Collision of convex objects for calculation of porous mesh in gas explosion simulation

We investigate the coupling of the flamelet combustion model with the collision distance algorithm based on Minkowski addition. The collision algorithm is coded to calculate the porosity of the geometry based on the PDR (Porosity Distributed Resistance) approach for modelling of complex geometries. The turbulent field generated by the interaction of the flow with the porous objects is used to calculate the wrinkling length scale of the flame via the fluctuating velocities. The turbulent fluxes are amended in accordance with assigned porosities at the cell faces. The combustion and porosity models are implemented in the framework of an in house Fortran code that solves the full set of Navier-Stokes equations. The code was named STOKES - Shock Towards Kinetic Explosion Simulator). Results are presented for non-reacting flows and reacting flows over different geometries. Numerical findings are compared with standard commercial CFD tools.     

Real-time leak detection in oil pipelines using an Inverse Transient Analysis model

Based on Inverse Transient Analysis (ITA) method, a real-time leak detection method is proposed to capture leak location and the associated leak rate in oil pipe conveyance systems. In the proposed approach, location and flow rate of leak (if any), the fluid properties, as well as physical parameters of the system, are calculated in consecutive periods through minimizing the discrepancy between the calculated and measured flow parameters of the system. The method of characteristics is employed to numerically calculate the transient responses of the system and the genetic algorithm is utilized as the optimization engine. The proposed approach was applied to several real pipeline systems in which the required transient flow data are either directly collected from the field or fabricated with a third-party numerical software. Extensive numerical explorations were conducted to investigate the performance of the proposed method in real-time leak detection and to determine the extent to which field data errors, stemming from Supervisory Control and Data Acquisition (SCADA) systems and measurement equipment, affect the leak flow rate and location detectability of the proposed approach. The results show that the proposed approach provides promising results under a variety of transient and steady-state flow conditions even in the case with small leak flow rate of around 2% of the line rate. The results also reveal that the noises in the measurement data and the errors originated from SCADA systems do not significantly compromise the leak detectability of the proposed approach, confirming that the proposed approach can be utilized in practice.     

基于遗传神经网络模型的大气环境质量评价方法

设计了用遗传算法训练神经网络权重的新方法,实验结果显示了遗传算法快速学习网络权重和全局搜索的能力,有效地解决了BP算法的局部收敛问题。误差反向传播的遗传——神经网络(GA—BP)模型用于大气环境质量综合评价,具有简便、准确、客观和适应性强等优点。     

人工神经网络在生态环境质量评价中的应用

以生态环境指标的各级评价标准作为训练样本,应用人工神经网络建立生态环境质量评价的BP网络模型。用训练好的BP网络对生态环境质量进行评价,并与模糊识别评价法进行比较。结果表明,人工神经网络用于生态环境质量的评价结果与环境质量实况相符。     

Cleaner production for continuous digester processes based on hybrid Pareto genetic algorithm

ＩｎｔｒｏｄｕｃｔｉｏｎＷｉｔｈｏｂｓｏｌｅｔｅｔｅｃｈｎｏｌｏｇｙ ,ｅｑｕｉｐｍｅｎｔ,ａｎｄｍａｎａｇｅｍｅｎｔ,Ｃｈｉｎａ′ｓｐｕｌｐａｎｄｐａｐｅｒｉｎｄｕｓｔｒｙｉｓｒｅｃｏｇｎｉｚｅｄａｓｂｅｉｎｇｏｎｅｏｆｔｈｅｍｏｓｔｈｉｇｈｌｙｐｏｌｌｕｔｉｎｇｓｅｃｔｏｒｓ.Ｉｔｓｐｌａｎｔｓｄｉｓｃｈａｒｇｅｄｏｖｅｒ3ｍｉｌｌｉｏｎｔ ａｏｆＣＯＤｉｎ 1 995,ａｃｃｏｕｎｔｉｎｇｆｏｒ 41 .8%ｏｆｔｈｅｏｖｅｒａｌｌｉｎｄｕｓｔｒｙｐｏｌｌｕｔｉｏｎｌｏａｄｏｆＣｈｉｎａ(Ｒｅｎ ,1 998) .Ｉｎｔ…     

示踪实验确定河流纵向离散系数的单纯形加速法

应用单纯形加速法结合实例求出福建闽江干流富屯溪洋口段纵向离散系数ＤＬ＝２．６２ｍ＾３／ｓ。研究结果表明,该法用于估计非线性模型参数优于“非线性逼近法”。由于单纯形法寻优过程不需要计算目标函数的偏导数,不受模型复杂程度的限制,因而计算简便,可广泛应用于环境,生态模型的拟合和参数估计。     

基于MCOA布置矿井瓦斯传感器

为了预防矿井瓦斯事故,确定增设瓦斯传感器的数量和位置。通过建立监测有效 等级来确定增设瓦斯传感器的数量,然后通过分析多个猴群同步并行搜索、简易猴群算 法的初步搜索和基本和声算法的二次搜索等方法,设计出猴群优化算法的计算流程,再 根据瓦斯传感器布置的目标函数确定增设瓦斯传感器位置。以大隆矿为例进行了现场工 业试验,结果表明,增设瓦斯传感器的数量与监测有效等级之间成正比例关系,但并非 随着监测有效等级的提高而持续增加。     

遗传梯度法在水质数学模型参数估值的应用

介绍一种混合遗传算法－－遗传梯度法,并将其应用于水质数学模型参数估值,较一般优化算法能稳定得到的全局最优解。     

基于人工神经网络的夜光藻密度预测模型

利用人工神经网络 BP算法 ,对各种理化因子与赤潮中夜光藻密度建立了人工神经网络预报模型 ,并利用该模型对各种理化因子与夜光藻密度的非线性对应规律进行了研究。结果表明 ,模型较好地反映了存在的对应规律     

基于改进FWA-NN的污水处理过程溶解氧浓度预测

为实现对污水处理过程溶解氧质量浓度的实时准确预测,提出了一种改进的混沌烟花混合优化算法并构建了基于改进算法的神经网络在线软测量模型.结合污水处理过程的数据特征,定义了一项新的样本相似度衡量指标用于提取更具代表性的建模数据.在改进算法中,为提高基本烟花算法初始成员的质量,定义了一种改进的两级正弦混沌映射并利用混沌运动的遍历性精选烟花算法的初始群成员;通过融合混沌算法改进了基本烟花算法的搜索机制,基于设定准则将寻优过程分为两阶段并采用两分群同时进行.测试结果表明改进算法的收敛速度和收敛精度有较大程度提高.将改进的软测量建模方法和样本数据提取方法用于污水处理过程溶解氧质量浓度软测量建模,应用结果表明该模型的均方根误差和平均泛化误差分别为0.0175和0.0118,具有较强的泛化性能.