System identification using evolutionary computation and its application to internal adaptive model control

The requirement for high-quality control of complex and/or structure-unknown plants is growing for real-world industrial machines. Indirect adaptive control (IAC), which identifies, models and updates the compensators automatically, is expected as one of the most promising ways to meet this requirement. Conventional IAC, however, requires information about the structure of the plant, i.e. the order of its transfer function, in advance. This paper presents a new IAC scheme which utilizes a genetic algorithm (GA) in its identification part and embeds it into a control system. In the proposed framework, the information on the order of the plant is not required, since the GA can find both the parameters of the plant and the structure of the plant dynamics autonomously. A two-degree-of-freedom internal model control (IMC) is adopted as the basic controller architecture, because an indirect adaptation mechanism can be achieved seamlessly. The effectiveness of the proposed framework is verified through some numerical simulations and experiments applied to the velocity control of a multi-mass system