基于改进GS-SVM的煤炭生产成本预测  

Prediction of Coal Production Cost Based on Improved Grid Search and Support Vector Machine

作　　者：何银银[1] 

机构地区：[1]辽宁工程技术大学工商管理学院,葫芦岛125105

出　　处：《世界科技研究与发展》2016年第3期701-705,723,共6页World Sci-Tech R&D

基　　金：国家科技支撑计划(2013BAH12F01);国家自然科学基金(51374121)资助

摘　　要：原煤生产成本同时受到多种因素的共同影响,导致原煤生产成本系统具有非线性、多维性等特点。为了对原煤生产成本进行更加科学、准确的预测,针对目前我国原煤生产成本预测中存在的问题,将支持向量机(SVM)引入到原煤生产成本预测中。为快速准确地选取支持向量机参数,在传统网格搜索(GS)算法基础之上提出了一种改进网格搜索算法,并建立了一种基于改进GS-SVM的煤炭生产成本预测模型。将该模型用于观台煤矿原煤生产成本预测中,模型预测误差均在5%以下,平均误差3.3673%,预测精度高于多元回归分析,而模型训练时间也远低于传统网格搜索算法和启发(粒子群)算法,能够满足实际原煤成本预测需求。The production cost of raw coal is influenced by many factors,and the production cost of raw coal is nonlinear and multi-dimensional.In order to make a more scientific and accurate forecast of raw coal production cost,the support vector machine（ SVM） is introduced into the prediction of coal production cost.In order to select the support vector machine parameters quickly and accurately,an improved grid search algorithm is proposed based on the traditional grid search（ GS）algorithm,and an improved coal production cost forecasting model based on improved GS-SVM is established.The model is used to forecast the cost of coal production in Guantai coal mine.The model prediction error is less than 5%,the average error is 3.3673%,the forecast precision is higher than that of multiple regression analysis.The model training time is far lower than the traditional grid search algorithm and heuristic algorithm parameters optimization method,which can meet the actual demand of coal cost forecast.

关 键 词：原煤生产成本 支持向量机 改进网格搜索 预测 

分 类 号：F406.72[经济管理—产业经济] F426.21