An LMI formulation for H2 based design of multivariable precompensators

Multivariable precompensators are effective in allowing extension of the application of decentralized control techniques to systems which are not weakly-coupled. This paper proposes an LMI method for the design of these precompensators which bridges the gap between the previously proposed techniques by offering a balanced compromise between speed of computation (important in large scale systems) and design conservatism (important to increase system performance). The method is based on the minimization of the H ₂ norm of a modified open loop transfer function. The approach is applied to a highly interacting real life complex benchmark problem.