In recent years, the needs associated with the development of new technologies in the manufacturing industry that utilize big data typified by deep learning and artificial intelligence (AI) have been increasing. The mold manufacturing industry satisfies the needs of customers, so that the shapes, the purposes of use, and the material types are diverse. Accordingly, more complicated designs can be manufactured. Many industrial products, such as dies, are designed with computer aided design (CAD), and the tool path (cutter location: CL) adopted until the desired dimensions are created by computer-aided manufacturing (CAM) are output in the form of a numerical control (NC) program. In reference to Japanese technologies, these processes can be performed relatively easily and automatically by non-experts with limited knowledge. In this case, the program output is input to the NC machine tool that creates tools with various shapes and purposes, holders that hold them, and machining using a jig that fixes the work material. As mentioned above, recent CAM systems have evolved so that even unskilled technicians can create the tool position information (CL) in the form of an NC program in a relatively easy manner. For example, kinematic information, such as the cutting speed V \[m/min\] and the feed rate per tooth f \[mm/tooth\] cannot be determined automatically. In other words, the cutting conditions of the NC machine tools often rely on tacit knowledge, and the knowledge and experience of skilled technicians, and are generally not systematized. Therefore, many on-site technicians often set conditions based on the recommended cutting conditions listed in the tool catalog. However, given that catalogs contain large-scale data on machining technology based on accumulated knowledge on tools from various tool manufacturers, conditions may vary extensively depending on the tool shapes and work materials. Accordingly, experience and tips on the selection of the setting conditions are based on trial-and-error, and are thus time consuming. Therefore, the database entries can be used as reference data. Correspondingly, these data constitute an implicit knowledge source for the determination of the cutting condition that summarizes the experiences of the field and the tool manufacturer.  
In recent years, a method called data mining has been developed and studied extensively that can extract new scientifically valuable knowledge, patterns, and relationships from accumulated large-scale data. Data mining methods composed of a number of statistical analysis methods have been used in manufacturing science, chemistry, and business administration. Therefore, in our previous research, we focused on tool catalog data that contained a vast amount of information—including the years of knowledge and experience of the field and tool manufacturers—and placed special emphasis on the clustering of quantitative variables that described these data. We have proposed catalog mining that applies a mining method. In this report, we focus on the cutting width and cutting speed as the cutting conditions, and use a random forest machine learning method to predict the cutting conditions of the radius endmill (used as an objective variable). Based on these, we constructed and made predictions. We also tried to visualize the process of the determination of the cutting conditions with the use of a decision tree method, a type of statistical analysis method.


Data Mining From Endmill Tool Catalog Information Based on the Use of a Machine Learning Method

Division / Sponsors: 
Design Engineering Division and Computers & Information In Engineering Division


Paper Proceedings: https://asmedigitalcollection.asme.org/IDETC-CIE/IDETC-CIE2020/volume/83907

IDETC2020 - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.