Optimizing a hybrid diesel power unit using extremum seeking control

Abstract

Abstract In this paper the problem of online-optimizing the steady-state efficiency of a hybrid single-cylinder diesel power unit is considered. The special characteristic of the proposed concept is using the time path of the crankshaft position during one combustion cycle as an optimization variable. The problem can be stated in terms of an optimal control problem with free terminal time, which is solved by applying the model-free optimization scheme extremum seeking control. It is shown that by using a parameterization of the reference input based on trigonometric basis functions the evolution of the performance index can be modeled as a discrete-time nonlinear and multivariable system with a non-constant sampling interval. To this system the discrete-time variant of ES can be directly applied. An advantage of the proposed method is that no modeling of the thermodynamic processes is required. The robustness and accuracy of the concept is at first studied through simulations. Furthermore, experimental results are presented demonstrating the need for a model-free optimization and the beneficial property of the concept’s adaptation capability.