针对耗时计算目标函数的约束优化问题,提出用代理模型来代替耗时计算目标函数的方法,并结合目标函数的信息对约束个体进行选择,从而提出基于代理模型的差分进化约束优化算法。首先,采用拉丁超立方采样方法建立初始种群,用耗时计算目标...针对耗时计算目标函数的约束优化问题,提出用代理模型来代替耗时计算目标函数的方法,并结合目标函数的信息对约束个体进行选择,从而提出基于代理模型的差分进化约束优化算法。首先,采用拉丁超立方采样方法建立初始种群,用耗时计算目标函数对初始种群进行评估,并以此为样本数据建立目标函数的神经网络代理模型。然后,用差分进化方法为种群中的每一个亲本产生后代,并对后代使用代理模型进行评估,采用可行性规则来比较后代与其亲本并更新种群,根据替换机制将种群中较劣的个体替换为备用存档中较优的个体。最后,当达到最大适应度评估次数时算法停止,给出最优解。该算法与对比算法在10个测试函数上运行的结果表明,该算法得出的结果更精确。将该算法应用于工字梁优化问题的结果表明,相较于优化前的算法,该算法的适应度评估次数减少了80%;相对于FROFI(Feasibility Rule with the incorporation of Objective Function Information)算法,该算法的适应度评估次数减少了36%。运用所提算法进行优化可以有效减少调用耗时计算目标函数的次数,提升优化效率,节约计算成本。展开更多
This research quantifies lake level variations in the Siling Co, Co'e and Bangor Co salt lakes in Central Tibet from 1976 to 2010, and most notably for the 2000-2010 periods. In particular, the effects of differen...This research quantifies lake level variations in the Siling Co, Co'e and Bangor Co salt lakes in Central Tibet from 1976 to 2010, and most notably for the 2000-2010 periods. In particular, the effects of different water replenishment modes on the lakes have been analyzed. Here we have provided new evidences for climate warming and accelerated glacial ablation on the Central Tibetan Plateau from 2000 to 2010. Based on fieldwork involving Differential Global Positioning System (DGPS) surveying and Remote Sensing (RS) interpretations of the lake area, we have drawn the following conclusions. (1) From 1976 to 2010, the process of lake level variation in Siling Co can be divided into two stages. From 1976 to 2000, the lake level rose 4.3 m in a steady fashion (from 4530 to 4534.3 m); the rise rate was 0.18 m/a. From 2000 to 2010, the lake level rapidly rose 8.2 m (from 4534.3 to 4542.5 m), with a dramatically higher rise rate of 0.82 m/a. Compared with the rapidly increasing lake level of Siling Co from 2000 to 2010, the fluctuations observed at Co'e and Bangor Co were smooth and inconspicuous. (2) From 1976 to 2009, the lake area of Siling Co experienced a steady-rapid-steady expansion pattern. The lake area of Siling Co increased 656.64 km2 in the 34 years to 2010, a proportional growth of 39.4%. This was particularly significant in the 2000-2010 period, when the lake area of Siling Co increased by 549.77 km2, a proportional growth of 30.6%. (3) According to correlation analysis, the rise in regional temperatures, which has led to the ablation of glaciers, is the main reason for the rapid rise in Siling Co lake levels in the 10 years to 2010. During this period, Siling Co rose approximately 8 m as the direct result of glacial melting. An increase in precipitation in the Siling Co catchment area is the secondary factor. This contrasts with Bangor Co, where the dominant factor in lake level change is the long-term increase in precipitation; here, the increasing temperature is the secondary factor.展开更多
文摘针对耗时计算目标函数的约束优化问题,提出用代理模型来代替耗时计算目标函数的方法,并结合目标函数的信息对约束个体进行选择,从而提出基于代理模型的差分进化约束优化算法。首先,采用拉丁超立方采样方法建立初始种群,用耗时计算目标函数对初始种群进行评估,并以此为样本数据建立目标函数的神经网络代理模型。然后,用差分进化方法为种群中的每一个亲本产生后代,并对后代使用代理模型进行评估,采用可行性规则来比较后代与其亲本并更新种群,根据替换机制将种群中较劣的个体替换为备用存档中较优的个体。最后,当达到最大适应度评估次数时算法停止,给出最优解。该算法与对比算法在10个测试函数上运行的结果表明,该算法得出的结果更精确。将该算法应用于工字梁优化问题的结果表明,相较于优化前的算法,该算法的适应度评估次数减少了80%;相对于FROFI(Feasibility Rule with the incorporation of Objective Function Information)算法,该算法的适应度评估次数减少了36%。运用所提算法进行优化可以有效减少调用耗时计算目标函数的次数,提升优化效率,节约计算成本。
基金supported by the Knowledge Innovating Program of Chinese Academy of Sciences (KZCX2-YW-12)the National Basic Research Program of China (2011CB403106)the National Natural Science Foundation of China (40940018)
文摘This research quantifies lake level variations in the Siling Co, Co'e and Bangor Co salt lakes in Central Tibet from 1976 to 2010, and most notably for the 2000-2010 periods. In particular, the effects of different water replenishment modes on the lakes have been analyzed. Here we have provided new evidences for climate warming and accelerated glacial ablation on the Central Tibetan Plateau from 2000 to 2010. Based on fieldwork involving Differential Global Positioning System (DGPS) surveying and Remote Sensing (RS) interpretations of the lake area, we have drawn the following conclusions. (1) From 1976 to 2010, the process of lake level variation in Siling Co can be divided into two stages. From 1976 to 2000, the lake level rose 4.3 m in a steady fashion (from 4530 to 4534.3 m); the rise rate was 0.18 m/a. From 2000 to 2010, the lake level rapidly rose 8.2 m (from 4534.3 to 4542.5 m), with a dramatically higher rise rate of 0.82 m/a. Compared with the rapidly increasing lake level of Siling Co from 2000 to 2010, the fluctuations observed at Co'e and Bangor Co were smooth and inconspicuous. (2) From 1976 to 2009, the lake area of Siling Co experienced a steady-rapid-steady expansion pattern. The lake area of Siling Co increased 656.64 km2 in the 34 years to 2010, a proportional growth of 39.4%. This was particularly significant in the 2000-2010 period, when the lake area of Siling Co increased by 549.77 km2, a proportional growth of 30.6%. (3) According to correlation analysis, the rise in regional temperatures, which has led to the ablation of glaciers, is the main reason for the rapid rise in Siling Co lake levels in the 10 years to 2010. During this period, Siling Co rose approximately 8 m as the direct result of glacial melting. An increase in precipitation in the Siling Co catchment area is the secondary factor. This contrasts with Bangor Co, where the dominant factor in lake level change is the long-term increase in precipitation; here, the increasing temperature is the secondary factor.
