Aiming at the influence of parameter uncertainty on the control systems, two methods of robust controller design are proposed to design output feedback controllers, regarding a ball grinding mill system to be the research object. One is the reversed-frame normalizing (RFN) method, which is based on the matrix singular value decomposition. A controller was fitted by using the least-squares technique to realize the desiring relational function of inputs and outputs. The other is the H-infinite method which is based on linear matrix inequalities (LMI). A method to design two independent SISO systems instead of a MIMO system is presented to eliminate the influence of the specific parameter uncertainties and disturbances. The results of simulation show that both of the two control systems achieve high robustness performances and the relationship between control signals is decoupled and no static error of outputs. However, the decoupling control systems, which have been widely used, are helpless under the condition of parameter uncertainty. |