A cost-effective planning method for automatic measurement based on task similarity and octopus optimization

Abstract

Measurement planning is vital for automatic and digitalized aircraft assembly, and in particular, for the key execution sequences. In this article, a novel planning method that considers both order and efficiency is proposed. General rules are defined for the analysis of measurement tasks and their elements. A Laplace kernel function is then utilized for similarity quantification, and based on iterative two-step elementary transformations of the similarity matrix, the clustering of measurement tasks is achieved. A virtual task is proposed to bridge task clusters, and the preliminary sequences are obtained naturally. Given that the adjacent association results may not be the most efficient, octopus optimization is proposed to solve this special case of the traveling salesman problem. The measurement tasks and clusters are shops and cities, respectively. The octopus has multiple tentacles that traverse all shops, and each tentacle carries one feasible execution order. Experiments performed using aircraft inspection indicate that the proposed method can output multiple orderly measurement sequences. Compared with the preliminary sequences, the efficiency is obviously improved in terms of a decrease in the total measurement time. In addition, the optimal sequences are more even, and the duty ratio of the measurement device is decreased.