Adaptive control of conventional and hybrid marine diesel engines subject to uncertain or time-varying dynamics

During the last decades, marine diesel engine emission regulations have become increasingly stringent due to environmental regulations imposed by the International Maritime Organization (IMO). As a result, emissions minimization has recently become one of the key targets for marine engine manufacturers. In this work, two topics dealing with the engine emission control have been studied, namely adaptation of control system to varying engine dynamics and multivariable control of novel turbocharged exhaust gas recirculation (EGR) system. The thesis focuses on the applied control theory with the emphasis done on control-oriented engine modeling, state estimation and design of control systems. Robustness of the control system in marine engines with respect to its time-varying dynamics has recently become an important research topic. The variation of dynamics, caused primarily by the mechanical wear of components or their faults originating from tough operating condition, can compromise the overall stability and/or performance of the control system. While this is typically not an issue in automotive industry where the components can be easily replaced, the marine engines cannot always be fixed fast enough. Thereby, in this work, a control adaptation is proposed to deal with the engine parameter variation in a way that the response of the original system is preserved. Two sources of engine parameter variation are studied, specifically the faulty actuator of a variable geometry turbocharger and the increased time-delay of a faulty lambda sensor. The stability and robustness of the proposed control system are studied via numerical simulations with the mean-value model of the combustion engine. Finally, the performance of the control concept is validated on a medium-speed diesel engine and a hybrid diesel-electric testbeds. Superior performance with respect to fixed-parameter PI-control is demonstrated. The second topic studied in this thesis is a turbocharged EGR topology installed on marine diesel engines. These topologies have been proposed in an effort to comply with the stringent emission legislation imposed on the turbocharged diesel engines by means of providing stable EGR flow at various operating points which was not possible with the standard high pressure exhaust gas recirculation systems. This work focuses on the development of a multivariable predictive control system which is capable of tracking the desired EGR fraction while keeping the pressure in the exhaust receiver low. The control design for the turbocharged EGR topology is also complicated by the fact that a number of states in the feedback path of the EGR cannot be directly measured due to harsh environment (high temperature, ash, etc.). Therefore, a nonlinear observer (unscented Kalman filter) is proposed for recovering the required system states. The proposed system is shown to successfully achieve the design targets