex physical processes, the explicit governing equations are either unknown or computationally unfeasible to solve. Learning control systems are ideally suited for this class of problems. The four main elements of a learning control system are described with particular attention to the synthesis of a nonlinear multivariate estimator-predictor model. Two industrial process control applications are presented to demonstrate the utility of this approach.