兰成渝腐蚀管道失效压力的GA-BP 神经网络组合预测方法

为准确掌握管道失效压力,保证管道安全运行,根据神经网络的非线性和良好的函数逼近特性,提出了基于遗传算法（GA）优化的BP神 经网络组合模型的腐蚀长输管道失效压力预测模型。组合模型将最佳组合阀值与权值隐含在网络的连接中,兼具遗传算法、人工神经网络预测 的优点,并克服了原始数据少对预测精度的影响,同时避免了神经网络容易陷入局部寻优的缺陷,也增强了网络的适应性,改善网络的收敛性 ,在客观地反应腐蚀油气管道失效压力变化趋势方面具有一定的优势。通过实例分析,结果表明:BP神经网络的预测值和Modified B31G计算结 果与真实值误差均较大,而GA-BP的预测值与实际结果的相对误差最大为6.12%,有很好的一致性,为管道的预防性维修提供了理论依据。     

广义多态事件树模型

针对工程系统中普遍存在的反馈关系 ,提出了广义多态事件树的概念 ,利用多值逻辑及集合理论 ,建立了广义多态事件树的数学模型。该模型拓展了一般的事件树模型 ,解决了存在反馈关系的系统的建树难题 ,并且可作为求解大规模复杂多态系统广义最小割集合的一种算法     

非均匀电场对土壤中基因工程菌的迁移与机理

以克隆有绿色荧光蛋白基因的Rm1021-GFP细菌作为标记细菌,研究了该细菌在非均匀电场作用下在土壤中的运移特征和存活性,分析了其运移机理和影响因素.结果表明,非均匀电场能有效地促进土壤中Rm1021-GFP细菌的运移,电泳、介电泳和随电渗析流迁移是其主要机理,土壤介质、电压梯度和电极反应是其主要影响因素,电极反应对电极附近细菌的存活性有不利影响.     

Screening of flocculant-producing strains by NTG mutagenesis

Screening of new microorganism being able to produce efficiently flocculants was carried out. A new model for screening efficient flocculant-producing strains was designed and tested. The results showed that this model for screening efficient flocculant-producing strains is very reliable and can greatly shorten the screening period. 13 flocculant-producing strains were isolated from activated sludge by conventional method. A strain, designated as HHE6, produced the bioflocculant with the turbidity removal 98% for kaolin suspension. Six of 13 strains selected as the original strains were treated with NTG as mutagen, and five mutant strains(HHE-P7, HHE-A8, HHE-P21,HHE-P24, HHE-A26) with high flocculation efficiency was obtained by selection, which exhibited the flocculation rate for kaolin suspension above 90%. Strains HHE6, HHE-P7, and HHE-P24 were classified as Penicillium purpurogenum, HHE-P21 as Penicillium cyclopium,HHE-A26 as Aspergillus versicolor and HHE-A8 as Aspergillus fumigatus, and it is hitherto unreported for biofloccutant-producing strains of Penicillium. The growth of the six strains(HHE6, HHE-P7, HHE-A8, HHE-P21, HHE-P24, HHE-A26) had similar curves, i.e. firstly increasing rapidly, keeping relatively constant then and finally decreasing gradually with cultivation time. The production of bioflocculants by strains showed the similar pattern to strain growth.     

大气质量污染损失率评价模型参数的GA优化

设定各项污染物在大气环境质量污染损失率评价模型中有相同参数值,采用具有全局优化的遗传算法(GA)对模型中的参数进行优化,得出了对各种大气污染物均适用的大气环境质量污染损失率计算公式及大气环境质量综合评价模型。该模型与其他多种评价模型用于大气环境实例进行评价比较,结果表明:该评价方法不仅原理直观,评价结果准确,而且计算简便,实用性强。      

INCORPORATING UNCERTAINTY IN THE DESIGN OF STREAM CHMNEL MODIFICATIONS1

ABSTRACT: The designs of stream channel naturalization, rehabilitation, and restoration projects are inherently fraught with uncertainty. Although a systematic approach to design can be described, the likelihood of success or failure of the design is unknown due to uncertainties within the design and implementation process. In this paper, a method for incorporating uncertainty in decision‐making during the design phase is presented that uses a decision analysis method known as Failure Modes and Effects Analysis (FMEA). The approach is applied to a channel rehabilitation project in north‐central Pennsylvania. FMEA considers risk in terms of the likelihood of a component failure, the consequences of failure, and the level of difficulty required to detect failure. Ratings developed as part of the FMEA can provide justification for decision making in determining design components that require particular attention to prevent failure of the project and the appropriate compensating actions to be taken.     

矿井通风系统风流控制的改进算法

介绍了矿井通风系统风流控制的基本数学模型和计算方法,指出了常规计算方法的不足,提出了一种能够避免大量矩阵运算、适用于实际通风网络的新算法,程序简单,计算时间省,并结合实例阐述了算法原理。     

反演同化技术在激光雷达中的应用研究

为增强模式预报准确性,提高激光雷达数据在化学模式中的应用水平,建立激光雷达同化系统,将激光雷达数据引入模式,利用WRF-Chem模拟2016年12月6日～10日一次污染过程,将激光雷达监测的气溶胶光学深度(AOD)反演成近地面PM2.5浓度,通过三维变分同化技术订正模式原始场,调整模式预报结果.实验对比发现,反演同化后...     

指数加权移动平均控制图在实验室中的应用及与其他质控图比较

阐述了指数加权移动平均(EWMA)控制图在环境分析实验室中的应用,并与休哈特控制图和Westgard多规则对于不同偏移量的ARL值作了比较.结果表明,EWMA控制图在检验不准确度,尤其在微小偏移上的优势值得推广.     

水质模糊贴近度模型中权值的遗传算法解

按照模糊综合评价中贴近度最小的原则,采用加速遗传算法优化权值。经过实例分析及对模型的验证、比较表明,用该法优化权值并进行水质评价,其结果与实际水质级别具有较高的一致性,评价方法更趋合理。