面向磨机负荷参数预测的多通道机械信号分析评估与优化组合  

Multi-channel Mechanical Signal Analytical Estimation and Optimal Combination for Mill Load Parameter Forecasting

作　　者：汤健 乔俊飞 刘卓 周晓杰 余刚 赵建军 TANG Jian;QIAO Junfei;LIU Zhuo;ZHOU Xiaojie;YU Gang;ZHAO Jianjun(Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China;Beijing Key Laboratory of Computational Intelligence and Intelligent System,Beijing 100124,China;State Key Laboratory of Synthetical Automation for Process Industries,Northeastern University,Shenyang 110004,China;State Key Laboratory of Process Automation in Mining&Metallurgy,Beijing 100089,China;Beijing Key Laboratory of Process Automation in Mining&Metallurgy,Beijing 100089,China)

机构地区：[1]北京工业大学信息学部,北京100124 [2]计算智能与智能系统北京市重点实验室,北京100124 [3]东北大学流程工业综合自动化国家重点实验室,沈阳110004 [4]矿冶过程自动控制技术国家重点实验室,北京100089 [5]矿冶过程自动控制技术北京市重点实验室,北京100089

出　　处：《北京工业大学学报》2020年第9期997-1007,共11页Journal of Beijing University of Technology

基　　金：国家自然科学基金资助项目(61703089,61803191);矿冶过程自动控制技术国家重点实验室和矿冶过程自动控制技术北京市重点实验室开放课题(BGRIMM-KZSKL-2018-06)。

摘　　要：针对球磨机系统中所采集的多源(多通道)机械信号在不同运行工况下的灵敏度具有差异性和蕴含磨机负荷参数(mill load parameter,MLP)信息存在冗余性与互补性等问题,提出了一种面向MLP预测(MLP forecasting,MLPF)的多运行工况多通道机械信号分析评估与优化组合方法.首先,基于多通道机械信号构建多运行工况下的数据驱动MLPF模型,采用验证数据集获取模型结构参数和验证误差;然后,定义对多通道机械信号预测性能和蕴含信息贡献率进行分析与度量的综合评估指标,并基于该指标结合预设定阈值进行有价值通道信号的初次选择;最后,对初选后的多通道机械信号进行优化组合以获得具有最佳预测性能的MLPF模型.通过实验球磨机的多通道机械信号仿真验证了所提方法的有效性.The sensitivity of multi-source(multi-channel)mechanical signals of ball mill system at different working conditions is different.Moreover,the valuable information for constructing mill load parameters(MLP)in them is also diverse.Aim at these problems,an optimal combination method for MLP forecasting(MLPF)and an analytical estimation of multi-channel mechanical signal at multiple working conditions were proposed.First,the data-driven MLPF models under multi-condition based on multi-channel mechanism signals were constructed,whose structure parameters and validation errors are calculated based on validation dataset.Then,a new combined evaluation index was defined to analyze and measure the prediction contribution ratio and hidden information contribution ratio of the multi-channel mechanical signals.By using the combined evaluation index and pre-set threshold,the initial selection of the multi-channel signals was achieved.Finally,these selected multi-channel mechanical signals were optimally combined and the best combination was used to construct MLPF model.The simulation results based on multi-channel mechanical signals of a laboratory scale ball mill verified the effectiveness of the proposed method.

关 键 词：磨机负荷参数预测 多通道机械信号 频谱数据 优化组合 评估指标 运行工况 

分 类 号：U461[机械工程—车辆工程] TP308[交通运输工程—载运工具运用工程]