Model Predictive Control with Gradient Descent Applied to DAB Converter Control

Abstract

The search for less polluting means of transport has intensified the research and promotion of electric vehicles (EVs). Unlike conventional vehicles, EVs have an electric power circuit composed of power electronic converters that together make up the battery charging system and the powertrain. These systems usually have different voltage levels, so it is attractive to use isolated converters, for example, the Dual Active Bridge (DAB) converter. Among the various techniques used to control converters, the application of model predictive control (MPC) has recently become evident. The advantages of predictive control are the design based on plant equations, the ease of implementation, and good precision and accuracy. In MPC the search for the desired setpoint is performed by minimizing a cost function. In the present study, the minimization of the cost function is achieved using the gradient descent technique. Also, computer simulations performed on the Matlab/Simulink platform with the DAB converter being controlled by MPC and MPC with gradient descent are presented and the results compared and discussed.