A novel PID gain tuning for a system with target specifications using evolutionary constrained optimization

Abstract

The PID control is one of the most widely applied control techniques in the control industry. To control the system precisely, PID gain parameters should be tuned finely based on the system model. Since complex or unknown system is hard to model, there are many methods proposed to tune PID gain parameters by checking error of the system without any modeling. These methods usually estimate minimum error of the control system without considering any other system specifications. In this paper, we propose a novel PID gain tuning method with satisfied target specifications. In this paper, two-phase evolutionary programming (TPEP) algorithm is utilized to optimize PID gains of a nonlinear and constrained system. In the first-phase of TPEP, general evolutionary programming is applied to a non-linear system. To satisfy constraints, augmented Lagrangian method is utilized in the second-phase. The proposed algorithm optimizes PID gains by establishing inequality constraints from target specifications. Thus, PID parameters are estimated so that the system has minimum error and the target specifications are satisfied at the same time. We demonstrate our proposed algorithm with DC motor simulator by comparing with a conventional gain tuning method.