Computer-Aided process planning group technology and delayed product differentiation for hybrid manufacturing using a hybridized machine learning and optimization approach

This research introduces a novel and integrative methodology for addressing the complexity inherent in hybrid manufacturing processes, offering potential benefits in terms of cost efficiency and customized production for highly variable product families. Specifically, this paper addresses the challenge of developing efficient process plans for hybrid manufacturing systems that embrace delayed product differentiation and multiple manufacturing strategies. In a dynamic market where product variants' demand is volatile, managing these variations becomes a critical issue. To tackle this, a two-phase hybrid optimization machine learning approach is proposed. In the optimization phase, a mathematical programming model is formulated to minimize overall production costs, encompassing both value-added and non-value-added activities. The model determines the processing of features forming the platform and the variants as well as the respective setups. The machine learning phase leverages the Logical Analysis of Data (LAD) algorithm to distinguish between high-quality and low-quality solutions generated by the optimization model. LAD's generated patterns can be used to guide the construction of favorable solutions and avoid non-favorable solutions, especially for challenging problem instances. An illustrative example is presented to explain the proposed hybrid approach, and a case study of a complex part family is demonstrated to validate the proposed approach. A simulation study is conducted on an extended industrial part to evaluate the performance of the proposed methodology on real parts.