摘要
为了提高粮食产量预测的准确性,针对我国粮食产量受到多因素影响且呈非线性关系的特点,提出主成分分析和极限学习相结合的粮食产量短期精准预测方法。首先计算各影响因素与粮食产量之间的相关系数,利用主成分分析方法构建影响粮食产量的主要成分,从而降低影响因子的维度。其次,根据粮食数据样本集少、输入变量与输出变量间呈非线性关系的特点,构建粮食产量的极限学习机预测模型并对模型参数进行比较优化。最后,通过对实际粮食产量数据的应用以及与其它预测方法的比较,研究模型的预测精度。结果表明,基于极限学习机的粮食产量预测模型的3年及5年预测误差分别为1.9%和2.08%,相比于BP神经网络模型和多项式拟合模型而言,预测精度大幅提高,能够实现粮食产量的短期精准预测。
To improve the prediction accuracy of the grain production, the model based on the combination of the principal component analysis （PCA） and extreme learning machine （ELM） is put forward to forecast the grain yield accurately in short time. Firstly, the correlation coefficient between the influence factor and the grain yield is calculated, and the main components are extracted by using the PCA, which reduced the amount of the input factors. Then, considering that the data in the sample is limited and the relationship between the inputs and out- put is nonlinear, the ELM model is established and the parameters of which is optimized. Lastly, the grain yield is estimated by the model and the prediction accuracy is compared with other methods. The results shown that the grain production can be estimated 3 years or 5 years in advance, and the prediction error is 1.9% and 2.08% respectively, which is more accurate than the methods of BP network and polynomial fitting, and can be used to predict the grain production accurately in short time.
出处
《粮食加工》
2017年第2期1-5,共5页
Grain Processing
基金
河南省科技攻关项目"河南省粮食产量预测模型研究及数字化平台开发"(162102210198)
国家粮食公益项目"数字化模拟粮食储备辅助决策体系及技术研究"(201413001)
关键词
粮食产量
主成分分析
极限学习机
预测模型
grain yield, principal component analysis, extreme learning machine, prediction model
