A rapid algorithm for multi-objective pareto optimization of modular architecture

Assigning components and functions to modules early can facilitate faster and higher quality results in the latter stages of development and manufacturing. Methods have been developed to suggest suitable architectures using a variety of metrics that measure the ideality of the modularity. However, different modularity criteria considered are often in conflict with each other and improving one is not achievable without a compromising effect on another. To investigate this, we explore using multiple metrics such as defined in the modular function deployment and consider modularization as a multi-objective optimization problem. We develop here a new multi-objective search algorithm that is able to quickly find non-dominated Pareto-optimal architectures. The algorithm is demonstrated for a cordless vacuum cleaner. We further compare the performance of the algorithm with previous work using the IGTA+ algorithm for multiple-objective clustering. The results show the new algorithm improved the computation time of generating the Pareto-optimal surface by more than two orders of magnitude, reducing the 54 component vacuum cleaner modularization search from 192 hours to 24 minutes.