改进蚁群算法设计拉式膜片弹簧

Improved ant colony algorithm for optimization design of pull-type clutch diaphragm spring

通过对拉式膜片弹簧载荷—变形特性的综合分析,考虑各种约束条件,提出了一种新的多目标优化设计数学模型.该模型以在摩擦片磨损极限范围内,弹簧压紧力变化的平均值最小及驾驶员作用在分离轴承装置上的分离操纵力的平均值最小为共同优化目标,使离合器后备系数稳定,离合器分离力的平均作用力较小.蚁群算法是一种新型的元启发式优化算法,该算法具有较强的发现较好解的能力,但同时也存在一些缺点,如容易出现停滞现象、收敛速度慢等.将遗传算法和蚁群算法结合起来,在蚁群算法的每一次迭代中,首先根据信息量选择解分量的初值,然后使用变异操作来确定解的值.最后,通过实例与其他优化方法的结果进行比较.结果表明,该算法有较好的收敛速度及稳定性.

This paper presents a new multi-objective optimization design mathematical model by comprehensively analyzing the load-deflection characteristics and possible constrains of pull-type clutch diaphragm spring. This model takes the minimum of average compressing force of spring within the scope of the friction slice wear and the driver's minimum manipulating force on separating bearings as optimization objectives, which can bring a more stable reservation coefficient and a smaller declutching force. As a novel meta-heuristic optimization algorithm, ant algorithm possesses powerful ability in searching better solutions coexisting with the disadvantages such as easily immersing into stagnation, slow convergence speed and so on. Ant algorithm is combined with genetic algorithm. In each iteration of ant colony algorithm, the first step is to choose initial values of components by adopting the trail information, and then determine the solution by cross and mutation operations. Finally, comparison between this solution and the ones gained from other optimization algorithms is conducted, the result of which indicates that this algorithm have a much higher stability and convergence speed.