【摘要】 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.更多还原

【Abstract】 To predict coal spontaneous combustion in caving zone, 8 relevant factors were selected from three aspects of temperature and indicator gases and drilling parameters. Logistic regression analysis wai\s used to extract 5 important parameters to be input. To improve generalization ability and prediction accuracy of extreme learning machine, particle swarm optimization algorithm was applied to optimize the input connection weights and hidden layer threshold of ELM,and PSO-ELM forecasting model was built. Experiment was performed on 28 groups of training data and 12 groups of testing data. The result shows that PSO-ELM forecasting model based on logistic regression analysis has high prediction accuracy, and it is an effective method for completing the prediction of coal spontaneous combustion.更多还原