A Hybird Method Based On Wavelet And Kalman Filter For Prediction Of The International Oil Futures Prices

For relevant departments accurate prediction can provide reliable and investors in the management and decision-making basis. the world governments and experts and scholars are very concerned oil futures prices market changes because its price changes brought about by the great influence .In present the oil futures price prediction methods typically include BP network, regression analysis method, the wavelet transform and the forecast method, etc. But the existing methods exist have computational complexity, slow convergence speed and so on; And in the process of price changes, the influence of randomness enough emphasis the accuracy of predicted results also needs to be improved.Wavelet analysis is using signal and noise on the scale of wavelet transform coefficient under different features, and unlosly the original signal in the premise, filter the processs of important information on the edges of the signal, eliminate noise information, ultimately reconstruct a clearer graphics, so as to improve the signal clarity. Its multiresolution characteristics widely in many fields, such as: the application of adaptive filter, speech synthesis, CT imaging, signal strengthen, etc. The wavelet analysis Will used in previous oil futures prices curve, can obtain more stable and real price changes, and the forecast results produced good auxiliary function. Kalman filter can be get the best estimate of the system BY mean-square error conditions, it is used in the time domain recursive way, small amount of calculation and facilitate real-time processing. Kalman filtering theory, as a kind of important optimal estimation theory has been widely applied in all kinds of fields, such as inertial navigation, global positioning system, target tracking, communication and signal process, finance and other applications. thought establish the reasonable vector of oil futures prices system, Kalman filtering method can real-time fast prediction.Based on international oil futures market volatility and Kalman filter is the dynamic real-time tracking characteristics, this paper puts forward Kalman filtering forecasting methods (W - K) based on wavelet analysis. According to obtain real-time theprice of observations, using the threshold of wavelet transform method to process the early changes, and obtaining more stable and real price rate observation value using interpolation methods, and then give feedback to the system. System established the state vector of observation through price observation and price rate.It finish the next phase of the price projections and system real state estimation using the iterative equations of kalman. Finally ,the numerical results show that this paper W - K prediction method can improve the reliability and accuracy of the prediction model and has certain practical value compares the traditional forecasting method .