Based on neural network time series forecasting methodsAuthor
1、Liaoning Technology University
Abstract:This is a time series based on neural networks to predict the kind of optimization problem, neural networks, parallel processing, a strong nonlinear mapping ability, that is, it could be a number of non-linear signal processing methods and tools for integrating, changing the a simple calculation of the nonlinear problems in the past to reduce the number of iterations, improve operational efficiency, the first to use kolmogorov theorem, making neural network time series prediction with the theoretical basis for the article, within five years through the establishment of Beijing precipitation in the specific neural network model, the use of BP hidden layer neural network with multi-layer neural network, a very good solution to the nonlinear problem, use the time series behavior of the system of quantitative data. Then, using matlab neural network toolbox, drawing specific values. Finally the use of error correction rules, inspection, correction model. With this algorithm which enables neural networks to predict the time series in the application of more fields, such as the stock market, weather forecast.