关键词: 人体姿态库, 花朵授粉算法, 乌鸦搜索算法, 聚类算法, 维修性验证

Abstract: For the poor task pertinence of present construction methods of human body posture examples in the simulation verification of equipment maintainability, a construction method of human body posture examples based on improved flower pollination algorithm (IFPA) and K-means is proposed. An IFPA is proposed by introducing the crow search algorithm (CSA) into FPA in order to improve the optimization accuracy of FPA. Test results show that IFPA can achieve wider search space and higher optimization accuracy compared with particle swarm optimization (PSO) and FPA. In order to avoid K-means falling into locally optimal solutions easily, the calculation steps of IFPA are designed by proposing an individual encoding method and defining an individual operation function, and IFPA is combined with K-means to deal with the clustering problem of maintenance postures of maintenance staff. The central data of each cluster is extracted, and it is taken as the input data of part activity of human body posture examples. And the human body posture examples having task pertinence for equipment maintainability simulation verification are constructed. Test results show that the proposed construction method for the human body posture examples based on PSO and K-means has better optimization accuracy and clustering effects compared with the other five construction methods, and it is suitable for the engineering practice of equipment maintainability simulation verification. Key

Key words: humanbodypostureexample, flowerpollinationalgorithm, crowsearchalgorithm, clusteringalgorithm, maintainabilityverification