Prediction method based on time sequence

Abstract

The invention discloses a prediction method based on a time sequence. The prediction method comprises the following steps of: continuing an original time sequence by using a mean generating function (MGF) method to obtain continuation sequences; screening the continuation sequences by using an optimal subset regression (OSR) method to obtain an optimal subset; by using the optimal subset which is obtained in the preceding step and using the original time sequence as output, training a back propagation (BP) neural network to obtain a BP neural network prediction model; and predicting by using the BP neural network prediction model. By combining the MGF method, the OSR method and the BP neural network, a novel MGF-OSR-BP prediction model is established. By focusing on the prediction model and construction of a learning matrix, the accuracy rate of prediction is higher, and reference can be provided for medium-and long-term prediction research of similar time sequence elements.

Description

technical field [0001] The invention relates to a forecasting method based on time series. Background technique [0002] At present, the medium and long-term forecast work at home and abroad is mainly based on statistical methods, but the statistical forecast methods widely used in medium and long-term forecasts are mostly applied to monthly and seasonal time scale forecasts. One of the major difficulties in climate prediction on yearly and above-year time scales is that it is difficult to find many objects related to long-term forecasting on time scales such as annual average precipitation, annual average temperature, earthquake cycle, and hydrological cycle. Corresponding, physically meaningful predictors. Therefore, the nature of this forecasting problem mostly depends on forecasting techniques such as time series and cycle analysis. At present, in addition to the conventional time-series autoregressive model, the time-series homogenous function method proposed by Wei F...

AI Technical Summary

Problems solved by technology

Jin Long and others used the neural network integrated forecasting method to study the spring precipitation in Nanjing. The results showed that the neural network integrated forecasting model fit or forecast results were better than other conventional integrated forecasting equations, but the selected neural network model structure, parameters and hidden nodes The determination problem has not been well solved [Jin Long, Chen Ning. Research and Comparison of Integrated Forecasting Methods Based on Artificial Neural Network, Journal of Meteorological Society, 1999, 57(7): 198-207]

Jin Long et al. established a precipitation prediction model in the flood season based on BP neural network, and the results showed that this method was better than the stepwise regression model in terms of historical sample fitting and forecasting effect. De, Jinlong. Research on Precipitation Prediction Model in Flood Season Based on BP Neural Network. Meteorological Science, 2002, 22(4): 461-467]

Images