Algorithms for the q-model clustering problem with application in switching cabinet manufacturing

Abstract

The model configuration problem (MCP) is a combinatorial optimization problem with application in the telecommunications manufacturing industry. The product is a switching cabinet, defined by a number of positions (slots) in which specific circuit packs are installed according to the customer requirements (configurations). Variety of customer requirements leads to a relatively large number of distinct configurations. In order to streamline the manufacturing process, a large number of switching cabinets with identical configurations (model cabinets) are produced in advance. A customer order is then filled by selecting a model cabinet whose configuration is relatively close to the customer configuration and performing any necessary circuit pack exchanges to make its configuration identical to the customer requirement. The manufacturing costs are proportional to the number of these circuit pack exchanges, and the q-model configuration problem is to design q different model configurations so as to minimize the total number of exchanges for a given collection of customer orders. We propose three heuristic algorithms for solving the q-model configuration problem and carry out a computational experiment to evaluate their effectiveness.