基于链式多重插补的WOA-ELM煤与瓦斯突出预测模型

WOA-ELM prediction model of coal and gas outburst based on multiple imputation by chained equations

为了提高缺失数据下煤与瓦斯突出预测准确率,提出1种基于链式支持向量机多重插补(MICE_SVM)的鲸鱼优化算法(WOA)-极限学习机(ELM)预测模型,以淮南朱集矿区为例,选取5个煤与瓦斯突出影响指标作为模型特征,采用提出的MICE_SVM算法插补突出事故数据中缺失值,利用WOA优选ELM输入层权值及隐含层阈值,构建煤与瓦斯突出预测模型,将插补后数据用于WOA-ELM模型的训练与测试,并与其他模型的预测效果对比。研究结果表明:MICE_SVM插补前、后的有突出数据预测准确率分别为83.02%,90.41%,MICE_SVM显著提高了有突出预测准确率,对无突出和整体的预测准确率提高不明显;数据插补后WOA优化ELM对无突出、有突出和整体的预测准确率分别为97.94%,96.25%,96.48%,较优化前分别提高了5.79%,5.84%,5.55%,数据插补后WOA-ELM为最佳预测模型。

In order to improve the prediction accuracy of coal and gas outburst under missing data,a whale optimization algorithm(WOA)-extreme learning machine(ELM)prediction model based on the chained support vector machine(MICE_SVM)was proposed.Taking Zhuji mining area in Huainan as an example,five indicators influencing coal and gas outburst were selected as the model characteristics,then the proposed MICE_SVM algorithm was used to interpolate the missing values of the outburst data,and the WOA was used to optimize the input layer weights and hidden layer thresholds of ELM.A prediction model of coal and gas outburst was constructed,then the interpolated data were used to train and test the WOA-ELM model,and the prediction effect was compared with other models.The results showed that the prediction accuracy of outburst data before and after MICE_SVM interpolation was 83.02%and 90.41%,respectively,and MICE_SVM significantly improved the prediction accuracy of outburst data,but the increase in the non-outburst data and whole prediction accuracy was not obvious.After the data interpolation,the prediction accuracy of WOA optimized ELM for non-outburst,outburst and whole was 97.94%,96.25%and 96.48%,respectively,which increased by 5.79%,5.84%and 5.55%than that before optimization,so the WOA-ELM after data interpolation was the best prediction model.