In this thesis, the feasibility of using model predictive control in a hybrid diesel-electric marine powerplant is investigated. Initially, the modeling procedure is divided into two parts. The aim of first part is to find linear models using datasets with experimental data at the nominal operation point of the hybrid powertrain and investigate the adoption of different signals considered as system disturbances, as the controller is capable to cope with multi-variable systems. Second part aims

more »

... o the development of a non-linear multivariable model which can describe the hybrid powertrain behavior over a wider range of operation. Finally, a output feedback manipulation is suggested in order to enable a better prediction of the future model outputs. Model Predictive Controller (MPC) is a model-based controller which tries to compute the optimal sequence of the control moves in order to succeed the optimal control performance of a plant over a finite prediction horizon. MPC is capable of tackling multivariable processes, satisfy input and output system constraints, deal with long time delays and utilize plant response to measured and unmeasured disturbances knowledge. Performance evaluation of the designed MPC controllers was firstly conducted through step response simulation against each controller's internal model, in order to fine-tune its parameters and ensure the system stability. As a next step, the controllers were evaluated through simulation using open-loop experimental data and the non-linear diesel engine model, using also the electric motor system model for the total power split determination. Last, the performance of the various MPC controllers was experimentally verified on the hybrid propulsion powertrain at LME. MPC response was tested at various load profiles, including alternating and propeller load, against static and dynamic reference tracking, evaluating disturbance rejection and efficiency in operating the plant within the desired exhaust emission and fuel consumption limits during closed-loop control.