基于改进BP神经网络的羊肉价格预测  

Prediction of Mutton Price Based on Improved BP Neural Network

作　　者：苏赫 段隆振[1] SU He;DUAN Long-zhen(School of Information Engineering,Nanchang University,Nanchang Jiangxi 330031,China)

机构地区：[1]南昌大学信息工程学院,江西南昌330031

出　　处：《计算机仿真》2020年第4期460-465,共6页Computer Simulation

基　　金：国家自然科学基金资助项目(61070139,81460769)。

摘　　要：在羊肉价格预测问题的研究中,羊肉价格有着严重的非线性、高噪声和影响因素难以确定等特点,高效准确的预测羊肉价格是十分困难的。传统方法对羊肉价格的预测往往主观性较强或过分依赖羊肉价格间的线性关系,导致预测的精度较低,不够准确。针对羊肉价格预测难题及BP神经网络存在的缺陷,提出一种主成分分析与LM(Lvevenberg-Marquardt)算法结合使用的BP神经网络改进模型。首先定性分析影响羊肉价格的因子,然后采用主成分分析方法消除噪声并筛选主要影响因子作为神经网络输入,最后采用基于LM算法的BP神经网络进行训练学习与预测。仿真结果表明,模型的预测值与实际值十分接近,预测精度良好,提高了仿真预测的效率,为羊肉价格的预测提供了一种可行且有效的方法。In the research of the price prediction of mutton, the price is characterized by serious nonlinearity, high noise and uncertain factors, it is very difficult to efficient and accurately predict the price of mutton. The traditional method for predicting the price of mutton is always strong subjectivity or overdependence on the linear relationship between the price of mutton, the prediction accuracy is low and inaccurate. Aiming at the difficult problem of mutton price predict and the defects of BP neural network, a modified BP neural network model based on principal component analysis and LM(Lvevenberg-Marquardt) algorithm is proposed. First qualitative analysis of the factors affecting the price of mutton, then principal component analysis is used to eliminate noise and screen the main influence factors as input of neural network, Finally the BP neural network based on LM algorithm is used for training and prediction. The results of simulation experiments show that the predicted value of the model is very close to the actual value and the prediction accuracy is very good, the training efficiency is improved, it provides a feasible and effective method for predicting the price of mutton.

关 键 词：羊肉价格 主成分分析 神经网络 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]