基于PCA-LWCA-LS-SVM回采工作面瓦斯涌出量预测  

PREDICTION OF GAS EMISSION IN WORKING FACE BASED ON PCA-LWCA-LS-SVM

作　　者：张瑞 李雅梅[1] Zhang Rui;Li Yamei(Faculty of Electrical and Control Engineering,Liaoning Technical University,Huludao 125105,Liaoning,China)

机构地区：[1]辽宁工程技术大学电气与控制工程学院,辽宁葫芦岛125105

出　　处：《计算机应用与软件》2018年第12期66-70,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(71371091);辽宁省重点实验室项目(LJZS003)

摘　　要：回采工作面瓦斯涌出量各影响因素之间存在多重共线性,这会对预测造成不利影响。为降低这种影响并提高预测精度,提出将主成分分析法(PCA)与双层狼群算法(LWCA)优化的最小二乘支持向量机(LS-SVM)相耦合。该方法利用PCA对数据进行降维处理,再利用LS-SVM泛化能力强的特点来进行瓦斯涌出量的预测。为了提升LS-SVM的性能,将LWCA与LS-SVM结合,利用LWCA来优化LS-SVM的参数。采用该模型对实际中的回采工作面瓦斯涌出量进行预测,并与LS-SVM预测模型、PCA与遗传算法优化的LS-SVM相耦合的预测模型进行比较,验证该模型的性能。实验表明:该模型预测结果中最大相对误差为2. 35%,最小相对误差为0. 26%,平均相对误差为1. 22%,预测精度得到显著提升。There are multiple collinearities among the influencing factors of the gas emission in the working face, and it has a negative effect on the prediction. In order to reduce the effect and improve the accuracy of prediction, the principal component analysis（PCA） was coupled with the least square support vector machine（LS-SVM） optimized by the two-layer wolf colony algorithm（LWCA）. The method used PCA to reduce the dimension of data. The volume of gas emission was predicted by using the generalization ability of LS-SVM. To improve the performance of LS-SVM, LWCA was combined with LS-SVM, which optimized the parameters of LS-SVM. The model was adopted to forecast the actual gas emission in the working face, and was compared with LS-SVM prediction model, PCA and LS-SVM prediction model optimized by genetic algorithm to verify the performance of the model. The experiment shows that the maximum relative error is 2.35%, the minimum relative error is 0.26%, the average relative error is 1.22% in the predicted results, and prediction accuracy is significantly improved.

关 键 词：主成分分析 最小二乘支持向量机 领导狼狼群算法 数据降维 回采工作面 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP712[自动化与计算机技术—计算机科学与技术]