The Improvements And Applications Of The Algorithm Of Parameters’ Optimization

Prediction is the process of building system models on previous results and makingpredictions on the basis of historical events. With the developing of science andtechnology, the prediction methods have been greatly improved. In the practice ofprediction, time-series often present in highly sophisticated dynamics nonlinear states. Forthis reason, fitting predictions can not be conducted well by single prediction method.Based on this situation, in order to improve the accuracy of fitting, on one hand, variousways of prediction methods modification are put forward in this dissertation. On the otherhand, if combined in a proper way, different prediction methods will become combinationforecasting methods. In this way, information collected in different ways could be puttogether, thus improves the accuracy of fitting. In this thesis, the improvement andcombination prediction model are applied in the stock market, demographic data and otherfields. The effectiveness of the improved method is verified by actual data fitting results.This paper studies the following three aspects.Firstly, nonlinear conjugate gradient method and the basic idea of combinationforecasting model, system analysis of time series model parameter optimization methodare studied. Two modified nonlinear conjugate gradient method are presented andsuccessfully applied to the time series model parameter optimization, namely the use of animproved conjugate gradient method to optimize the parameters of the model. Then, thefeasibility and superior of this method is proved by a serial of examples.After that, gray systematic model and BP neural network model are studied, and aprediction method is given on the basis of them. At the end of the paper, Using severalexamples, the feasibility of this method is proved successfully.Finally, based on the traditional time series prediction method, BP neural networkprediction method is applied to correct the accuracy of fitting. the combination model ofthe forecasting method improve the prediction results of the original model, so that thedata processing is more effective.