摘要
多目标演化算法(MOEA)不能很好地解决强约束问题的多目标求解。为此,采用一种新的集复合形方法和几何选择为一体的多目标Pareto进化算法(GPEC),对常规型抽油机的上冲程最大扭矩因数和上冲程悬点加速度进行多目标优化设计。该算法中,对抽油机杆件尺寸采用浮点数编码方案,利用复合形法在可行域内构造演化算法的初始种群,在杂交和变异过程中,始终保持种群中具有足够多的有效个体,并且在每一代进化过程中,挑选出与无穷远处的某点距离最远的点输出,最后再筛选出有效的Pareto前沿点。从优化计算结果看,该算法能有效地找出足够多的Pareto前沿点。
Multi-objective problem (MOP) of strong constrained condition can not be properly solved by using multi-objective evolutionary algorithm (MOEA). Therefore a new algorithm of geometrical Pareto evolution algorithm with complex (GPEC) based on geometrical characters is used to make a multi-objective optimal design for maximum torque factor and polished rod acceleration on the upstroke of a conventional pumping unit. In the calculation the dimension of rod is coded by using the method of floating point encoding, and the initial population for the evolutional algorithm is established in the feasible domain by using composition method. During the cross and variation, enough valid individuals are always maintained in the population, and the output of the farthest point from the earliest given point is selected and eventually the valid Pareto frontier point is selected. The result of optimized calculation shows that the method can be used for finding enough Pareto frontier points.
出处
《石油机械》
北大核心
2006年第8期21-24,共4页
China Petroleum Machinery
关键词
常规型抽油机
优化设计
演化算法
多目标优化
conventional pumping unit, optimized design, evolutional algorithm, multi-objective optimization
作者简介
李克清，副教授，博士生，生于1966年，主要研究方向为网格计算、计算机网络和算法设计。地址：（430072）湖北省武汉市。E—mail：likq03@126．com。
