柴油机乙醇选择催化还原系统控制策略的研究

设计了一种氮氧化物(NOx)选择催化还原(SCR)试验台,简单介绍了试验台的系统组成和功能。为车载柴油机排气SCR净化系统提出了一种先进的控制策略,该控制策略主要包括还原剂喷射控制和助燃机剂喷射控制组成。对控制策略进行了试验研究,结果表明:使用该控制策略的SCR净化系统对车用柴油机排气NOx有较好的净化效果。     

燃气轮机燃烧基准温度估算方法

针对重型燃气轮机DLN(Dry Low NOx)燃烧控制中的燃烧基准温度计算问题,介绍了GE公司Mark系列控制系统中燃烧基准温度的计算方法,并基于其原理提出了一种简单的改进拟合方法。该方法可应用于电厂燃烧调整和国产燃气轮机的低污染燃烧室设计。     

基于EMD与ELM的光伏电站短期功率预测

针对传统光伏电站功率预测方法精度不高的问题,提出一种基于经验模态分解(EMD)与极限学习机(ELM)组合功率预测方法。该方法中,首先利用EMD分解分辨率为15 min的功率序列,得到一组相对平稳的分量,减少不同功率影响因素间的相互影响;然后针对各分量的不同特性,考虑相应气象因素作为输入,利用ELM神经网络建立不同的预测模型,分别预测各分量值;最后对ELM预测的各分量值求和,从而得到最终预测结果。算例仿真表明,该方法比传统的预测方法具有更高的预测准确度。     

极限学习机在初始地应力场反演中的应用

针对BP网络收敛速度慢、网络结构难确定、过拟合等缺点,提出将极限学习机(ELM)应用于区域地应力场反演计算中。利用FLAC3D软件进行正分析平衡计算,以实测点处应力值为网络输入,边界条件和岩石力学参数为网络输出,建立ELM反演模型。将该模型应用于某水电站地下厂房区地应力场反演分析中,计算结果与实测应力值拟合较好,说明该模型简单、实用,具有良好的泛化性和反演精度,可满足工程设计要求。研究成果为初始地应力场的反演计算提供了一种新思路。     

FCC再生装置NOx生成机理及控制技术

氮氧化物(NOx)是一种主要的大气污染物.在石化炼油企业中,催化流化再生装置(FCC)是NOx排放的主要来源.介绍了FCC再生装置中NOx的生成机理,认为再生过程中生成的NOx主要来自裂解过程中沉积在催化剂表面的含氮化合物的氧化.现有的控制技术包括原料的预处理、设计新型的再生器、添加deNOx助剂、安装独立的尾气净化装置等,其中添加deNOx助剂的方法具有使用方便、投资小等特点,被认为是最有前景的FCC再生装置NOx控制技术.     

基于特征选择的RELM风速短期预测

准确的风速预测是风电场功率预测的基础,对大规模风电并网具有重要的价值。文章提出一种基于信息增益(IG)的正则化极限学习机(RELM)短期风速预测方法。首先采用信息增益对32维风速属性序列进行特征选择,并对其进行加权;然后将正则化系数引入极限学习机(ELM)网络,构建RELM风速预测模型;最后结合美国风能技术中心的实测数据进行仿真,与传统ELM网络、BP神经网络相比,该方法具有较高的准确性和预测精度。     

Urea-SCR系统NO_x传感器的NH_3交叉感应研究

针对氨选择性催化还原(Urea-SCR)系统中NOx传感器对NH3产生的交叉感应现象,通过原理分析和试验观察,研究了排气温度对这种交叉感应特性的影响,建立了与排气温度有关的NOx传感器模型,结合SCR催化器模型,提出了一种运用扩展卡尔曼滤波预测催化器下游实际NOx体积分数的方法.仿真结果与试验结果对比表明,这种基于扩展卡尔曼滤波的方法能够显著提高NOx传感器读数的准确性,为NOx传感器满足Urea-SCR系统控制和车载诊断的要求提供了一种解决方案.     

内燃机燃料研究的新进展

2004年后对NOx排放的要求更加严格,采用以下两种措施有望达到法规要求.一种是使用NOx后处理器,另一种是改变柴油机扩散燃烧过程.本文介绍了柴油机燃烧过程研究的新近展情况,包括均匀充量压缩燃烧(HCCI)、可控自燃(CAI)、预混合稀气燃烧(PREDIC)和调谐动力(MK)等燃烧模式.     

在线监测燃煤锅炉NOx排放的自适应支持向量机模型

提出了一种基于改进的在线支持向量机自适应建模方法,并应用于电站锅炉NOx排放连续监测和特性分析.对常规在线支持向量机方法进行了改进,提出了新的样本剔除规则,保证了训练集内样本分布的均匀性.通过该改进方法对基于试验数据的常规向量机模型预测余差进行了连续估计,并预估煤质等因素引起的NOx排放特性的变化,从而补偿了实际工况与试验工况的差别,以便正确给出锅炉NOx排放特性.     

船用增压锅炉NO_x的生成机理及其排放量的近似计算方法

在分析船用增压锅炉NOx生成的化学反应机理基础上,提出了一种船用增压锅炉NOx排放量的近似计算方法。船用增压锅炉中NOx的生成主要包括热力型和燃料型两种,热力型NOx生成量可通过炉膛中有效温度、N2和O2在炉膛有效温度下的平衡浓度、燃烧产物在炉膛中的平均停留时间来近似计算;燃料型NOx生成量可通过燃料中氮的含量和平均转变率来近似计算。炉膛有效温度系数m″为0.9,燃料型NOx生成量平均转变率为0.36。对某船用增压锅炉NOx生成量的实际测量结果和计算结果的最大误差为5.6%,表明该计算方法可以满足船用增压锅炉NOx排放量的近似估算要求。     

基于连接权值算法的燃烧调整参数敏感性研究

为确认燃烧调整过程中影响燃气轮机运行状态的主导因素,首先将影响运行状态的天然气压力、天然气温度、压气机排气温度、压气机进口温度等14个参数作为输入变量,将表征燃气轮机运行状态的功率、燃烧室、加速度、NO_x质量浓度作为输出变量,建立粒子群算法优化的Elman神经网络模型,得到隐含层与输入层、输出层之间的连接权值;然后利用Olden方法处理神经网络的连接权值,获得各因素对燃气轮机运行状态影响显著性的量化值表达式,建立了燃烧调整过程中燃气轮机运行状态影响因素显著性分析的方法;最后结合燃气轮机运行数据进行计算分析。结果表明：燃气轮机的运行状态主要受排气温度、预混气压力及流量、值班气流量、压气机进口差压以及进气导流叶片开度5个因素的影响,并且燃烧调整过程中需要统筹调整输入参数,以保证燃气轮机燃烧始终处于稳定、低NO_x排放区域。     

基于改进PSO-ELM算法的混凝土坝变形非线性监控模型

针对混凝土坝变形模型高度非线性问题,将极限学习机(ELM)用于混凝土坝变形监控模型的构建中,由于极限学习机的精度受输入权值和隐含层阈值的影响,引入改进的粒子群算法(PSO)进行最优求解,从而建立基于改进PSO-ELM算法的混凝土坝变形非线性监控模型。实例应用结果表明,该模型不仅可行、有效,且具有较强的学习能力和泛化能力。     

基于IMRAN的电站锅炉效率与NO_x排放模型

电站锅炉高效低NOx燃烧优化技术越来越受到人们的重视,而锅炉燃烧效率和NOx排放模型是高效低NOx燃烧优化的基础.从提高网络的泛化能力着手,对最小资源分配网络算法的隐节点删减策略进行改进,加入惩罚策略和合并策略,并把改进的MRAN算法应用到对电站锅炉NOx排放与效率的实时建模上.仿真结果表明,改进的MRAN算法除了具有一般MRAN算法的优点外,还具有比MRAN网络更加紧凑的结构.提出的网络算法具有多输出结构,可同时预测NOx排放与效率,适于用在电站锅炉的NOx排放与效率的燃烧实时整体优化中.     

基于PLS与MPGA-ELM的汽轮机排汽焓预测模型

针对汽轮机排汽焓参数难以直接测量的问题,提出了一种利用偏最小二乘法(PLS)和多种群遗传算法(MPGA)优化极限学习机(ELM)的汽轮机排汽焓预测模型。先将采集到的数据进行预处理,然后通过PLS将多维的输入变量降维成低维相互独立的变量,再利用MPGA对ELM的初始权值与阈值进行优化,最后使用优化后的模型进行训练。并将该模型预测结果与BP、ELM以及SVM等模型预测结果进行对比,结果表明MPGA-ELM具有更低的误差、更高的模型预测精度及更强的泛化能力。     

基于CNN-LSTM的燃气轮机NOx排放预测研究

为了建立精准的NO_x预测模型,解决燃气轮机电站存在NO_x超标排放的问题,提出一种基于卷积神经网络(Convolutional Neural Network, CNN)和长短期记忆神经网络(Long Short-term Memory, LSTM)组合模型的NO_x排放预测方法。将NO_x排放历史数据和燃气轮机燃烧的状态参数通过滑动窗口法构建成特征图格式输入到CNN中,利用其卷积层和池化层提取表征NO_x动态变化的特征向量,并转化为时间序列格式输入到LSTM中进一步挖掘内部规律,从而实现NO_x的排放预测。以某三菱燃气轮机的历史运行数据进行试验。结果表明:CNN-LSTM的相对均方误差e_(RMSE)为1.811 mg/m~3,并通过与PCA-BP,PCA-RNN和PCA-LSTM模型进行比较,验证了方法的可行性。     

Prediction model for methanation reaction conditions based on a state transition simulated annealing algorithm optimized extreme learning machine

Methanation is the core process of synthetic natural gas, the performance of the entire reaction system depends on precise values of the reaction condition parameters. Accurate predictions of the CO conversion rate of the methanation reaction can eliminate time-consuming and complex steps in experiments and speed up the discovery of the best reaction conditions. However, the methanation reaction is an uncertain, highly complex, and highly nonlinear process. Thus, this paper proposes a machine learning prediction model for the methanation reaction to facilitate the subsequent search for optimal reaction conditions. The reaction temperature, pressure, hydrogen–carbon ratio, water vapor content, CO₂ content, and space velocity were selected as the condition variables. The CO conversion rate was the optimization objective. An extreme learning machine (ELM) was selected as a prediction model. Because the input weights and bias matrices of the ELM are randomly generated, an ELM based on a state transition simulated annealing (STASA-ELM) algorithm is proposed. The STASA algorithm was used to optimize the ELM to improve the accuracy and stability of the model. Five additional sets of experimental data were designed for the experiment, and the error between the experimental and predicted values was small. Thus, the STASA-ELM algorithm can accurately predict the conversion of CO for different values of reaction conditions.     

Inverse neural network for optimal performance in polygeneration systems

In this paper, inverse neural network (ANNi) is applied to optimization of operating conditions or parameters in energy processes. The proposed method ANNi is a new tool which inverts the artificial neural network (ANN), and it uses a Nelder-Mead optimization method to find the optimum parameter value (or unknown parameter) for a given required condition in the process. In order to accomplish the target, first, it is necessary to build the artificial neural network (ANN) that simulates the output parameters for a polygeneration process. In general, this class of ANN model is constituted of a feedforward network with one hidden layer to simulate output layer, considering well-known input parameters of the process. Normally, a Levenberg–Marquardt learning algorithm, hyperbolic tangent sigmoid transfer-function, linear transfer-function and several neurons in the hidden layer (due to the complexity of the process) are considered in the constructed model. After that, ANN model is inverted. With a required output value and some input parameters it is possible to calculate the unknown input parameter using the Nelder-Mead algorithm. ANNi results on three different applications in energy processes showed that ANNi is in good agreement with target and calculated input data. Consequently, ANNi is applied to determine the optimal parameters, and it already has applications in different processes with a very short elapsed time (seconds). Therefore, this methodology can be useful for the controlling of engineering processes.     

Application of extreme learning machine for estimating solar radiation from satellite data

In this paper, a simple and fast method based on extreme learning machine (ELM) for the estimation of solar radiation in Turkey was presented. To design the ELM model, satellite data of the National Oceanic and Atmospheric Administration advanced very high‐resolution radiometer from 20 locations spread over Turkey were used. The satellite‐based land surface temperature, altitude, latitude, longitude, month, and city were applied as input to the ELM, and the output variable is the solar radiation. To show the applicability of the ELM model, a performance comparison in terms of the estimation capability and the learning speed was made between the ELM model and conventional artificial neural network (ANN) model with backpropagation. The comparison results showed that the ELM model gave better estimation than the ANN model for the overall test locations. Moreover, the ELM model was about 23.5 times faster than the ANN model. The method could be used by researchers or scientists to design high‐efficiency solar devices such as solar power plant and photovoltaic cell. Copyright © 2013 John Wiley & Sons, Ltd.     

Diluted Air Combustion and NOx Emission in a HiTAC Furnace

CFD modeling of NOx emission via N₂O-intermediate mechanism was developed to predict the NOx formation in an experimental furnace equipped with high temperature air combustion (HiTAC) system. The good agreement between the predicted and measured results illustrates the superiority of using a N₂O-intermediate model in prediction of NOx emission during low peak temperature, which happens in HiTAC systems. Moreover, the CFD and measured results show that the flame volume as well as NOx emission significantly depends on temperature and oxygen concentration. Lower NOx emission was experimentally and numerically obtained at lower input oxygen concentration conditions.     

基于支持向量机和遗传算法的燃煤电站锅炉多目标燃烧优化

采用支持向量机方法建立350MW燃煤电站锅炉NOx预测模型和锅炉效率预测模型,并采用遗传算法对NOx和锅炉效率进行多目标优化,表明支持向量机和遗传算法可以用于指导参数调节,进行燃烧优化。