多模型软测量的顶板来压预测模型  

Study on the pressure prediction model based on soft measurement and multiple model

作　　者：郝秦霞[1] 张金锁[2] 张光耀 邢书宝[4] 

机构地区：[1]西安科技大学通信与信息工程学院,陕西西安710054 [2]延安大学管理学院,陕西延安716000 [3]陕西南梁矿业有限公司,陕西西安719400 [4]西安科技大学管理学院,陕西西安710054

出　　处：《西北大学学报（自然科学版）》2015年第5期716-720,共5页Journal of Northwest University（Natural Science Edition）

基　　金：国家软科学研究计划基金资助项目(2013GXS4D151);陕西省教育厅专项基金资助项目(14JK1457)

摘　　要：为了对中国煤矿顶板安全进行合理的预测,建立有效的预警机制,提出采用多模型软测量的来压预测方法,方法中首先利用EEMD方法对非线性、非平稳来压监测信号进行模态分解,得到多个IFM;其次根据IFM的特点,对非线性的固有IFM序列运用SVM方法进行预测,对于线性趋势项序列运用ARIMA进行预测,最终将各预测值合成重构得到系统模型的预测输出。实测数据分析表明,基于EEMD-SVM-ARIMA软测量的来压预测模型具有较高的精度,能很好地反映来压变形规律,能满足安全生产的需求。The paper analyzes the currently problems of pressure prediction in the case of deep mining,the increasingly intensified roof pressure,the deterioration of roadway maintenance and the increased frequency of coal mine accidents,and then puts forwards a pressure prediction method based on soft measurement and multiple models. First,non-linear,non-stationary signals of pressures are mode decomposed through EEMD,resulting in multiple mode function sequence（ IFM）,then the paper predicts the non-linear IFM original sequence via support vector machine according to the characteristics of model function sequence. While for the linear parts,the trend of the sequence,the paper applies auto-regressive integrated moving average model（ ARIMA） to conduct the linear prediction. At last,the system prediction output results from the reconstruction of each prediction synthesis. The analysis of the measured data shows that the prediction model based on EEMD-SVM-ARIMA has obtained a higher accuracy,which is a better reflection to pressure deformation law and meets the needs of safe production.

关 键 词：顶板来压 EEMD ARIMA SVM 预测 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]