基于机器学习机床机械加工特征信息与加工材料关联性研究

Research on the Correlation Between Machining Feature Information of CNC and Machining Materials Based on Machine Learning

为了开展数控机床机械加工特征信息检测与分析方面的研究,以健康状态的机床对不同工件材料进行切削加工,通过数据采集系统采集在不同切削状态下机床主轴输出的机械特征信息,利用机器学习的方法对特征信息进行分析和判断,提出一种基于机床主轴振动信号与机床主轴负载电流特征信息融合的工件材料精确识别判断模型。首先,获取机床在不同加工状态下的主轴振动信号以及主轴负载电流信号,利用变分模态分解(VMD)算法对其进行分解获得本征模态分量(IMF)并计算各个IMF的多尺度加权排列熵(MWPE)进行信息融合构建特征向量;然后使用灰狼优化(GWO)算法对传统支持向量机进行优化并对4种常见工况进行识别判断。试验结果表明:基于信息融合的特征提取与GWO-SVM相结合的方法能够利用机床加工状态输出的数据特征信息对正在加工的材料种类进行精确识别判断。

In order to carry out the research on the detection and analysis information of CNC machine tools,the machine tool in a healthy state was used to cut different workpiece materials,the data signal acquisition system was used to collect the mechanical characteristic information output in different cutting states,and the machine learning method was used to analyze and judge the characteristic information,an analysis and judgment model of workpiece material based on the fusion of machine tool spindle vibration signal and machine tool spindle load current characteristic information was proposed.Firstly,the spindle vibration signal and the load current signal of the machine tool were obtained,the algorithm of variational mode decomposition(VMD) was used to decompose the acquired signal under different processing condition,the multi-scale weighted permutation entropy(MWPE) of each acquired intrinsic mode function(IMF) was calculated for information fusion.Finally,the grey wolf optimization(GWO) algorithm was used to optimize the traditional support vector machine,and four common operating conditions were identified.The result shows that the method of combining feature extraction based on information fusion and GWO-SVM can accurately identify and judge the types of materials being processed by using the data feature information output of the machine tool processing state.