Generalised predictive controller (GPC) design on single‐phase full‐bridge inverter with a novel identification method

Abstract

The design of an on-line generalised predictive control (GPC) technique with a novel identification method is presented in this paper for a single-phase full-bridge inverter in the presence of different disturbances. The controller uses system discrete-time model to reach the control variables with a prediction over these values, followed by computing a cost function for control aims. However, in this controller, the need for the mathematical model of the system is removed since the black-box identification strategy is used. Moreover, GPC structure has many advantages including low computational complexity, systematic design procedure, and fixed switching frequency that makes it a good alternative for practical applications. Various disturbances can have a negative impact on a DC–AC inverter; thus, considering robust dynamics and ease of implantation, the GPC scheme is used along with an improved exponential regressive least square identification algorithm as an adaptive strategy in the controller. Moreover, the prediction horizons of this controller have been increased, resulting in its low steady-state error and better performance. Furthermore, harmonics in the sinusoidal signal can decrease the total efficiency of the system; thus, an LC filter is used to reduce the level of total harmonic distortion. However, the stability of the filter is the most challenging issue in designing a suitable controller. Finally, the strength of the current controller is verified using experimental and simulations results.