质子交换膜燃料电池膜电极组件温度分布的神经网络预测模型

质子交换膜燃料电池膜电极组件表面的温度分布会影响质子交换膜燃料电池的性能、寿命和可靠性.为探究质子交换膜燃料电池传热规律,本文提出了一种基于神经网络的质子交换膜燃料电池膜电极组件温度分布的预测模型.本研究选取径向基函数神经网络(RBF)和广义回归神经网络(GRNN)两种神经网络,以电流密度、温度点的位置作为网络输入,不同位置的温度作为网络输出,对平行流道质子交换膜燃料电池、蛇形流道质子交换膜燃料电池分别建立了神经网络预测模型.结果显示,RBF神经网络预测的均方根误差平均为0.464、平均绝对百分误差为1.179％,GRNN神经网络预测的均方根误差平均为0.7155、平均绝对百分误差为2.27％;相较于GRNN神经网络,RBF神经网络精度更高;基于RBF神经网络的平行流道质子交换膜燃料电池膜电极组件温度分布预测模型预测值与96％的实验值的相对误差在5％以内.基于RBF神经网络的蛇形流道质子交换膜燃料电池膜电极组件温度分布预测模型预测值与95％的实验值的相对误差在5％以内.     

自然吸气式CNG发动机主充模型的研究

为了准确预测自然吸气式压缩天然气(Compressed Natural Gas,CNG)发动机的空气流量,基于汽油机进气系统平均值模型,构建了CNG发动机的主充模型。根据CNG发动机进气系统的实际工作环境,引入了温度修正系数、体积修正系数、燃气量修正系数等参数,计算了不同工况下CNG发动机的空气流量。通过CNG发动机台架试验,测量了不同转速、不同进气歧管压力下的空气流量;对比分析了空气流量的计算值和试验值的方法,评估了模型的预测性能。结果表明,所建立的主充模型能较好的预测不同转速下空气流量随CNG发动机进气歧管压力的变化规律,空气流量的预测值与实验值的最大误差小于3%,模型具有较高的预测精度。     

基于BP和RBF神经网络的乙烷脉动热管对比研究

为研究BP和RBF神经网络对脉动热管热阻的预测及改善脉动热管性能,将加热功率、倾角及工作温区作为输入参数,热阻作为输出参数,建立BP和RBF神经网络模型。利用大量实验数据对BP及RBF神经网络进行训练并预测,将预测值与实验值比较,以验证BP和RBF神经网络预测性能。结果表明:BP和RBF神经网络均能较好地预测热阻;采用RBF神经网络,训练数据及测试数据线性回归决定系数R~2分别为0.999 44和0.969 76,预测结果相对误差分别为0.89%和2.97%,均方误差分别为1.43×10~(-7)和3.13×10~(-6);与BP神经网络相比,线性回归决定系数R~2更接近1,相对误差和均方误差更小,能更精确地预测热阻。     

基于不同优化算法下风电预测精度对风水电协调影响研究

研究了持续法、ARIMA方法、改进BP神经网络3种不同的风电预测模型,在相同条件下,经实例仿真发现,改进BP神经网络模型的预测精度好于ARIMA预测模型,而ARIMA预测模型的预测精度好于持续法预测模型.基于上述3种不同的风电预测模型,建立了风-水发电联合协调运行的模型.采用遗传粒子群和混合粒子群2种不同的优化算法来研究风电预测精度对风-水电协调影响,通过仿真实例发现,风电预测模型精度越高,得到的理论值与实际值偏差越小;在考虑2种不同优化算法的情况下,遗传粒子群优化算法得到的数值与实际值偏差比混合粒子群大,同时基于不同风电预测模型下的误差也要比混合粒子群大.     

基于果蝇优化算法的GRNN水电机组状态趋势预测

针对水电机组振动的非线性、非平稳特性,提出了一种基于果蝇优化算法（FOA）的广义回归神经网络（GRNN）模型（FOAGRNN ）,实现了GRNN分布参数的优化选择,并对四川省新政航电工程3台机组5个不同部位的振动序列峰峰值进行了预测,与BP神经网络预测结果的均方误差（MSE）对比结果表明,FOAGRNN预测精度较高。     

车用汽油机过渡工况空燃比的神经网络多步预测控制策略

车用汽油机空燃比存在传输延迟,直接用于反馈控制影响空燃比的控制精度.为此,提出了一种基于神经网络的空燃比多步预测控制策略,首先建立了基于BP神经网络的空燃比多步预测模型,利用空燃比预测模型预报空燃比的未来值,利用预测值与期望值的误差及误差变化率,采用模糊控制器对空燃比实施多步预测控制.对HL495发动机两种典型过渡工况实验数据进行仿真,结果表明,该方法能将过渡工况空燃比控制在理论空燃比的±3%以内.     

自适应粒子群支持向量机风速组合预测模型

提出一种优化的支持向量机风速组合预测模型,首先通过模糊层次分析法对参与组合的单项预测模型进行遴选,在当前风速样本集下自适应决策预测效果较优的单项预测模型的输出值作为支持向量机的输入,将实际风电场风速值作为支持向量机的输出,并采用粒子群算法优化支持向量机组合模型的参数。基于实际运营的风电场数据进行仿真分析,自适应遴选出BP神经网络、RBF神经网络、小波神经网络和遗传算法优化BP神经网络这4种单项预测模型参与支持向量机组合,结果表明所提方法的预测精度不仅高于单项模型,且高于线性组合预测模型和神经网络组合预测模型。     

BP/RBF神经网络与模糊规则耦合的电站锅炉燃烧控制

针对电站锅炉NO_x浓度和发电效率的非线性及复杂耦合关系问题,分别建立某320 MW火电机组RBF神经网络模型、BP神经网络模型和模糊规则模型。采用满负荷70%～80%的常规工况进行训练,RBF神经网络有效地预测了发电效率及NO_x排放浓度,平均相对误差分别为2.03%和2.41%。根据专家经验制定25条模糊控制规则,将RBF神经网络的输出值作为模糊控制器输入值,对锅炉运行参数进行调整,并将调整后的值输入BP神经网络进行预测。结合RBF/BP神经网络和模糊控制规则建立了综合优化模型,使NO_x调整值相对于实际值平均下降了7.89 mg/m~3,发电效率提高了1.08%。     

基于最优权系数的风电功率组合预测方法

摘要： 风电功率的短期预测对于电力系统的安全稳定运行具有重要意义。提出了一种基于最优权系数的风电功率短期预测组合方法,该方法将ARIMA时间序列、BP神经网络、RBF神经网络和支持向量回归机这4种单一预测模型进行综合,并根据预测误差信息矩阵,以误差平方和最小为原则得到组合预测模型中的最优权系数,以此构成组合预测模型,该模型能够有效地综合各单一预测模型的优势,降低预测风险。仿真实例表明：所提组合预测模型预测精度高,能够方便快速地确定最优权重系数值,降低预测误差。     

基于IFOA-GRNN的电力系统短期负荷预测

考虑到电网负荷与诸多因素有关,设计了一种带有温度、气象、日期类型的广义回归神经网络（GRNN）负荷预测模型。为了提高该模型的预测精度,提出了一种改进果蝇优化算法优化广义回归神经网络（IFOA-GRNN）的方法,即在利用果蝇优化算法（FOA）进入迭代寻优时,通过改进搜索距离优化该算法的性能和稳定性。利用改进的FOA优化GRNN的光滑参数,然后利用训练好的预测模型对甘肃省某地区进行了短期负荷预测,并与FOA-GRNN和误差反向传播神经网络(BPNN)模型结果进行了误差比较。结果表明, IFOA-GRNN具有较高的预测精度,能够满足电力系统短期负荷预测的要求。     

天然气发动机神经网络控制策略及试验研究

天然气发动机本身具有低排放的特点,所以不需要由排放参数反馈修正发动机的空燃比,减少了控制的反馈系统。传统的混合器式天然气发动机无法高精度地控制发动机的空燃比,采用神经网络控制理论对天然气发动机的空燃比实施控制,使其空燃比始终保持在理论空燃比高精度的状态,提高了天然气发动机的性能。以175F天然气发动机为例,试验结果表明:采用神经网络控制的天然气发动机的性能高于混合器式天然气发动机。     

基于RBF神经网络的电喷天然气发动机空燃比控制

电喷天然气发动机空燃比在工况突变时会剧烈波动,为提高控制精度,使用RBF神经网络和前馈PID控制算法相结合,由当前工况决定燃料基本喷射量,再由RBF神经网络预测的空燃比信号传递给PID控制器进行反馈调节。此外利用Matlab/Simulink软件进行仿真,建立了电喷天然气发动机空燃比仿真模型。仿真结果表明,该控制方法在节气门及发动机转速突变的情况下,过渡时间较PID算法明显减小,控制精度亦有提高。     

Optimization control of hydrogen engine ignition system based on ACO-BP

With the development of new energy, hydrogen fuel engines have become a research boom in the automotive field. But there are abnormal problems such as backfire and pre-ignition during the combustion of hydrogen engines. This paper is based on the Ant Colony Optimization-Back Propagation (ACO-BP) algorithm to study the influence of different speed and load conditions on the ignition advance angle, so as to optimize the control of the hydrogen engine. The experimental system is established on a hydrogen engine converted from a 492Q gasoline engine. The prediction of the optimal ignition advance angle was obtained through experiments, and the optimal ignition MAP diagram of the hydrogen engine is constructed. The optimal ignition advance angle under different working conditions can effectively avoid the occurrence of hydrogen engine pre-ignition. The accuracy reaches 0.0018209 when the training reaches 14 times, the fitness between the actual value and the predicted value of ACO-BP training is 0.99921, the verification accuracy reaches 0.99913, and the test accuracy reaches 0.99932. Compared with the three optimization methods, the convergence speed and error accuracy of ACO-BP are significantly better than the BP neural networks and Genetic Algorithm-Back Propagation (GA-BP). This method realized the model of the nonlinear mapping model from hydrogen engine speed and load to optimal ignition advance angle, which is of great significance for solving the problem of abnormal combustion in hydrogen engine.     

基于优化PSO-RBF的柴油机故障诊断方法

针对传统RBF算法收敛速度慢,易于陷入局部极值的问题,提出了一种经优化的粒子群算法PSO,对RBF神经网络粒子群的改进参数、权值线性递减参数和标准参数进行训练寻优,构建出最优PSO-RBF神经网络,并将其用于柴油机的故障诊断预报。对MAN B&W 6L23/30H柴油机三种不同工况下第一缸试验参数的训练表明:改进的PSO-RBF神经网络在柴油机故障诊断中判别率更高,故障诊断的准确性与可靠性得到提高。     

天然气电控发动机的匹配研究

提出了正交试验法与神经网络相结合建立发动机性能与运转参数间的动力性模型、经济性模型及排放模型的方法,并以这些模型为基础建立目标函数和约束条件,采用遗传算法进行运转参数实时优化。在913CNG天然气电控发动机上进行的动力性能和排放性能匹配试验结果表明其动力性能达到了设计指标,排放性能满足国家标准要求,采用的方法切实可行。     

柴油机涡轮增压器匹配程序中排气温度预测模型的研究

介绍了三种常用柴油发动机排温计算模型在自主开发的增压系统匹配软件TC-Match中的应用和对比,利用此三种模型对三种发动机分别进行了排温计算,通过计算结果与实验数据比较,分析了排温模型对不同发动机的计算误差以及参数对增压器匹配计算准确度的影响,选定热力学第一定律修正模型为TCMatch程序的排气温度预测模型,提高了TCMatch软件的计算精度,可以满足不同发动机的排温计算要求。     

CNG-diesel engine performance and exhaust emission analysis with the aid of artificial neural network

This study investigates the use of artificial neural network (ANN) modelling to predict brake power, torque, break specific fuel consumption (BSFC), and exhaust emissions of a diesel engine modified to operate with a combination of both compressed natural gas CNG and diesel fuels. A single cylinder, four-stroke diesel engine was modified for the present work and was operated at different engine loads and speeds. The experimental results reveal that the mixtures of CNG and diesel fuel provided better engine performance and improved the emission characteristics compared with the pure diesel fuel. For the ANN modelling, the standard back-propagation algorithm was found to be the optimum choice for training the model. A multi-layer perception network was used for non-linear mapping between the input and output parameters. It was found that the ANN model is able to predict the engine performance and exhaust emissions with a correlation coefficient of 0.9884, 0.9838, 0.95707, and 0.9934 for the engine torque, BSFC, NO_x and exhaust temperature, respectively.     

基于RBF神经网络的厂修舰船柴油机状态判断技术

利用柴油机爆压、排温、振动噪声等参数,建立了RBF柴油机技术状态识别网络,以大量维修实践中得到的样本进行网络训练,检验结果证明该网络町实现柴油机技术状态判断,取得了预期的效果.     

Neural network model for a commercial PEM fuel cell system

Performance prediction of a commercial proton exchange membrane (PEM) fuel cell system by using artificial neural networks (ANNs) is investigated. Two artificial neural networks including the back-propagation (BP) and radial basis function (RBF) networks are constructed, tested and compared. Experimental data as well as preprocess data are utilized to determine the accuracy and speed of several prediction algorithms. The performance of the BP network is investigated by varying error goals, number of neurons, number of layers and training algorithms. The prediction performance of RBF network is also presented. The simulation results have shown that both the BP and RBF networks can successfully predict the stack voltage and current of a commercial PEM fuel cell system. Speed and accuracy of the prediction algorithms are quite satisfactory for the real-time control of this particular application.     

Experimental study of hydrogen enriched compressed natural gas (HCNG) engine and application of support vector machine (SVM) on prediction of engine performance at specific condition

The effect of excess air ratio (λ) and ignition advance angle (θ_ig) on the combustion and emission characteristics of hydrogen enriched compressed natural gas (HCNG) on a 6-cylinder compressed natural gas (CNG) engine has been experimental studied in an engine test bench, aiming at enriching the sophisticated calibration of HCNG fueled engine and increasing the prediction accuracy of the SVM method on automobile engines. Three different fuel blends were selected for the experiment: 0%, 20% and 40% volumetric hydrogen blend ratios. It is noted that combustion intensity varies with the excess air ratio and the ignition advance angle, so are the emissions. The optimal value of λ or θ_ig has been explored in the specific engine condition. Results show that blending hydrogen can enhance and advance the combustion and stability of CNG engine, and it also has some benefic influence on the emissions such as reducing the CO and CH₄. Meanwhile, a simulation research on forecasting the engine performance by using the support vector machine (SVM) method was conducted in detail. The torque, brake specific fuel consumption and NO_x emission have been selected to characterize the power, economic and emissions of the engine with various HCNG fuels, respectively. It can be seen that the optimal model built by the SVM method can highly describe the relationship of the engine properties and condition parameters, since the value of the complex correlation coefficient is larger than 0.97. Secondly, the prediction performance of the optimal model for torque or BSFC is much better than the case of NO_x. Besides, the optimal model built by the PSO optimization method has the best prediction accuracy, and the accuracy of the model obtained based on the training group with 20% hydrogen blend ratio is the best compared with those of others. The upshots in this article provide experimental support and theoretical basis for the adoption of HCNG fuel on internal combustion engines as well as the application of intelligent algorithmic in the engine calibration technology field.