协同演化免疫算法提取水库调度规则研究

Reservoir Operating Rule Extraction based on Co-evolution Artificial Immune System

从复杂适应系统角度探讨了水库调度的复杂性,分析比较了水库调度规则提取过程与人体免疫反应中抗体种群协同演化的相似性,提出了基于协同演化人工免疫系统(CoEISM)水库调度规则提取模型,并应用于某水库供水调度规则提取,所获规则对训练与检验样本识别率分别为92.5%和84.4%,缺水指数为2.23(1014m6),与神经网络(RBF)提取结果相当。进一步研究表明,CoEISM用于水库调度规则提取是可行、有效的,不仅所获规则易于理解,能提供更为丰富的决策信息;而且调度规则自适应调整有利于调度人员做出适应调度环境变化的水库蓄泄决策。

The background of reservoir operation is becoming complex.The release strategies has to be continuously adjusted to be adaptable for the change of reservoir operation enviroments.In this paper,the complexity of reservoir operation was discussed from the viewpoint of complex adaptive system and the similarity was analyzed between the process of extracting reservoir operating rule and self-adaptive learning of antibody and antigen in human immune system.A reservoir operating rule extraction model based on the co-evolution immune system（CoEISM） is developed and applied to extract the water-supply reservoir operating rule of a water-supply reservoir in North China.The obtained operating rules acquire the identification rate of 92.5% for training samples and 84.4% for testing samples,respectively,which is comparable to that by the RBF.The reseaches indicate that the co-evolution artificial immune system is feasible and effective for extracting reservoir operating rule.The obtained rules are easier to be interpreted and can provide more rich decision-making information.The self-adaptive adjustment of rules is helpful for decision-makers to make reseasonable release strategies to fit for environmental changes.