The Simulation And Research Of Excitation System Based On Artificial Neural Network In Ship Power Station

At present, with the increasing tonnage of ship and more and more ships propelled by electric power, the capacity of a generator in a marine power station is getting larger and larger and the control system of voltage stability in vessel must be with better performance. The performance of excitation control of generators is the key to the voltage stability of power system in vessel electrical network. Presently, the wide-applied SCR excitation control is unchangeable in the whole control process after rectified. But in the practical control process, when the state of system and parameters are changed, the uncertainty of the parameters will not be up to the strict requirement of system. However, because of the ability of dealing with nonlinear system and fault tolerance, the artificial neural network will be a new method to deal with the uncertainty nonlinear system and control problems.Take the excitation control system of COSCO HANBURG vessel as an referenced object, combines ANN with linear optimal control to design the new adaptive excitation controller. It takes active power, reactive power and the end voltage of generator as the feedback inputs. After trained offline based on lots of samples, it works out Riccati equation and the optimal feedback gains, gets the best control methods.Finally, The paper makes some simulated researches of BP ANN control and SCR excitation control with the software of MATLAB. Two control effects are compared and researched according to the simulated result. BP ANN control strategy has better voltage regulation precision and dynamic characteristic in the course of load started, load changed, electrical network fault and so on.