摘要
针对最优权组合模型可能出现不符合实际的负权重,增加权重大于等于0的约束条件,放松权重之和等于1的约束条件,采用遗传模拟退火算法搜索最优权重建立组合模型。实例说明:该方法合理可行,所建立的组合模型与最优权组合模型的拟合精度和预报精度相近,克服了最优权和基于神经网络组合模型的缺点。
In view of the untruthful negative weighting that may occur in optimum weighting composite model, constraint condition of nonnegative weighting is added, and constraint condition that sum of the weightings equals to 1 is relaxed. Then genetic simulated annealing algorithm is used to search optimum weightings to build composite model. By the example, the method is feasible. Fitting and forecasting precision of the composite model built by use of the method are close to that of the optimum weighting composite model. The method settles the problem existing in optimum weighting and composite model based on neural network.
出处
《水电能源科学》
2003年第4期4-6,22,共4页
Water Resources and Power
基金
国家自然科学基金重点项目(50139030)
教育部博士点基金资助项目(20020294005)
河海大学科技创新基金(2002404443)。
关键词
大坝安全监控
遗传算法
模拟退火算法
组合模型
dam safety
genetic algorithm
simulated annealing algorithm
composite model
