Electric Propulsion Ship Power Load Forecasting Based On Chaos Theory

On the base of analying chaotic characterics, the aim is to establish a ship power load forecasting system combine with ship power load character, which did assure secure and economic operation of electric propulsion ship. The forecasting system covers three aspects: the chaos character analysis, the historial data beforehand processing and the forecasting of ship power load.Chaos character of electric propulsion ship power load is analyzed through the quality and ration angle. Power spectrum approach to quality angle is presented, correlation dimension and the most Lyapunov exponent are computed to ration angle, which did validate chaos character of electric propulsion ship power load. So it gives us a new acquaintance to the complexity of ship power load time series, and the ship power load could be forecasted through chaos theory.The reliability and reality of load historial data is the foundation of electric propulsion ship power load forecasting. But, the impact load in running ship power system, and the disturb data in collecting load data may cause fault data in load historial data. Focusing on beforehand processing historial load data, the singularity of ship power load is analyzed using wavelet method, and the ship power load data can be of self-adaptability to be apart from its noises based on fractal dimension, then fault data can be eliminated, which did provide better data basis for the electric propulsion ship power load forecasting.The univariate time series of electric propulsion ship power load forecasting model is set up. The self-similar trajectory is looked for, and the correlation degree is acted on chaos local forecasting model. Moreover, the similar days load is input into chaos local forecasting model. The practical examples express that the above methods have more precision to univariate forecasting of ship power load. To counter the influence of different factors on ship power load, the multivariate chaos forecasting model of electric propulsion ship power load is set up. Based on the phase space reconstruction of univariate time series, phase space of multivariate time series is reconstructed, and then chaos forecasting model of multivariate time series can be established to forecast the electric propulsion ship power load. Moreover, to cunter the varying rule of the ship power load itself, wavelet transform is applied to decompose electric propulsion ship power load. Ship power load is divided into normal load and dead weight influence load. A chaos local correlation forecasting model is built for normal load, and a polynomial regressive model is built for dead weight influence load. The practical examples express that the methods are effective and feasible.