论文部分内容阅读
针对采空区煤炭自然发火的预测问题,从温度、标志气体浓度以及钻孔参数3个方面选取了8个相关因素,利用Logistic回归分析从中提取出5个相对重要的因素作为预测模型的输入,运用极限学习机算法进行预测,并采用粒子群算法对极限学习机的输入权值及隐含层阈值作优化选取,以提高其泛化能力及预测精度,以此建立了PSO-ELM自然发火预测模型.选用28组训练样本和12组检验样本进行模型的预测实验,结果表明,基于Logistic回归分析筛选指标后的PSO-ELM模型有较高的预测精度,是预测采空区自然发火的一个有效方法.
According to the prediction of spontaneous combustion of coal in goaf, eight related factors were selected from three aspects of temperature, marker gas concentration and drilling parameters. Logistic regression analysis was used to extract five relatively important factors as the input of prediction model. Finally, PSO-ELM spontaneous combustion prediction (PSO-ELM) is used to predict the PSE-ELM spontaneous combustion prediction by using Particle Swarm Optimization (PSO) to optimize the input weights and hidden layer thresholds of the ELI machine to improve its generalization ability and prediction accuracy. Model.Experimental results show that the PSO-ELM model based on Logistic regression analysis shows that the PSO-ELM model has high prediction accuracy and is an effective method for predicting spontaneous combustion of goaf method.