Research On Tool Wear Status Identification Of GWO-SVM Based On Feature Selection Of Genetic Algorithm

Massive Precision Machining Smart Manufacturing is designed to integrate big data,advanced analytics,high performance computing,and more into traditional manufacturing systems to create high quality,low cost products.The development and implementation of mechanical prediction and maintenance management is critical because wear and failure of mechanical systems or components often results in higher costs and lower productivity.The prediction of tool wear is the most important in cutting process.In this paper,on-line identification of tool wear condition is taken as the research object,constructs recognition model by combining artificial intelligence algorithm through analyzing cutting force signals,training and verifying the validity of the model with the experimental data.It has important application value for the research of transforming the traditional manufacturing system into the intelligent manufacturing system with artificial intelligence.This paper mainly carries on the following research:(1)Analyze and clarify the method and framework of the tool wear state recognition model studied in this paper,including feature extraction module,feature selection module,Gaussian segmentation module and state recognition module,and introduce the specific content of each module.It lays a theoretical foundation for the on-line identification of tool wear.(2)A milling experiment platform is built and the tool wear experiment is carried out.The wear form of the tool during the milling process is studied,and a Gaussian segmentation method is proposed,which is successfully applied to the division of the tool wear stage,which provides a basis for the output value of the tool wear state recognition model.(3)Using the experimental analysis,the key techniques for determining the input of the tool wear state identification model are elaborated.The extracted milling force signals are extracted in the time domain,frequency domain and time-frequency domain,and 84 eigenvalues are extracted.In order to reduce the dimension of input features,14 typical features are selected as input vectors of subsequent wear state recognition model based on genetic algorithm.(4)The Gaussian segmentation wear data is used as the output,and the genetic algorithm(GA)feature selection data is input.The multi-classification model of the grey wolf optimization support vector machine(GWO-SVM)is established using the Gaussian segmentation wear data as the output and the genetic algorithm(GA)feature selection data as input.The results show that compared with the single model and other combined models,the proposed combined model(GA-GWO-SVM)can better extract tool wear information and has higher recognition accuracy.