A novel SPSA-based IMC-PID data-driven control method

This paper proposes a new data-driven control (DDC) based on SPSA(simultaneous perturbation stochastic approximation). Inheriting the essence of PID controller: simple and practical, an IMC-PID controller is conducted as Function Approximator in SPSA tuning. All the parameters in this polynomial controller are tuned online simultaneously directly via system I/O data, rather than only the filter coefficients tuned off-line. IMC concept introduced into DDC structure help provide initial value, the range of controller parameters and the random disturbance vectors of SPSA. This effort improves the SPSA stochastic approximation performance and every IMC-PID controller parameter has physical meaning. Compared with the traditional SPSA DDC, this method is more simple and practical and is expected to be applied in nonlinear, time-varying and large time delay process.