A methodology of estimating hybrid black-box - prior knowledge models of an industrial processes

Abstract

The paper presents a combined prior knowledge - black-box approach to modeling of the dynamic systems. Proposed methodology assumes two basic steps: first an optimization of a physical model, using Particle Swarm Optimization algorithm is performed. Next, a black-box model is estimated so as to reduce the remaining error of the physical model. Proposed methodology was presented on the example of improving the accuracy of a pneumatic actuator physical model.