Optimal Control To Reduce Emissions In Gasoline Engines: An Iterative Learning Control Approach For Ecu Calibration Maps Improvement.

Control of emissions in gasoline engines has become more stringent in the last decades, especially in Europe, posing new and important problems in the control of complex nonlinear systems. In this work a preliminary investigation is conducted on the idea of exploiting Iterative Learning Control to optimize calibration maps that are commonly used in the Engine Control Unit of gasoline engines. In this spirit, starting from existing maps, we show how to refine them using a gradient-descent iterative learning control algorithm, considering additional constraints in the optimization problem. The outcome of this procedure is a control signal which can be integrated in a modified map. The performance of the proposed technique is validated on the provided training signal and cross-validated on different reference signals. Simulation results show the effectiveness of the approach.