A Stochastic Approach to Shipboard Electric Loads Power Modeling and Simulation

This work proposes a method to model, predict and simulate shipboard electric load power profiles for specific users in order to overcomes the limitations encountered with traditional methods applied at the design phase of the ship. A probabilistic characterization of the user's operating modes is proposed combined with a stochastic process called Hidden Markov Model (HMM) in order to model and simulate the load power profiles. Main inputs for the user's characterization are the initial state pi of every user, the state transition probability matrix A(i,j) and, finally, the corresponding emissions at each operating state e(i). All these inputs can be derived by the traditional information available from the electrical power load analysis (EPLA) or by experimental readings obtained through field measurements performed on-board ships. The method has been applied, tested and validated on a typical on-board laundry service of which experimental readings are available. The simulation of the load power profile shows a significant accuracy if compared with the recorder data (e.g. a mean absolute percentage error equal to the 11.1%).