Study On Short-term Forecasting Of Fruit Pirce Based On Artifical Neural Networks

Since the reforming and opening-up in1978,especially after joining the WTO in2001, agriculturalmarket activity is enhancing in China and the price of agricultural products is instability, while theoverall Scale of the agricultural market is expanding all the time.The fruit, a typical of fresh agriculturalproducts, is the third largest farm produce next to grain and vegetables. The price of fruit,especially short-term price,is volatile and fluctuant.Which is adverse effects for farmers andconsumer.So the short-term price forecasting of fruit is a great research,and it’s helpful for framer andconsumer.In view of this,the artificial neural network method is introduced and used in this article.Withthat,the future prices of fruit is predicted and the development trend is learned intelligently.The task ofresearch is done as the following aspects:First of all, the present situation, characteristics and the fluctuation in fruit market of China arediscussed and analysed. Specifically, market structure and lags behind of development are analysed inaspects of production, logistics and consumption; factors to price are talked in natural,social,economic,unexpected events and so on.How they worked are ratiocinated too.Second, the presentation of ANN and BP-NN are introducted.So are their characteristics anddifference with the traditional method. Specifically,some characteristics such asself-organization, self-learning,memory and parallel computing and something about that how to getthese work are described in detail.All the works build up the theoretical basisfor the fruit price short-term forecasting model.Third, forecasting model based on BP neural network is designed with the price data of Fuji applein wholesale market of main producing areas.The performance of forecasting model is discussed.Thedefects of model and the way how to correct that are analyzed too.From the research above of this,here are some conclusions:First of all, the fruit market has become a great market relating to people’s livelihoodin China,and the volatility of fruit price is closely related with production operators and consumers.Butnow,the research about fruit price and price forecasting based on mathematical are rare,while the studiesabout production, logistics and trade of fruit are rich.The old methods such as qualitative analysis andstatistics are more used than intelligent methods in price forcasting studies.The price forecasting offruit based on ANN maybe a better way while the short-term price forecasting is so difficult and thetraditional methods cannot work well.Second, artificial neural network, a modern intelligent prediction method,is good at researchingcomplex issues depending on its advantages like self-organization, self-learning, memory, parallelcomputing etc..A good price forecasting model of fruit is designed well base on BP-NN in this paper.Third,it works out that the short-term price forecasting model of fruit has a high predictive accuracy.But it’s also found that some problems Which are a slight lag,random factors and large errorson individual points,are existent. Although the rate of improvement is limited, the method of gettingaverage on multiple forecasting experiments can achieve the purpose of improvement.Maybe it willwork well enough some day.