Multi-step Prediction Based on Chaotic Time Series

In order to get a better performance of interference isolation,the prediction of platform carrier motion attitude was studied in the paper.After judging the chaotic characteristics of time series,multi-step prediction of roll attitude was carried out with weighted one-order local-region method.Aiming at the problem of error cumulating in predicting process,error series predicting was carried out while time series predicting,and the roll attitude predicting values were compensated by error predictions in time.Two methods of multi-step predicting with error compensation were given in the paper,and the simulating results of experimental cistern datum show that the predicting precision of the weighted one-order local-region method with error compensation is improved,and that the error cumulating is controlled.Furthermore with the relative mean square error as the evaluating index,the paper points out that the multi-step prediction with LPC compensation is the best and it is valuable for practice.