A two-step hydrogen powered fully electric ship energy management framework combined with probabilistic interval load forecasting and model-free predictive control

A B S T R A C T In recent years, in response to the implementation of global energy conservation and emission reduction plans, the application and promotion of hydrogen powered fully electric ships (HPFES) have received great attention. However, the current drainage volume of HPFES is generally small, mainly used in urban inland waterways with more complete hydrogen refueling station facilities, and the factors affecting the fluctuation of operating load are more complex. Therefore, it is urgent to develop a reasonable energy management strategy (EMS) for the combined power supply system of proton exchange membrane fuel cell (PEMFC) and lithium batteries (LB). In this regard, this article combines probabilistic interval load forecasting (PILF) with model-free predictive control (MFPC) and proposes a two-step EMS framework based on improved empirical dynamic modeling and data-enabled predictive control (IEDM-DeePC) to achieve optimized operation of HPFES power systems. The first step is to establish an IEDM model that combines HPFES system operation status data and channel hydro meteorological data for PILF. The second step is to use the PILF upper and lower predicted load sequences as disturbance input vectors for DeePC, and construct a Hankel matrix together with the control input vector and output vector to perform rolling optimization calculations for DeePC. Finally, the method was validated using actual data from an HPFES vessel in the Yangtze River channel, and the results showed that the constructed IEDM-DeePC framework outperformed multiple traditional methods.