基于模态分解和时间卷积网络的瓦斯涌出量组合预测  

Combined Prediction of Gas Emergence Based on Modalihj Decomposition and Temporal Convolutional Networks

作　　者：毛智强 徐耀松[1] 王丹丹[1] 田楚汉 黄明宇 MAO Zhiqiang;XU Yaosong;WANG Dandan;TIAN Chuhan;HUANG Mingyu(Faculty of Electrical and Control Engineering,Liaoning Technical University,Huludao Liaoning 125105,China)

机构地区：[1]辽宁工程技术大学电气与控制工程学院,辽宁葫芦岛125105

出　　处：《传感技术学报》2024年第10期1795-1802,共8页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(51974151);辽宁省高等学校创新团队项目(LT2019007)。

摘　　要：为有效地分析和处理煤矿中产生的瓦斯涌出数据,实现精准、可靠的回采工作面绝对瓦斯涌出量预测,以提前规避瓦斯灾害,提出自适应噪声完整集成经验模态分解对瓦斯涌出量序列进行分解,对分解得到的各分量分别构建时间卷积网络模型。利用IGJO算法对TCN模型的相关超参数进行寻优,建立各分量的预测模型。使用Logistic混沌映射生成金豺种群,引入柯西-高斯变异算子,更新金豺位置并选择最优位置,增强算法搜索能力,避免种群陷入局部最优。将各分量的预测输出值叠加,得到最终的瓦斯涌出量预测值。测试结果表明,CEEMDAN-IGJO-TCN组合预测方法,降低了预测的复杂度同时提高了预测精度。In order to effectively analyze and process coal mine gas emission data,which is crucial for maintaining safety in mining opera tions,complete ensemble empirical mode decomposition with adaptive noise is proposed for decomposing the gas emission volume sequence.Once each component,obtained through decomposition,is analyzed,a temporal convolutional networks model is constructed,the purpose of which is to establish prediction models for each component.The Logistic chaos mapping,which is employed to generate a golden jackal population,plays an essential role in the optimization process.When the Cauchy-Gaussian mutation operator is introduced to update the golden jackal positions and select the optimal location,it enhances the search capability of the algorithm and avoids local optima that could hinder accurate predictions.Consequently,the prediction output values of each component are superimposed to obtain the final gas emission volume prediction.Test results indicate that the CEEMDAN-IGJO-TCN combined forecasting method not only reduces the com plexity of the prediction but also improves its accuracy,making it a valuable tool in monitoring gas emissions in coal mines.

关 键 词：瓦斯涌出量预测 经验模态分解 时间卷积网络 金豺优化算法 柯西-高斯变异 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]