Integration of Predictive Control and Interconnected Structure for Autotuning Velocity Controller

Abstract

In the servo system, the autotuning controller based on a generalized predictive control (GPC) usually requires a specified reference instruction and the accurately linear system model. The former seriously limits the servo applications, while the latter's lack easily reduces the control performance. This article proposes a variable-structure proportional–integral (VSPI) velocity controller based on a GPC and an interconnected structure to address the above limitations. First, the VSPI controller is designed to track the arbitrary-order instruction without a static difference. Second, based on the Taylor series and the higher order difference operators, an improved prediction model, whose prediction output only includes the velocity error, is constructed to liberate the GPC from the reference instruction. Then, the control parameters of the VSPI controller can be adjusted online by the rolling optimization. Furthermore, the interconnected structure using three modified sliding-mode observers (MSMOs) is designed to efficiently provide the system model for the GPC, where an adaptive gain and an equivalent derivative designed in each MSMO make it possible for the identification parameters to converge exponentially. Finally, the third MSMO is used to identify the load disturbance for promoting the interconnected structure and the disturbance compensation, which can ensure the identification accuracy and the linearization of the system model. Simulation and experimental results demonstrate the effectiveness of the proposed method.