In this paper, a general framework for LTI (linear time invariant) MIMO (multi-input multi-output) ontrol system design is proposed based on constrained multi-objective genetic algorithms (CMOGA). A general control problem consists of controller parameters and control structure designs. Also, there always exist several objectives (transient and steady state) and constraints in time and frequency domains. In this work, we consider this general control system design problem as a mixedvariable constrained multi-objective nonlinear optimization problem. Using Q-parameterization, CMOGA intelligently searches for the optimal controller in the class of all stabilizing LTI controllers. The result is an LTI MIMO control system that benefits from conventional tools of system analysis. In this paper, the drawbacks of conventional approaches such as LQG, 2 H and ∞ H are also considered and compared with the new approach. To demonstrate the efficiency of the proposed method, an optimal control system is designed for a 3-stroty seismically excited benchmark building.

Keywords