关键词: 复杂对象, 改进型粒子群算法, 计算消耗, 局部极小值, 复合评价

Abstract:

A composite evaluation strategy for complex object parameter identification problem is proposed. By hierarchically design optimized index evaluation function, the requirement of anti-local minimal ability for iterative search algorithm is reduced. Identification accuracy is significantly improved while small increase of calculate consumption. A multi-joint manipulator space robot is used as a representative of complex objects. An improved particle swarm optimization is used to identify the parameters of its crawl target inertial parameters without linear momentum measurement information. 115 group identification simulation example statistical results show that using the composite evaluation strategy, the recognition accuracy is increased dramatically by little increased of the calculate consumption. Qualitative analysis considers that swarm intelligence algorithms have better performance with the strategy, which is expected to be applied in a broader objects and fields.

Key words: calculating consumption, complex objects, improved PSO, local minima, composite evaluation