基于BP神经网络的赣南脐橙价格预测研究

发布时间：2018-04-08 21:36

本文选题：赣南脐橙　切入点：BP神经网络　出处：《华中农业大学》2017年硕士论文

【摘要】：赣南脐橙是我国重要的柑橘类水果,2015年,赣南脐橙在农产品公共区域品牌价值排行榜位列第一,为提高赣南地区旅游品牌知名度,带动赣南苏区人民脱贫致富,促进柑橘产业转型升级发挥了重要作用。截止到2016年,赣南脐橙种植规模逾160万亩,产业产值接近100亿元,种植、加工、销售规模逐渐增加,赣南脐橙在水果市场中的消费量逐年提高,市场规模逐渐扩大,然而,赣南脐橙价格波动剧烈,短期波幅大等,易导致出现“橙贵伤民”,“橙贱伤农”两种极端情况。赣南脐橙价格的稳定,关系着生产者、加工者、销售者以及普通消费者收益及生活的稳定。对赣南脐橙价格波动进行预测研究,不仅有利于市场政策、销售策略的制定,对于保证产地生产、农户收入稳定具有十分重要的意义。本文依据价格形成理论,根据2007年-2017年赣南脐橙月度价格数据,以及2016年11月-2017年2月日度价格数据,分析其波动特征及影响因素等。基于BP神经网络,进行赣南脐橙价格预测研究,并比较分析BP神经网络与ARIMA模型的预测精度,根据预测结果,分析赣南脐橙价格走势,并使用MATLAB软件进行实证分析。因此,本文研究内容与结论如下:(1)探究赣南脐橙产业发展现状及价格变化;从价格波动、产业结构、市场推广等方面分析了赣南脐橙产业面临的问题;从市场因素、自然因素、人为因素三方面探讨影响赣南脐橙价格的主要因素。(2)预测赣南脐橙价格变化。根据赣南脐橙价格特征,确定适合的隐含层节点数,搭建基于L-M改进算法的BP神经网络价格预测模型。研究发现,BP神经网络的预测结果误差较低,月度价格预测相对误差在2.5%左右。通过与ARIMA模型预测结果的对比,在日度价格预测上,BP神经网络预测相对误差为1.5%,低于ARIMA模型2%的相对误差。(3)针对赣南脐橙价格波动提供促进其价格平稳发展的对策。基于价格预测结果及波动原因,本文认为赣南地区应完善信息建设,加强赣南脐橙价格波动的预警研究,并提高规模化经营程度,降低生产成本,以此提高价格波动风险的应变能力。本文创新点主要有:(1)在研究内容上,采取最新的赣南脐橙价格数据,减少了时间跨度过长带来的影响;此外,目前学术界关于水果与柑橘的价格预测研究较多,而关于赣南脐橙价格预测的研究较少。本文将近年来发展较为迅速的智能预测方法具体的应用到赣南脐橙价格预测领域,不同于以往的定性分析。(2)在预测方法的选取上,比较基于L-M改进算法的BP神经网络与标准BP网络的优度,以及BP神经网络与ARIMA预测模型的预测精度,选取合适的预测模型,避免模型单一化。
[Abstract]:Gannan navel orange is an important citrus fruit in our country. In 2015, Gannan navel orange ranked first in the list of brand value of agricultural products in public area. In order to improve the popularity of tourism brand in Gannan area and promote the people of Gannan Su District to get rid of poverty and become rich,Promoting citrus industry transformation and upgrading has played an important role.Up to 2016, the scale of navel orange planting in Gannan is over 1.6 million mu, and the industrial output value is close to 10 billion yuan. The scale of planting, processing and selling is gradually increasing. The consumption of navel orange in Gannan is increasing year by year, and the market scale is gradually expanding, however,The price fluctuation of navel orange in Gannan is violent, and the short term fluctuation is large, which may lead to two extreme conditions: "orange is expensive to injure people" and "orange is cheap to hurt agriculture".The stability of navel orange price in Gannan is related to the income and life of producers, processors, sellers and ordinary consumers.To predict the price fluctuation of navel orange in Gannan is not only beneficial to the formulation of market policy and sales strategy, but also of great significance to ensure the production of producing area and the stable income of farmers.Based on the price formation theory, the monthly price data of Gannan navel orange from 2007 to 2017 and the daily price data from November 2016 to February 2017, the fluctuation characteristics and influencing factors are analyzed.Based on BP neural network, the price prediction of navel orange in Gannan is studied, and the prediction accuracy of BP neural network and ARIMA model is compared. According to the forecast results, the price trend of navel orange in south Jiangxi is analyzed, and the empirical analysis is carried out with MATLAB software.Therefore, the contents and conclusions of this paper are as follows: (1) to explore the present situation and price change of the navel orange industry in Gannan; to analyze the problems faced by the navel orange industry in Gannan from the aspects of price fluctuation, industrial structure and market promotion; to analyze the market factors, natural factors, and so on;The main factors influencing the price of Gannan navel orange were discussed from three aspects of human factors. (2) to predict the price change of navel orange in south Jiangxi.According to the price characteristics of navel orange in Gannan, the suitable number of hidden layer nodes is determined, and the BP neural network price prediction model based on L-M improved algorithm is built.It is found that the prediction error of BP neural network is low and the relative error of monthly price prediction is about 2.5%.By comparing with the prediction results of ARIMA model, the relative error of BP neural network in daily price prediction is 1.5, which is less than 2% of ARIMA model.Based on the result of price forecast and the reason of fluctuation, this paper thinks that the information construction should be perfected, the early warning study of price fluctuation of navel orange should be strengthened, the degree of large-scale operation should be improved, and the production cost should be reduced.In order to improve the risk of price fluctuations in response to the ability.The main innovation of this paper is: (1) in the research content, we adopt the latest price data of navel orange in Gannan to reduce the influence of long time span. In addition, there are more researches on the price prediction of fruits and oranges in academic circles at present.However, there is little research on the price prediction of navel orange in Gannan.In this paper, the intelligent forecasting method, which has been developed rapidly in recent years, has been applied to the price prediction of navel orange in south Jiangxi, which is different from the qualitative analysis in the past.The advantages of BP neural network based on L-M improved algorithm and standard BP neural network, as well as the prediction accuracy of BP neural network and ARIMA prediction model are compared. The suitable prediction model is selected to avoid the simplification of the model.
【学位授予单位】：华中农业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;F323.7

【参考文献】