Research On Diesel Speed-regulating Based On Reinforcement Learning PID Controller

The diesel engine eletronic control can greatly improve the operational reliability, betterthe fluel-economy and emission performance, having became one of the most importantdirections of the current diesel engine technology. And the choice of the system’s controlstrategy is the key to affecting the diesel speed-regulating performance.The current available diesel engine speed-regulating systems mostly adopt PID controllerwhich has many characteristics such as simple structure, reliable, stable and so on. However,it not only has the shortcomings of the large scope changes of load or model parameters, butalso difficult in Parameters optimization and can’t ovecome the effect of non-linear factors.Based on the characteristics of reinforcement learning algorithm solving problem, the paperproposed the diesel speed-regulating methods of the reinforcement learning PID controller,furthermore, designed the modle of reinforcement learning PID Controller and developSimulation research. According to this paper work requirement, the paper establishedturbocharge dieselengine average model and eletronic speed control systems model viastuding and comparing the characterics of different type models. And the paper also gavesome Simple and practical solutions to the problems in modeling such as compressorcharacteristi ccurve and instructions thermal efficiency curve.Reinforcement Learning PIDcontroller is conducted digital simulation in accordance with the established eletronic speedcontrol systems model. Simulation results showed reinforcement Learning algorithm canadjust PID controller parameters online, and acquire good diesel speed-regulating.