Model-free predictive current control of a voltage source inverter

Abstract

This work presents a model-free predictive current control (MFPCC) that is suitable for a voltage source inverter with a decoupled three-phase load. With this method, the load currents and the load current differences are available to predict the future load currents associated with eight possible switching states, generated by the inverter. Unlike model-based predictive current control (MBPCC), the proposed method is not based on any discrete model of the system. Clearly, it is only based on detecting the load currents and the current differences. In other words, it does not require any parameter of the system, such as resistance, inductance, or back-EMF. In addition, the proposed MFPCC only needs twice current measurements in each sampling interval. Similar to the MBPCC that the optimum switching state is chosen by minimizing a cost function, the MFPCC also undertakes this optimization process. The proposed MFPCC ensures that the load currents tracks the reference ones with high accuracy reducing the transient and steady-state errors. Using MATLAB® software, comparative simulations of the above two methods can be obtained to show and validate the feasibility and correctness of the proposed strategy.