基于大数据挖掘的数控机床多工况载荷谱系研究

项目名称： 基于大数据挖掘的数控机床多工况载荷谱系研究

项目编号： No.51605173

项目类型： 青年科学基金项目

立项/批准年度： 2017

项目学科： 机械、仪表工业

项目作者： 武滢

作者单位： 华中科技大学

项目金额： 20万元

中文摘要： 载荷谱是数控机床可靠性分析的基础。数控机床载荷谱系包括机床各功能部件在各种工况下、不同切削状态下的载荷谱，是数控机床系统可靠性失效相关分析和多状态分析的基础。传统小样本下的载荷谱分析难以将各载荷特征对机床可靠性的影响进行详细分析，难以表征各类信号与功能部件性能状态之间的复杂映射关系。本项目应用大量传感器采集机床实际生产中的载荷包括切削力、温度、振动等数据，采用深信度神经网络方法，结合全寿命周期数据，对切削力-时间历程中的极值载荷、静态载荷、动态载荷进行自识别，挖掘切削力多尺度载荷特征与机床性能之间的映射关系；对切削力、温度、振动等载荷数据进行相关性分析，提出反映不同载荷特征参数动态相关性的相关函数。建立多工况下反映不同加工参数、不同性能状态包含多源载荷信息的数控机床的载荷谱系，为数控机床可靠性分析和优化设计提供依据，为数控机床的复杂工况识别及预测奠定基础。

中文关键词： 数控机床；载荷谱系；大数据挖掘；可靠性分析

英文摘要： Load spectrum is the basis of the reliability analysis of CNC machine tools. The establishment of the load spectrum series in various conditions and different cutting conditions is the basis of dependent failure analysis and multiple status of system reliability of machine tools. Traditional load spectrum analysis using small sample cannot describe the effect of load characteristics on the reliability of machine tool, it is difficult to characterize the complex mapping relations between all kinds of signals and performance status of components. A large number of sensors to collect actual load such as cutting force, temperature and vibration data are applied; using neural network method, combined with the whole life cycle data, the maximum load and static load and dynamic load in the cutting force -time history are recognized. The mapping relationship between multi-scale load characteristics of cutting force and machine performance is mined. Analyzing the dependency between cutting force, temperature and vibration, dynamic correlation function is proposed. To build the load spectrum series reflecting different processing parameters, different performance statuses provide evidence for reliability analysis and optimization design of CNC machine tools.

英文关键词： CNC machine tool;load spectrum series;big data mining;reliability analysis