基于BP神经网络的数据挖掘方法在需求预测中的应用研究  

Application of Data Mining Technology Base on BP Algorithm for Demand Prediction

作　　者：付斌[1,2] 张吉军[1] 钟健[3] 黄长略 于智博 

机构地区：[1]西南石油大学经济管理学院,四川成都610500 [2]西南油气田公司天然气经济研究所,四川成都610051 [3]西南油气田公司重庆气矿,重庆400021 [4]西南油气田公司社会保险中心,四川成都610051

出　　处：《情报科学》2017年第11期132-135,共4页Information Science

基　　金：国家科技重大专项(2016ZX05062)

摘　　要：【目的/意义】数据挖掘是从庞大数据中挖掘出有潜在价值信息的信息处理技术,它包括神经网络、遗传算法、粗糙集、支持向量机和决策树等多门技术,其中神经网络法具有良好的自学习和含联想储存功能,能够高速寻找优化解,有效提高需求预测准确率。【方法/过程】本文利用BP神经网络具有的优异非线性逼近能力,以我国天然气需求量预测为例,对需求量数据进行训练并采用L-M算法进行改进,最终提高对天然气需求量的预测精度。【结果/结论】实验结果证明,利用数据挖掘中的BP神经网络非线性预测优势能准确捕捉天然气需求预测的变化趋势,为精准预测需求提供了一种有效的工具。【Purpose/significance】Data Mining is an information-processing technology that aims to dig potential valuable information from a huge data base. It includes many kinds of technologies,such as neural network, genetic algorithm（GA）,rough set theory, support vector machine（SVM） and decision-making tree and so on,among which the neural network technology is characterized with favorable self-learning and association storage function that could seek optimizing solution at a high speed and improve the accuracy of the demand prediction.【Method/process】This dissertation therefore takes the quantity demanded for natural gas as an example and applies the excellent nonlinear approximation capability of BackPropagation neural network, trains the required data and applies the L-M algorithm to improve the data. This could finally contributes to improving the prediction precision of natural gas demand.【Result/conclusion】The results show that by applying the BP Algorithm nonlinear prediction in data mining we could accurately capture the trend of natural gas annual demand, it is an effective tool to forecast annual demand and give some advices to government to formulate related policy.

关 键 词：BP神经网络 数据挖掘 需求预测 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] G250.2[自动化与计算机技术—计算机科学与技术]