The Simultaneous Perturbation Stochastic Approximation‐Based Data‐Driven Control Design

In this chapter, a multivariate stochastic optimisation technique is utilised in the design of data‐driven control systems. In this chapter, simultaneous perturbation stochastic approximation (SPSA) will be used in the data‐driven feedback control framework to obtain the optimised controller parameters. The SPSA algorithm uses the gradient approximation and only requires two measurements of the cost function. Proportional‐Integral‐Derivative (PID) is the most widely used control methodology in the industry. The first data‐driven control technique of the early last century was PID‐based. Model predictive control is the most widely used advanced control technique in the industry. The chapter selects the liquid slosh and the ball and beam under‐actuated systems to show the procedure and effectiveness of the SPSA‐based data‐driven controller design.