rSOESGOPE Method Applied to Four-Tank System Modeling

It’s known that the design of identification signals plays a fundamental role in the estimation quality of dynamic systems models. Well-designed signals are able to excite the system’s dynamics to be later identified and represented in a model. This work presents the application of the rSOESGOPE (robust Sub-Optimal Excitation Signal Generation and Optimal Parameter Estimation) Method, which proposes the identification of robust parametric models from the use of multiple identification signals. In this perspective, the identification experiment is composed of optimized signals of the type Amplitude-modulated Pseudo Random Binary Signal (APRBS), designed by an approach composed of the Particle Swarm Optimization (PSO) and by the Interior-Point Method (IPM). To verify the effectiveness of the methodology, it was decided to study the classic problem of modeling the four-tank system, investigating the use of multiple optimized identification signals.