Dynamic modeling and simulation of marine diesel engine using Elman networks

CODAD propulsion plants are complicated systems with hull, engine and propeller operating together. It is very important to control the loads transferring from one diesel engine operating mode to twin engines operating. During the development phase of the diesel load sharing controller, the system simulation can be used as a tool to support test and verification of the complex control algorithms. The simulation models should accurately reflect the dynamic performances of the system, especially diesel engines which is high nonlinear, time-variant and cannot be linearized owing to its wide operating range. This paper presents a modeling method, based on Elman network, of diesel engines represented only by measurement data. Comparing the measurements from a test bench and the simulation results shows that the model based on the neural network modeling method can represent dynamic behaviors of the diesel engine with good precision.