基于LSSVM与CPSO的瓦斯涌出量组合预测

Combination Prediction of Gas Emission Amount Using LSSVM and CPSO

为了提高矿井瓦斯涌出量的预测精度,提出了一种基于最小二乘支持向量机(LSSVM)与经典粒子群优化算法(CPSO)的矿井瓦斯涌出量非线性组合预测方法。该方法应用最小二乘支持向量机建立了一个多输入单输出的瓦斯涌出量非线性组合预测模型,对3个常用的单项预测数据进行非线性组合并作为最终预测结果,模型的参数由经典粒子群算法与学习样本的平均绝对误差最小原则进行智能优化。实验结果表明,所提出方法最大绝对误差为0.0183m3·t-1、平均绝对误差为0.0039m3·t-1,达到了高精度预测的要求,对矿井瓦斯涌出量预测研究具有较好的实用价值。

In order to improve the forecast accuracy of gas emission amount, a combined forecasting method using Least Square support vector machine （LSSVM） and canonical particle swarm optimization （CPSO） was proposed. LSSVM is used to build a MISO （multi--input and single--output） non--linear forecasting model. The parameters of model are intelligently optimized by both training samples set and the principle of mean absolute error （MAE） minimization, to combine 3 original predition as final predicting result. The experiment shows that the method can improve the forecast accuracy of gas emission amount with the Maximum absolute error is 0. 0183 m^3· t^-1 and Mean absolute error is 0. 0039 m3 · t^- 1 , which is much higher accuracy than common methods and has good practical value.