本文在工程模糊集理论的基础上,融合香努(Kohonen) 聚类网络与自适应谐振理论(Adaptive Resonance Theory,ART) 的优点给出了一种新的模糊聚类神经网络,并将其应用于区域水资源评价中。网络模型融入模糊识别理论,增强了处理复杂水资源问题的能力,同时模糊竞争的学习算法改善了Kohonen网络学习效果。结合ART理论使得网络模糊类别数可以增加,因此具有较强的可塑性。最后,通过水资源丰富度评价与水资源承载能力评价实例验证了本文提出的模型与方法的可行性与有效性。

建立了以模糊优选、BP神经网络及遗传算法有机结合的智能预报模式与方法。在应用该方法进行中长期水文智能预报时,首先选取训练样本的数量,根据预报因子与预报对象的相关关系得到相对隶属度矩阵;再将其作为BP神经网络输入值以训练连接权重;最后将得到的连接权重值用于预报检验。计算结果表明,智能预报模式与方法的运行速度、精度及稳定性都达到了实际应用的要求。

Lectotype optimization of offshore platforms is of particular importance in the concept design process. Lectotype optimization involves multiple objectives with uncertainty and so is a problem of multiple attribute decision making. To date, there have been few published works on this topic in the context of offshore engineering. This paper develops a framework and methodology for evaluation of offshore platform alternatives, where a fuzzy optimum model is proposed to integrate the influence of each objective in the criteria set, and a new weight-assessing method is developed to mimic the decision maker’s experience and preference based on complementary pairwise comparisons. A case study shows that the new framework and methodology is scientific, reasonable and easy to use in practice.

Due to significance of the engineering system for flood control and the multi-dimensional synthesis of the risk evaluation issue that the paper selects typical risk indexes to classify risk degree of flood-control engineering. Under global view of system that the variable fuzzy sets method is presented to set up comprehensive evaluation model for flood-control engineering system (FCES). The method can scientifically and reasonably determine membership degrees and relative membership functions of disquisitive indexes at level interval that relating to engineering, also it can fully use one’s experience and knowledge, qualitative and quantitative information of index system to obtain weights of indexes for operating comprehensive risk evaluation for FCES. The numerical example shows that the proposed method is feasible and effective, and the evaluation results are reasonable.

This paper aims to develop a fuzzy dynamic programming approach for multiobjective multistage decision making problems. Fuzzy dynamic programming usually convert the problems into corresponding single objective problems by an aggregate of hybrid objective values that expresses the performance of the particular stage decision. Weightassessment is an important problem arising from the evaluation of multiple objectives. Therefore, a fuzzy iteration model is first developed to simultaneously provide the objective weights and the evaluation of alternatives with multiple objectives. Then a fuzzy dynamic programming is derived by applying the fuzzy iteration model to classic dynamic programming to evaluate the decisions at each stage in the dynamic process of decision makings, where the objective weights are initially elicited from the information implicit in alternatives and can be interactively adjusted to reflect the changes of decision situations by using a non-structured decision making analysis method if necessary. Finally, the fuzzy dynamic programming is validated through a problem of water resource allocation.