An Application of aModel-Prediction-Based Reference Modification Algorithm to Diesel Engine Air Path Control

Abstract

Engine air path systems have a lot of constraints coming from hardware and actuator limits. In order to explicitly treat the constraints, we apply a reference governor (RG) to the engine air path systems. The RG algorithm optimizes a cost function that evaluates the degree of constraint satisfaction based on the gradient descent method. Experimental results using a real diesel engine show improvement in control performance with constraint satisfaction.