Algorithms for reduced order robust H∞ control design

While there has been much work done on robust control synthesis, most frameworks produce controllers that are full order. The paper presents LMI-based algorithms that explicitly synthesize robust controllers that have reduced order. That is, the controller transfer functions have fewer poles than the plant. This reduction is accomplished by treating the controller order as part of a multi-objective optimization. As expected, the reduced controllers typically achieve degraded performance as the order is reduced. This trade-off is explored in the paper using three design approaches: H_∞, robust H_∞, and Popov/H_∞ design techniques, and the results are presented using trade off curves for a typical benchmark problem. The reduced order design algorithms are useful in the cases when robust performance is critical but the controller order must be constrained due to limitations of control hardware, or excessive order of the plant