基于BPANN的4-CBA软测量模型研究

本文提出了一个基于3层BP人工神经网络(BPANN)新的4—CBA软测量模型。在MATLAB软件平台上，利用两种改进BP算法以及由正交试验和机理模型仿真所产生的大量数据样本对不同网络结构模型进行了学习训练和验证，并对算法的训练效果进行了比较研究。仿真结果表明基于Levenberg-Marquardt学习规则的合适结构BPANN软测量模型，不仅学习快速，预测精度也远高于经验回归模型，为实现精对苯二甲酸(PTA)产品中4—CBA含量的实时、精确控制提供了可能。     

链霉素发酵过程中软测量技术的应用

链霉素发酵过程是一个典型的生化过程,众多状态参数难以量测.分析了链霉素发酵过程的建模特点,建立了基于人工神经网络的产物浓度软测量模型.利用工业生产数据进行了应用仿真,证实了该方法的有效性,同时对BP网络和RBF网络进行了比较.     

设施农业中的磷酸根浓度软测量研究

无土栽培中的作物营养环境优化控制要求磷酸根浓度能现场测量,由于没有磷酸根离子选择电极,目前难以在现场测量磷酸根浓度.本文从机理分析出发,结合人工神经网络方法,建立了磷酸根浓度的软测量模型.基于该模型的软测量实验结果表明这种软测量方法是合理有效的.     

基于RBFNN建模的动态流量软测量方法研究

本文通过对粘性流体在圆管中的层流和湍流流量方程对比研究发现,动态流量主要与管道中摩擦导致的压头损失、管道中最大的流速、流体温度变化有关,依据这一原理设计了基于径向基函数人工神经网络(RBFNN)的软测量模型.在伺服阀动态性能实验台上构建了数据采集系统,在新型动态流量测量管上安装超声波、压差、温度传感器来采集各种信息,其中流速信息采用一种新颖的超声波类时差法获取,用于标定的实际流量由无载液压缸的速度传感器获取.基于NeuroSolution软件中的RBF网络模块组成软测量RBFNN,选用部分采集数据作为学习样本对RBFNN进行训练,建立了动态流量的软测量模型.利用采集的数据的测试样本对RBFNN进行测试,通过流量预测曲线和实际曲线的对比,验证了该软测量模型具有很高的逼近精度.该软测量方法为动态流量的测量提供了一条新的途径.     

融合机理与数据的篦冷机温度软测量模型

针对篦冷机内部温度难以在线测量的问题,提出一种融合渗流换热机理与神经网络数据辨识的篦冷机温度软测量模型。首先考虑局部非热平衡效应,并结合多孔介质渗流换热理论建立床层内气固换热的机理模型;其次以床层边界条件与机理模型的温度预测值作为网络输入量,机理模型的温度预测偏差作为输出量,建立双并联前馈神经网络,并采用快速学习网算法训练网络参数;将辨识后的人工神经网络作为篦冷机气固渗流换热机理的补偿器,建立混合温度软测量模型。利用篦冷机生产运行数据进行了二次风温的软测量实验,实验结果表明,与机理模型相比混合软测量模型测量精度有较大提高,并且与实验数据的误差较小,混合软测量模型对于篦冷机温度具有较好的测量精度。     

基于RBF神经网络的汽包水位软测量模型研究

热工系统往往表现出非线性和不确定性．难以建立精确的数学模型。以汽包水位为对象．结合机理分析确定原始变量作为神经网络的输入．通过k均值聚粪法则与梯度下降法实现了网络的学习功能，并最终建立了基于RBF神经网络的软测量模型。阐述了RBF神经网络在汽包水位洲量中的建模与应用。仿真实验表明该模型具有简单易行．精度高、训练时间短、运算速度快的特点．为汽包水位测量提供了一种新的方法。     

基于PCA—改进BP算法的软测量技术

文中针对基本BP网络建立软测量模型时所存在的一些问题,例如：输入变量之间可能存在的线性相关等冗余性、基本BP算法收敛速度较慢而且易于限于局部最优等,本文尝试将主元分析与变尺度改进BP算法相结合,以提高软测量模型的训练速度和外推能力,为软测量技术的在线应用提供更大的方便。实验结果表明PCA方法与变尺度的改进BP神经元网络相结合的软测量建模方法在训练速度和外推能力方面有较大的改善。     

基于自适应模糊聚类的多模型染纱能耗软测量

针对染纱生产的工艺能耗测量问题,提出一种基于自适应模糊聚类的多神经网络的染纱能耗软测量方法.该方法采用自适应模糊C均值聚类算法,基于实时采集的样本数据,将训练集划分成不同聚类中心的子集,并自适应修正.每个子集用径向基函数网络训练得到子模型,然后根据聚类后的隶属度,将各子模型的输出加权求和获得最后结果.通过对染缸能耗软测量建模,并对其进行仿真和典型实例研究,表明该方法具有良好的预测精度和鲁棒性,且与制造执行系统结合具有良好的在线测量能力.     

粗集理论在污水参数软测量中的应用研究

用粗糙集理论约简属性,消除冗余信息后建立了污泥体积指数的神经网络软测量模型.用某城市污水厂实际水质参数进行仿真实验.仿真结果表明,与未采用粗糙集进行预处理的模型相比,应用该模型不仅测量值的误差更小,而且输入参数从9个降至4个,大大降低了输入数据的维数,减少了神经网络的训练时间及训练步数,有利于软测量模型的实用化.     

基于神经网络软测量的动态流量测量方法研究

文章以圆管中的粘性流体为研究对象,分析了管道中压差、温度、速度的变化对动态流量的影响,设计出了用于软测量建模的RBF神经网络结构;与此同时,还设计了动态流量测量管,并在伺服阀动态性能试验台上采集了相关数据,其间采用超声波类时差法测量了流速信息;利用采集的数据基于NeuroSolution软件对RBF网络进行了学习训练、检验和测试,最后建立了动态流量的软测量模型;通过将流量预测曲线与实测曲线相对比,结果发现该软测量模型具有很高的逼近精度,它可为动态流量的测量提供一条新的途径.     

Prediction of the discharge of side weir in the converge channels using artificial neural networks

Considering the common use of side weirs in irrigation and drainage networks, estimation of the discharge of the side weirs has always given a consideration by water engineering researchers. Another issue about side weirs is the change in flow conditions in the weir and downstream channel. To optimize the flow conditions in the side weir, this structure is established in a converging channel to reduce the channel width and compensate the reduced discharge. The geometrical parameters assumed as variables in this study are: weir length, weir height, convergence angle and downstream channel width. About 248 experiments were performed. Three neural network models were used to estimate the discharge from the side weir. The model was constructed using MATLAB, and the dimensionless variables that were the geometrical and hydraulic ratios of the model were selected as input parameters. Four ratios were selected as inputs to the model to estimate the discharge coefficient and the discharge from the weir. Considering the outputs of the model, the neural-fuzzy networks have the least error compared to the other models, and this model estimates the discharge of side weir overflowing with 99.8% accuracy.     

Analytical and experimental study of flow through elliptical side orifices

Discharge estimation is important for water management. Side orifices are commonly used in irrigation and drainage networks for distributing the water. Despite the vast amount of theoretical and experimental studies published, no generally applicable discharge equations are available for elliptical sharp-crested side orifices. When the length (diameter) of the circular side orifice is not sufficient to divert the water, an elliptical side orifice is a good alternative. In this paper, the elliptical sharp-crested side orifices were studied theoretically and experimentally. Several models were developed to predict the discharge coefficient of elliptical side orifices based on the Buckingham's theorem of dimensional analysis. A series of laboratory runs (588 runs) were conducted for different values of orifice geometry. The main channel discharge used in the tests ranged from 13.8 to 39.6 l/s and the side orifice discharge ranged from 3.66 to 21.4 l/s and upstream Froude number ranged from 0.22 to 0.77. Using measurements obtained by laboratory runs carried out in this investigation the proposed models of elliptical side orifices were calibrated under free outflow conditions. The model that includes the approach Froude number had an average error of 1.74%, while other model that does not include the approach Froude number had an average error about 2.43%. Moreover, suitable models were proposed for design procedure when measurement data for flow depths above the centroid of the orifice are not available. In this case, the model that includes the approach Froude number had an average error of 1.92%, while other model that does not include the approach Froude number had an average error of 2.24%. It was found that the proposed stage-discharge relationships were in an excellent agreement with the experimental results.     

基于径向基神经网络的明渠流量软测量方法

为克服现有明渠流量测量方法在监测自动化、测量准确度、测量成本和适用范围等方面存在的不足,在新兴的大尺度粒子图像测速(large-scale particle image velocimetry,LSPIV)技术的框架下,设计了一种基于径向基神经网络(radial basis function neural networ...     

The measurement of discharge using a commercial digital video camera in irrigation canals

This paper presents measurements of the discharge by image techniques on the surface velocity field and water stage in irrigation canals. The velocity and stage gauge are obtained from a commercial digital video camera. The time series of the surface velocity and stage were collected simultaneously. Particle image velocimetry (PIV) was used to determine the surface velocities in the irrigation canal. PIV proceeds by using bubbles floating on the water surface as tracer particles, and making a cross-correlation analysis between two continuous images. The whole surface velocity distribution in the irrigation canals can be obtained. The water stage of the canal is obtained from the digital video camera images by making use of image segments to separate the stage gauge and the background. The discharge is computed by using the surface velocities and water stage via open channel velocity distribution theory. Comparing the discharge measured using image techniques with Parshall flume data shows that the differences are less then 5%. The results suggest that the image measurement techniques developed can be used in applications to estimate the discharge in irrigation canals effectively.     

Discharge characteristics of sharp-crested circular side orifices in open channels

A side orifice is a flow diversion structure provided in one or both side walls of a channel to spill/divert water from the main channel. It is widely used in irrigation and environmental engineering. Analytical and experimental studies related to the discharge characteristics of sharp-crested circular side orifices in open channels under free flow conditions have been presented in this paper. Considering the side orifice as large, the discharge equation for the side orifice is derived analytically. Experiments were performed to estimate the coefficient of discharge which depends on the approach flow Froude number and ratio of the diameter of the orifice and bed width of the channel. Relationships for the coefficient of discharge, considering the orifice as large and small were developed. Such relationships were used to compute the discharge through the orifice for data not used for proposing such relationships for the coefficient of discharge. The computed discharges were within ±5% of the observed ones. The average percentage error in computation of discharge through the orifice considering it as large and small are, respectively, 1.59% and 1.66% which are practically the same. Therefore, it is recommended that the discharge through the side orifice can be computed considering it as a small orifice within the range of data used in the present study.     

Automatic recognition of machining features using artificial neural networks

We report on the development of an intelligent system for recognizing prismatic part machining features from CAD models using an artificial neural network. A unique 12-node vector scheme has been proposed to represent machining feature families having variations in topology and geometry. The B-Rep CAD model in ACIS format is preprocessed to generate the feature representation vectors, which are then fed to the neural network for classification. The ANN-based feature-recognition (FR) system was trained with a large set of feature patterns and optimized for its performance. The system was able to efficiently recognize a wide range of complex machining features allowing variations in feature topology and geometry. The data of the recognized features was post-processed and linked to a feature-based CAPP system for CNC machining. The FR system provided seamless integration from CAD model to CNC programming.     

Estimating the uncertainty of discharge coefficient predicted for oblique side weir using Monte Carlo method

Side weir is a hydraulic structure, which is used in irrigation systems to divert some water from main to side channel. It is installed at the entrance of the side channel to control and measure passing water into the side channel. Many studies provided side weir water surface profile and coefficient of discharge to measure water discharge diverted into the side channel. These studies dealt with different side weir shapes (rectangular, trapezoidal, triangular and circular), which were installed perpendicular to the flow direction. Recently, some studies dealt with skew side weir, but these studies still need to more investigation. Here we report to investigate oblique side weir theoretically using statistical method to supported other studies in this case. Measurement uncertainty discharge coefficient C_d was obtained by two methods: analytical according to the ‘Guide to the expression of uncertainty in measurement’ and the Monte Carlo method. The results indicate that all experimental results are consistent with the analytical results. The relative expanded uncertainty of the discharge coefficient C_d does not exceed 2%.     

Experimental study of simple flumes with trapezoidal contraction

There are different devices available for measuring the flow discharge. Static measuring devices such as weirs and flumes (without any moving parts) play a significant role in discharge measurement in open channels. Many researchers have focused on application of flumes in irrigation networks. This investigation set out to study the flow discharge through a trapezoidal cut-throated flume (TCTF). The flume is simply constructed by placing two triangular plates on either side of a rectangular open channel to form a trapezoidal throat. The channel cross section is rectangular while throat cross section of the flume is trapezoidal. The proposed flume is simple, low-cost, easy to install and does not require high maintenance. The present study was designed to determine the effect of different variables on the flume discharge. For this, an experimental study was carried out under free outflow conditions and under upstream subcritical flow regime to formulate the flume discharge. Three models are defined for flume discharge relation based on the Buckingham's theorem of dimensional analysis, and then calibrated using the experimental data collected during this study. The first discharge model has an average error of 1.81%, while the second and third improved models have average errors of 0.96% and 1.44% respectively. To reliably estimate the flume discharge, free-flow and submerged-flow conditions should be distinguished. For this, suitable equations with an average error less than 2.23% were presented to estimate the submergence threshold. The results of this study indicate that while the downstream wall slope influences the submergence threshold, the flume discharge is not influenced by this variable. The proposed models are simple and accurate and present appropriate estimation of discharge for flows through the TCTFs. The findings of this study will be of interest for practical hydraulic engineers.     

Discharge prediction in flow measurement flumes with different downstream transition slopes

Experimental testing of 9 different rectangular compound cross-section flow measurement flumes with different downstream slopes was conducted to yield the coefficient of discharge and the approach velocity coefficient. The aim of the experimental research was the determination of stage–discharge relationship in compound cross-section flow measurement flumes with different downstream slopes. One empirical predictive model for each of the coefficient of discharge and the approach velocity coefficient for the 9 cross-sections have been derived using one dimensionless parameter for the coefficient of discharge and another one dimensionless parameter for the approach velocity coefficient as the single independent variable. This approach is preferred as it allows the estimation of discharge by only measuring the water depth at the head measurement section. All obtained predictive models statistics have indicated the high reliability of the derived models in estimating discharge in an open channel flume of a rectangular compound cross-section using the predicted coefficients.