The Application Of BP Neural Network In Short-term Forecasting Of Exchange Rate

With the development of China's economy, the exchange rate becomes the key economic variables and contacts with macro and micro economy. RMB exchange rate plays an important role to maintain balance in China's national economy, especially on July 21, 2005, China's Central Bank imposed "currency reform". Then China experienced an economic crisis in the world in 2008.The whole things made the RMB exchange rate issue affect the key issues of our relations with important trading partners. Therefore, exploring the inherent laws of the exchange rate system and better accurate prediction of exchange rate fluctuation and amplitude have great practical significance and value.Due to the limitation of traditional linear time series models, we cannot effectively reflect the exchange rate data for the non-linear characteristic. So this paper respectively uses traditional of time sequence model--ARIMA model and BP neural network model for fitting time sequence of nonlinear features to fits the RMB exchange rate data. Then paper analysis the two model of forecasting errors and fitting condition. Research result shows that the effect of BP neural network in the nonlinear data fitting is better than traditional time series model.This paper is mainly composed of three parts: in the first part, the paper introduces the concept of RMB exchange rate and reviews the development of the RMB against the U.S. dollar and analyzes the main factors of exchange rate fluctuations and the influence of exchange rate fluctuations on the internal and external economy of the country. The second part, the paper introduces the relevant theories of the traditional linear time series models and summarizes the modeling steps of fitting ARIMA model, and then introduces the characteristic and research of neural networks and the structure and features of BP neural network in detail. The third part is the empirical part. The paper chooses the RMB/USD exchange rate data after the reform of currency and does the model fitting and forecasting. After comparing the fitting and forecasting results.The model can accurately predict the price level and fluctuation trend of the RMB exchange rate in China. It has a certain guiding significance for foreign enterprises or banks and other investment institutions to formulate the correct monetary policy and to avoid the risk of exchange rate fluctuations.