08.基于LSSVM的啤酒企业能耗预测方法研究

基于LSSVM的啤酒企业能耗预测方法研究

胡冬梅¹牛国成^1*白晶¹张凤晶²

（1北华大学电气信息工程学院 吉林吉林 132021

2 空军航空 大学特种专业系 吉林 长春130022）

关键词　最小支持向量机；能耗预测；交叉验证；网格搜索

中图分类号：TQ083+.4              文献标识码：A

摘　要：为了提高生产效率与能源利用率，提出了采用最小二乘支持向量机（Leastsquares support vector machines LSSVM）算法对啤酒厂煮沸车间蒸汽量的消耗预测。根据历史数据样本，采用RBF核函数作为LSSVM的核函数，交叉验证结合网格搜索来优化参数，用所建立决策函数作为预测模型。实验结果表明，数据样本的预测集和测试集的均方差均达到0.0060，拟合相关参数达到99%以上，LSSVM方法能够快速准确的预测该车间在生产旺季的蒸汽消耗，为企业节能控制方案的制定提供了理论依据。

Research onenergy consumption forecasting method of beer enterprise using theLSSVM model

HU Dong-mei¹,NIU Guo-cheng¹, BAIJing¹， ZHANGFeng-jing²

(¹ College ofElectrical and Information Engineering，Beihua University，Jilin132021，China)

（²Special Information Department，Air ForceAviation University , Changchun 130022，China）

Keywords Leastsquares support vector machines (LSSVM); Energy consumption forecasting;cross validation; grid search

AbstractLeast squares support vectormachines (LSSVM) algorithm was proposed to predict the steam consumption in theboiling workshop of the brewery to improve production efficiency and energyutilization ratio. According to samples of historical data, the RBF kernelfunction was used as the kernel function of LSSVM and cross validation combinedwith grid-search was used to optimize the parameters, with the establisheddecision function being the forecasting model. Experimental results show thatMSE of the training set and prediction set reaches 0.0060；correlation coefficientsreached above 99%. LSSVM method can quickly and accurately predict the energy consumptionof the plant in the busy season, providing a theoretical basis for theenterprise energy saving control scheme.