人工智能在洪水预报中的应用

洪水预报对防洪减灾，拦洪储存水资源等具有重要意义。对用遗传算法（GA）来改进神经网络的算法进行了分析，建立了用GA动态寻求权重的前馈网络模型（Ⅰ）和用GA优化初始权重的前馈网络模型（简称模型Ⅱ）。借助于Matlab中的神经网络包和遗传算法包，编制了改进模型的计算机程序，对湖北省黄龙滩水库14场洪峰和洪水总量进行了模拟，用另外5场洪水检验了改进的模型，并与传统的前馈神经网络模型进行了比较。结果表明：用模型Ⅰ对洪峰流量的检验效果明显优于模型Ⅱ和传统的前馈神经网络模型；模型Ⅱ对中低流量及洪水总量的检验效果相对较好。因此，本次改进的模型可以用于洪水预报：     

基于遗传神经网络的城市用水量预测研究

介绍了BP(误差反向传播算法)和GA(遗传算法)及GA-BP 3种神经网络,并以此分别对城市用水量进行预测.实验结果表明,基于GA-BP算法的神经网络方法应用于城市用水量的预测问题,能采用遗传学习算法优化BP神经网络模型的初始权重,即先利用遗传学习算法进行全局训练,再用BP算法进行精确训练,使网络收敛速度加快和避免局部极小.GA-BP神经网络在收敛速度和预测精度等方面均优于BP和GA网络,从而为未来短期城市用水量负荷的准确预测提供了新的思路与方法.     

改进Elman神经网络在径流预测中的应用

针对传统静态前馈神经网络动态性能较差的缺点,提出一种基于遗传算法（GA）优化Elman神经网络连接权值的GA-Elman多元变量年径流预测模型.以新疆伊犁河雅马渡站径流预测为例进行实例分析,并构建传统Elman,传统BP和GA-BP多元变量年径流预测模型作为对比模型,预测结果与文献IEA-BP网络模型预测结果进行对比.结果表明:①GA-Elman模型的拟合及预测效果略优于文献IEA-BP模型,该模型用于多元变量年径流预测是合理可行的,具有较好的预测精度和泛化能力.②在相同网络结构及传递函数等条件下,GA-Elman模型的预测精度和泛化能力优于GA-BP模型,传统Elman模型优于传统BP模型,表明具有适应时变特性的Elman反馈动态递归网络预测性能优于BP网络;GA能有效优化Elman神经网络连接权值,使网络的预测精度和泛化能力有了较大提高.     

基于遗传神经网络的河流冰凌预报

在分析影响凌汛成因因素的基础上,选取合适的预报因子,建立了基于遗传算法(GA)和Levenberg-Marquardt BP网络(LMBP)相结合的冰凌预报神经网络模型.GA-LMBP算法先通过遗传学习算法进行全局训练,再用LMBP算法进行精确训练,克服了传统BP算法收敛速度慢、易陷入局部极小等缺陷.把该模型应用到松花江依兰、佳木斯段开河预报中,取得了良好的效果.     

基于改进PSO-RF算法的大坝变形预测模型

针对传统随机森林参数寻优方法的不足,引入均衡惯性权重和自适应变异对粒子群优化算法进行改进,提出了一种基于改进粒子群优化算法和随机森林算法(改进PSO-RF算法)的大坝变形预测模型。实例验证结果表明,在计算效率方面,与传统网格搜索法相比,改进PSO-RF算法显著提升了模型的寻优速度;在预测精度和稳定性方面,基于改进PSO-RF算法的大坝变形预测模型明显优于长短期记忆网络、支持向量机和BP神经网络模型。     

改进模糊神经网络模型及其在大坝监测中的应用

鉴于传统模糊神经网络计算繁琐、模型精度较低、难以收敛等问题,在大坝位移的监测数据分析中应用改进的模糊神经网络及改进的反向传播(BP)网络算法,并建立了相应的网络模型.大量的数据分析计算表明,该模型收敛快,且精度优于常规的模糊神经网络和传统的统计模型.     

基于神经网络与改进D-S证据理论的水质评价模型研究

针对水质监测数据多源、非线性、不确定性大的特点，提出了一种基于神经网络与改进D-S证据理论相结合的水质评价模型。该模型首先利用3种前馈神经网络对水质监测数据进行初步评价，将初步评价结果归一化后作为基本概率分配，然后引入证据权重修正冲突证据，根据D-S合成规则得到融合评价结果，最后利用迭代思想修正评价结果。其中，基于证据权重修正冲突证据并进行融合结果的迭代修正，能有效解决传统D-S证据理论无法处理高冲突证据的缺陷。冀南地区5个监测断面水质评价结果表明，该水质评价模型能够提高水质评价准确性。     

Matlab仿真平台下大坝位移BP神经网络模型研究

基于人工神经网络的非线性映射能力,应用Matlab7.1网络仿真平台,结合辽宁省白石水库多年大坝位移实测数据,建立了3种不同改进BP算法的多层前馈神经网络模型。并通过LM算法、BR算法、GDX算法的BP网络模型的拟合、预报结果,对3种模型的应用效果进行了比较分析,得出了LM算法的BP网络更适合用于建立坝顶位移监控模型的结论,以实现对大坝位移实时、有效的监控。     

Matlab仿真平台下大坝位移BP神经网络模型研究

文中基于人工神经网络的非线性映射能力,应用Matlab7.1网络仿真平台,结合辽宁省白石水库多年大坝变形实测数据,建立了三种不同改进BP算法的多层前馈神经网络模型,分析其拟合、预报结果,选择更为合适的网络模型,实现对大坝位移实时、有效的监控。     

Matlab仿真大坝位移BP神经网络模型研究

文章基于人工神经网络的非线性映射能力,应用Matlab7.1网络仿真平台,结合辽宁省白石水库多年大坝变形实测数据,建立了三种不同改进BP算法的多层前馈神经网络模型。分析其拟合、预报结果,选择更为合适的网络模型,实现对大坝位移实时、有效的监控。     

生物预处理技术在嘉兴地区给水厂的应用比较

对嘉兴地区主要给水厂近年来应用立体弹性填料、LT型悬浮球填料、YS型悬浮填料、卵石填料等的生物预处理工艺从填料及配套设施、工艺运行参数、运行效果、成本等方面进行了比较.认为弹性填料生物预处理更适用于低浊(50 NTU以下)、低氨氮(1.5 mg/L以下)原水,LT型悬浮球填料生物预处理适用于低中浊(100 NTU以下)、低中氨氮(2.5 mg/L以下)原水,YS型悬浮填料生物预处理及"SAF"滤池对低浊、中高氨氮(2.5 mg/L以上)原水有较好的适应性.最后探讨了弹性填料存在的短流、积泥,悬浮填料存在的部分填料流化不均匀,富氧的生物工艺单元内水生动物泛滥等问题.     

中国部分河口及其近海水域缺氧现象研究

对长江、珠江等河口及其近海水域缺氧现象与影响因素进行研究,结果表明,主要分布于长江与珠江河口及其邻近海域的缺氧现象,夏季(6—8月)达到最盛,秋冬季消失;长江口的缺氧区域主要位于30.75°N~32°N,122.5°E~123.25°E附近海域,具有南北2个缺氧中心;珠江口缺氧程度相对较轻,缺氧区域主要在广州黄埔区河段至虎门水域和伶仃洋;辽河、钱塘江、海河等河口区域也时而出现零星的缺氧现象;水体缺氧受控于多项环境要素,物理层化作用是缺氧产生的最初诱因之一,陆源污染物的分解则是重要的耗氧负荷;各河口区特征各异,潮汐、洋流、营养盐、叶绿素a等要素对缺氧区的范围、程度和持续时间影响程度不一。     

无资料流域水文预报(PUB计划)研究进展

无资料流域水文预报，简称为PUB(Predictions in Ungaugecl Basins)，从2003年7月开始成为国际水文科学学会新的国际水文十年计划主题．文中首先简述了PUB计划的主要研究内容，然后回顾了国际和国内已有的PUB研究进展，最后展望了对水文学发展有重要意义的几个PUB前沿研究，包括不确定性研究、对水循环各个环节因素的预报和对无资料流域水文过程的预报。     

Changes in urban stream channels in Zimbabwe

Although channel change has been documented downstream of urban areas there have been few previous studies which have focused upon the spatial variation of erosion and upon examples of such channel change in tropical areas. In the Avondale stream basin, which includes part of Harare the capital of Zimbabwe the way in which the channel network has evolved from the extensive dambo system in the period 1891 to 1984 is reconstructed showing that drainage density which was initially between 0.35 and 0.80 km km^?2 has increased to 3.15 km km^?2 including stormwater drains. Characteristics of erosion of the modified urban channels are described and the downstream reaches indicate an average channel widening of 1.7 times involving average rates of bank erosion of 0.33 m per year. Peak discharge is estimated from channel characteristics to be increased to two times former values and it is concluded that channel type and the relationship of channel change to the accommodation of peak discharges requires further research.     

北京某学校雨水利用设计

以北京市某中学为例,介绍一种学校雨水利用的方法。根据校区水量平衡计算结果,对利用校园内汇集的雨水作为冲厕、绿化和景观用水等进行设计;采取相应的技术措施,对雨水径流的污染进行控制,保障所利用雨水的水质达到要求,并采用生态净化的方式,使学校景观水体循环起来。虽然雨水利用工程投资较高,但其社会和环境效益显著。     

水利高性能计算的进展

针对水利信息化对精细模拟与高性能运算提出的愈来愈高的要求,在简单介绍高性能计算的硬件、软件环境及其在水利上的几个应用实例后,着重介绍了并行有限元法、并行边界元法的一些进展以及网格技术的最新进展。指出从网络到网格是高性能计算发展的最新方向,应以网格平台支持、促进水利信息化进程。     

中国内地水库淤积的差异性分析

为了解水库的淤积现状,根据115座具有代表性的中国内地水库淤积资料,对这些水库按类型和区域进行统计分析,结果表明,中小型水库比大型水库淤积严重,不同流域水库淤积的空间差异明显,黄河中下游地区水库淤积比例最大,西南地区水库年均淤积率最大。截止到2003年,根据代表性水库淤积的计算结果推算出中国内地水库的平均淤积比例约为20%,库容年均淤积率为0.76%,相当于每年损失1座库容近42.3亿m3的超大型水库。     

Trends in ecological river engineering in Korea

Ecological river engineering can be defined as the design and implementation of river works and river restoration works for the benefit of human society. It also guarantees the sustainable ecological functions of a river, such as its habitats and self-purification of its water. It is currently in the beginning stages in Korea, utilizing scientific knowledge on the processes of aquatic ecosystem degeneration and a methodology for solving the ecological problems in artificially altered rivers currently under development. The changes in river management and work practices in Korea may be best explained with a chronologically progressing sequence of ‘Natural’, ‘Disaster-prevention’, ‘Occupied’, ‘Park’, and ‘Close-to-nature’ rivers. Since the 1960s, the focus on river management and work has shifted from flood control only, to both flood control and riverine habitat conservation and restoration. Five research topics have been selected for this article, and the progress of each research area is briefly described with a representative picture in each topic. They are as follows: (1) flow resistance due to vegetation, (2) environmental flow, (3) floodplain vegetation modeling, (4) small dam removal, and (5) river restoration. For the future prospects of research on ecological river engineering in Korea, a necessity of further research on floodplain vegetation recruitment and succession, which can explain the so-called ‘white river’ and ‘green river’, is underlined, among others. Finally, two ongoing large research programs on river ecosystem restoration, of which are sponsored by the Government of Korea, are briefly introduced, followed by the introduction of a near-prototype experiment facility recently completed mainly for research on ecological river engineering.     

吉林省2001年汛期降水量趋势预测

本文根据历年6－9月降水量距平值累计曲线和历年6－9月降水量特征值变化，对我省今后汛期全省各地区降水趋势及降水量进行了预测，可供有关部门参考。     

Lake—Wide Seasonal Changes in Limnological Conditions in Lake Michigan in 1976

Data collected on lake-wide cruises in 1976 were used to study seasonal and vertical variations in water temperature, transparency, chlorophyll a, and nutrients in Lake Michigan. Data were analyzed according to subsets corresponding to the northern and southern open lake. Comparisons (t-tests) of data from the open lake indicated that the average water temperature was cooler and average water transparency was greater in the northern lake than in the southern, but with the exception of total phosphorus, average nutrient concentrations did not differ between the northern and southern parts. It was found that physical-chemical characteristics of nearshore and Straits of Mackinac stations differed significantly from open lake stations. Seasonal phytoplankton dynamics in the open lake were related to seasonal and vertical changes in silica and nitrate nitrogen. The spring phytoplankton bloom occurred before the lake was strongly stratified thermally. After thermal stratification was well developed, epilimnetic concentrations of chlorophyll a decreased, probably due to some combination of nutrient limitation and zooplankton grazing, and maximum chlorophyll a concentrations were found below the thermocline. Epilimnetic silica concentrations decreased after thermal stratification and diatoms were replaced in the phytoplankton assemblage by green and blue-green algae in late summer. Total phosphorus averaged only 8 μg/L on a lake-wide basis and thus only small but significant reductions in absolute concentration can be expected from phosphorus control programs. However, over a period of several years, these small reductions in concentration may be difficult to verify from total phosphorus measurements which have relatively large sampling errors and variances.