Machine Learning-Based Control Framework For Boost Converters Applying Particle Swarm Optimization

Due to the shortage of fossil fuels and their jeopardizing environmental effects, microgrids have gained popularity in the recent century since they make use of renewable energy sources. DC-DC boost converters have been applied in microgrids to satisfy Loads” needs. Although classic controllers have been used to control output voltage, they are usually not effective when the plant model changes. One such controller is model predictive control (MPC) which controls output based on predictions. As such, machine learning methods have been proposed to counter classic controllers” limitations. In this paper, a decision tree model is developed to control output voltage with the help of a Proportional Integral (PI) controller which is tuned by the Particle Swarm Optimization (PSO) algorithm to remove the steady-state error. Required data for training the model is prepared by a DC-DC boost converter controlled by an MPC, then the tree model is trained and tested in MATLAB, after which it is implemented in the Simulink model previously controlled by the MPC. Finally, simulation results are provided as voltage reference and load change to show the system's reliability.