Included-Angle-Based Decomposition and Weighting in Multimodel Predictive Control of Hammerstein Systems

A measurement of nonlinearity (MoN) method is proposed for single-input single-output (SISO) Hammerstein systems based on the included angle, which is helpful for both analysis and control synthesis of Hammerstein systems. Based on the MoN method, a systematic multimodel decomposition method is put forward to get a set of local models to approximate the considered Hammerstein system. Then linear model predictive controllers (MPCs) are designed based on the local models. Finally, an included angle based weighting method is initiated to combine the MPCs into a global multimodel MPC controller (MMPC), which can be employed for set-point tracking and disturbance rejection control. Two SISO Hammerstein systems are investigated to illustrate the effectiveness of the proposed methods. Simulations prove that the proposed method is superior to the common nonlinearity inversion method.