Ai-enhanced nonlinear model predictive control of power split and thermal management of vehicle powertrains

Abstract

Presented are model predictive control (MPC) systems, methods, and devices for regulating operation of hybrid powertrains. A method of controlling a hybrid powertrain includes a controller determining path plan data, including a vehicle origin, destination, and predicted path. Based on this path plan data, the controller estimates vehicle velocities for multiple rolling road segments of the predicted path and, based on the estimated velocities, determines an estimated power request—transmission input torque and/or vehicle axle torque—for each road segment. The controller calculates a minimum cost function such that total fuel consumption to generate the engine power output is minimized. Minimizing the cost function is subject to battery pack current limits and state of charge (SOC) terminal costs at the ends of the rolling road segments. Command signals are sent to the engine and motor to output engine and motor torque based on the calculated minimum cost function.