基于核主成分和灰狼优化算法的刀具磨损状态识别

Tool wear condition recognition based on kernel principal component and grey wolf optimizer algorithm

为了提高刀具磨损状态实时监控的准确性和泛化能力,提出一种基于切削力特征间接识别刀具磨损状态的方法。该方法建立了切削力信号与刀具磨损的非线性映射关系,并进行刀具生命周期内的性能试验,采集切削力信号,提取信号的时域、频域和小波域特征,采用核主成分分析法进行数据降维,利用灰狼优化的支持向量机得到刀具磨损的分类等级。最后与其他文献中的方法进行对比,结果表明该模型能够更准确地反映刀具的磨损状态,具有较高的泛化能力。

To improve the accuracy and generalization ability of real-time monitoring of tool wear state,a method based on cutting force characteristics to indirectly identify tool wear state was proposed,which established a non-linear mapping relationship between cutting force signal and tool wear.Tool performance test in tool life cycle was operated to collect cutting force signals,and the time domain,frequency domain and wavelet domain features of signals were extracted.The dimensionality reduction features of the data with kernel principal component analysis were evaluated,and the classification grade of tool wear was obtained by using the support vector machine classifier optimized of grey wolf optimization algorithm.Compared with other methods in the literatures,the experimental results showed that the proposed model could reflect the tool wear state more accurately and had higher generalization ability.