生命线工程网络系统抗震可靠性分析方法综述

生命线工程系统地震反应分析的一个重要任务是计算系统的抗震可靠度,因此,寻找一种高效合适的系统可靠性分析方法至关重要.对生命线工程网络系统抗震可靠性分析中目前常用的几种算法进行了系统的比较、分析和综述.     

基于遗传算法的大气颗粒物的源解析

遗传算法是一种全局优化的随机搜索方法,适用于模型参数寻优。将遗传算法用于大气颗粒物源解析的CMB方程组中参数的优化,得到各污染源对大气颗粒物的优化贡献率。该方法应用于成都市大气颗粒物的源解析结果与用其他多种源解析法得出的结果基本一致。理论分析和应用实践表明了遗传算法用于大气颗粒物源解析研究具有简便、准确和实用性强等特点。      

改进型量子遗传算法应用于区域水资源可持续利用评价

提出了一种对多项指标都能普遍适用的区域水资源可持续利用评价模型,将实数编码混沌量子遗传算法应用于区域水资源可持续利用评价模型中参数的优化,并用优化好的模型对区域水资源可持续利用做出评价。通过对中国12个区域实测数据的评价实例研究及与其它评价方法得到的结果进行对比,表明该模型具有实用性和可行性。     

MODELING AND SENSITIVITY ANALYSIS FOR PLANNING DECISIONS IN WATER RESOURCES EXPANSION1

ABSTRACT. Individuals and organizations concerned with the expansion of the facilities of a river basin (such as a river basin authority) need to determine optimal strategies of operation and capital investment. They also need to examine the sensitivity of whatever planning decisions are contemplated. This paper extends the applicability of an algorithm that had been previously applied to the deterministic river-basin expansion problem to include the feature of a sensitivity analysis. The algorithm, containing a partial enumeration search technique and a network analysis code, gave a construction sequence of reservoirs, canals, and treatment plants, and an operating policy that maximized the present value of net earnings consistent with certain underlying assumptions. A river basin was chosen that had an existing configuration of unregulated streams and rivers, reservoirs, canals and treatment plants, and sites for future additional facilities. A series of representative synthetic flow sequences, future demand profiles, interest rates and reservoir costs that served as inputs to or parameters in the system were each perturbed by various factors (for a total of 24 cases). The sensitivity studies showed that the immediate planning decision of what facility to construct next was insensitive to variations in future demands and costs and independent of later decisions. Thus, decision-making was adaptive in the sense that by always making the optimal proximate decision, the management of the river basin is optimized.     

OPTIMAL CAPITAL INVESTMENT IN THE EXPANSION OF AN EXISTING WATER RESOURCES SYSTEMS1

The study of the optimal expansion of existing water resources systems is of continuing importance because of the rising demand and limited supply of water in many areas of the world, particularly in the southwestern part of the United States of America. This study is concerned with the investigation of the optimal expansion of a realistic water resources system to meet an increasing demand for municipal and industrial use, irrigation, energy, and recreation over a planning horizon of T years. A number of possible dam sites are available for the further regulation of river (canal) flows in the basin and/or the regulation of imported waters into the basin. To maximize, over the set of alternative projects, the sum of discounted present value of net earnings subject to the demands and various institutional, physical and budgetary limits, an optimization problem (Problem I) was formed as a 0-1 mixed integer programming problem and was decomposed into the set of all feasible combinations (Problem II). The economic return was determined for each combination (Problem III). Problem II was solved by a branch and bound procedure which selected each feasible combination of dams while the optimal return for each such combination (Problem III) was found by a network analysis code.     

门限自回归模型在海洋冰情预测中的应用

给出了建立门限自回归模型（ＴＡＲ） 的一套实用方法, 用作者提出的改进遗传算法可同时优化门限值和自回归系数,成功地解决了ＴＡＲ建模过程所涉及的大量复杂寻优工作这一难题。实例计算的结果说明了, 这套方法在海洋冰情预测中是可行的和有效的, 在各种自然灾害非线性时序动态预测中具有重要的理论意义和广泛的实用价值。     

基因算法在环境优化问题中的应用

给出了处理环境优化问题的一种改进基因算法，并通过实例对该法与传统优化方法进行了比较，结果表明，该法直观，简便，适应性强，是一种优秀的非线性优化方法。     

煤与瓦斯突出危险性计算机识别与预报系统

根据反向传播神经网络模型的原理，设计了煤与瓦斯突出危险性的计算机识别与预报系统，借助于对已知突出实例的训练学习，得出反映突出预测指标和突出危险程度之间复杂的非线性映射关系的网络系数（预报规则），可以对突出危险性进行多因素识别预报。通过外推预报检验，表明系统具有较高的可靠性和实用性     

基于B—P网络的大气环境质量评价

选取大气环境质量标准作为运算样本，以大气污染物各级标准值作为样本输入信息，建立了大气环境质量的Ｂ－Ｐ网络评价模型，该模型用于武汉市大气环境质量评价，并与用模糊数学评价结果比较，表明Ｂ－Ｐ人工神经网络用于大气环境质量评价具有通用性、合理性和实用性。     

基于序列模式的土地利用变化分析——以广西壮族自治区为例

土地利用变化是一个长期的过程,同时具有一定的复杂性。传统的转移矩阵方法只能在两期土地利用数据之间相互比较而不能总结长期整体的变化规律,频繁项集不能导出变化序列。针对这些方法的不足,本文提出了基于序列模式的土地利用变化序列分析模型。首先给出土地序列数据库的定义,然后根据土地序列数据实际特点和垂直格式的序列模式SPADE算法,给出土地变化序列模式的计算过程和方法。在以中国广西为实例的研究中,计算了1970s—2015年共7期22种二级土地利用类型的变化序列。研究区土地利用变化主要发生在林地之间,部分林地转换为厂矿、采石场、交通道路等建设用地类型;城镇用地主要由旱地和水田转换而来;在研究期内没有任何一个土地单元转化成水田。