Investigations of sustainable machining and ultra-precision machining using social network analysis and unsupervised learning

Abstract

The manufacturing industry is one of the main contributors to greenhouse gas emissions and the consumption of natural resources. With an increasing demand for high technological products nowadays, the natural resource consumptions from ultra-precision machining (UPM) predictively go up and therefore sustainable machining has become progressively important in terms of reducing the induced negative environmental issues. On the other hand, sustainability includes three dimensions: economic sustainability, environmental sustainability, and social sustainability. However, the complicated interactive connections among the sustainable items of these three dimensions regarding UPM cause infeasibilities and difficulties to execute sustainable UPM practically. Up to now, it still lacks studies about the solutions or measurement for resolving interactive relationships between sustainable items of UPM. Therefore, it is necessary to develop a model to evaluate the sustainable UPM parameters considering their connections. In this study, the influential parameters of sustainable machining and their interactive relationships were identified. Several centrality metrics of social network analysis (SNA) were utilized to evaluate the role of each parameter from a systemic view. What's more, the calculation results of centrality metrics can be utilized as the feature data of some parameters to predict the feature of other parameters by using machine learning algorithms. By these steps, the SNA - machine learning model can be established for sustainable parameters investigation. Moreover, with the information from the obtained main metrics results, some managerial implications for improving the sustainability level of the UPM process were raised. In this study, it is the first time the SNA method was introduced in the research area of sustainable manufacturing and UPM. And the unsupervised learning approach was also applied firstly to classify the centrality metrics results. Moreover, the roles and importance of sustainable manufacturing and UPM parameters have been evaluated to help companies to achieve optimal settings of them. By using link prediction metrics of SNA, the potential values of the undiscussed relationship between two sustainable machining factors in previous studies can be discovered to support the researchers to do the topic selection. In the previous research on sustainable manufacturing, only a few factors were studied to find out their impacts on the overall sustainability level. It still lacks a study that can analyze the importance of various sustainable machining parameters in the same model with considering their influencing relationships. In this work, by using the method of SNA, this research gap can be resolved. Based on the evaluation of sustainable manufacturing factors, it was found that cutting quality is the parameter with the highest value of the centrality index, which is the overall measurement of centrality. It indicates that cutting quality should be considered as the key factor in the manufacturing system. In the case study of the UPM optimal setting, Material recovery was discovered as the UPM parameter which has the highest betweenness result, which shows that it performs as a gatekeeper to collect the impacts from the upstream UPM nodes and can be observed before getting machining outcomes. Thus, it plays a key role as one significant indicator for researchers to obtain optimized UPM output. Besides, the influencing relationships among the sustainable machining parameters were selected based on relevant literature. As these relationships are performed as the "edges" in the SNA model, two non-adjacent nodes mean that their relationship is not discovered in current literature. In this work, the link prediction metrics were used to find out the probability of the existence of the hidden relationships between these two parameters. Therefore, this work can determine the undiscussed latent relationships among sustainable machining parameters with high potential values to be investigated. Thus, this work provides a reference for developing more research topics in the sustainable machining area.