Research And Application Of Prediction Technology About Temperature And Air Pressure Based On Chaos Theory

Accurate prediction of air temperature and air pressure for economic development and national defense construction is of great significance. Based on the chaotic time series prediction theory as the basis, to Lianyungang region temperature and air pressure data for time series experimental samples, studied the time delay and embedding dimension two phase space reconstruction parameters and three kinds of chaotic time series prediction algorithm to compare research. The main work of this paper is as follows:(1) On time delay and embedding dimension selection gives a full theory analysis and experimental study. Phase space reconstruction of chaotic time series is predicted based, time delay and embedding dimension are in phase space reconstruction of two important parameters, selecting reasonable parameters for phase space reconstruction quality plays an important role. In view of Lianyungang region of the temperature and pressure data, using mutual information algorithm, we get three a time delay. Through the G-P algorithm to calculate the embedding dimension, for temperature and pressure were calculated for three reasonable embedding dimension, from the prediction results, the reasonable parameters, the prediction error is acceptable.(2) Using the weighted zero order local method and weighted one-rank local-region method to forecast research. For the reconstruction of the phase space, were calculated for the center point to the adjacent domains of each point in the weights in the weighted, zero order localization method, directly through the various weights and phase coordinates to calculate the next phase coordinates, whereas in the weighted one-rank local-region method by least square method, phase change linear coefficient, combined with the weights and point coordinate calculation of a future phase coordinates, to predict time series with a value, in this paper, experimental results show, weighted one-rank local-region method weighted zero rank local-region method in prediction of air temperature and air pressure data has a higher precision.(3) The maximum Lyapunov exponent model of forecasting research, calculating the longest predictable step, and with the local prediction model comparative study. Using Wolf algorithm to get the model of the maximum Lyapunov exponent and the longest predictable step, calculation of distance from the center point of recent phase point, using the maximum Lyapunov exponent and the recent phase point and its next phase distance between points to predict future phases, from the prediction of the phase point coordinates in isolated predictive value, in this paper, experiments show that, as the local prediction model in the presence of the accumulated error and linear limitations, so based on the maximum Lyapunov exponent prediction method in the prediction of air temperature and air pressure data than the weighted zero order local method and weighted one-rank local-region method in accuracy are improved.