A Weighting Factor Design of Model Predictive Control for LCL-Type Grid-Connected Inverter Based on PSO and RBFNN Algorithms

One of the advantages of finite control set model predictive control (FCS-MPC) is that it allows for the consideration of multiple constraints simultaneously, thereby facilitating the achievement of multi-objective control. The configuration of weighting factors (WFs) in its cost function is the key to ensuring a balance between various control objectives and achieving optimal performance. However, there is a paucity of mature theoretical guidance to determine these WFs, which usually depends on experience or trial and error. This has the consequence of leading to a cumbersome debugging process and difficulty in obtaining optimal control effects. The paper proposes a strategy for the design of WFs, based on the particle swarm optimisation algorithm (PSO) and the radial basis function neural network algorithm (RBFNN). This strategy is intended to develop a set of scientific and systematic design processes for the selection of WFs. To verify the effectiveness of the proposed method, a three-phase LCL grid-connected inverter is selected as the control object, and a large number of simulations and experiments are carried out.