Internal Model Controller Design Based On H₂/H_∞ Optimization

H2 control developed since the 70s and H∞ control developed since 80s are important achievement of modem control theory. The former is well developed, and the latter remains as a very active research field in present years. Because They emphasized some important facts of practical process, such as uncertainty of model, these theories led to a new era of feedback control development. On the other hand, internal model control (IMC) developed since 80s has been applied to control engineering broadly, because of its simple structure and design procedure.Comparatively speaking , H2/H∞ control and IMC have their own individual features. Having experienced development of over 20 years, H2IH∞ control has built a strict theory frame for system analysis and synthesis, but its design procedure is relatively complicated. In contrast, the design of IMC is easier, but it lacks strict theory support for system analysis and systhesis.lt is the objective of this paper to simplify the design procedure and realize optimization of the system performance by absorbing their own advantages. The work of this paper is to design IMC controller to improve system performance by taking advantage of theory achievement of H21 H∞ optimization. With the widespread application of computer, the research of data-sampled system has more practical importance, thus data-sampled system is chosen as the research object of this paper. In the course of reserch, we start with the single input single output (SISO) case, the results of the research show that the controller designed with H2 optimization can lead to smooth response of thesystem, and that with H∞ optimization can result in fast response of the performance. Based on the research of SISO case, we employ the method of feedforward compensation to decouple multiple input and multiple output(MIMO) plant. It is the key point to design H2 -optimization compensator with the approach of l2 -norm optimization so as to apply directly the result of SISO case to the decoupled system. The research results show that the approach proposed in this paper is effective too with the MIMO case.