基于因子分析法的瓦斯涌出量预测指标选取

摘要

为解决瓦斯涌出量预测过程中存在的预测指标过多而导致预测精度降低的问题,构建因子分析与BP神经网络相结合的瓦斯涌出量预测模型.采用SPSS因子分析法对瓦斯涌出量影响因素进行了分析降维,并对BP神经网络模型进行训练及预测.结果表明:因子分析能使BP神经网络的输入变量从10个降为3个有实际含义的因子,经因子分析后预测模型的预测速度及精度均高于未处理的样本数据,预测性能明显改善,其平均误差为3.8％,最大误差为4.9％,表明所采取瓦斯涌出量预测指标的选取方法是可行和有效的.%In order to solve the problem of redundant prediction index which leads to low prediction precision in the gas emission prediction,a gas emission prediction model was built by factor analysis combined with BP neural network.SPSS factor analysis method was used to reduce the dimension of the gas emission factors,and the model was trained and predicted.The results show that the input data of BP neural network can be reduced from 10 to 3 with practical meaning by factor analysis,and the prediction speed and calculation accuracy are higher than those of the neural network having not underwent factor analysis,with the averaged biases of 3.8％ and the maximum error of 4.9％.It is indicated that the index selection of gas emission prediction is feasible and effective.