基于EMD-TCN的多能供热系统太阳辐照度预测模型

针对太阳辐照度的非平稳性和非线性影响多能供热系统运行效率和可靠性问题,该文提出一种基于经验模态分解（EMD）和时间卷积网络（TCN）的太阳辐照度混合预测模型EMD-TCN,更精准地从气象数据中提取太阳辐照度非线性和非平稳的隐含特征,获得更佳的预测精度。该研究利用逐时气象数据对所提出的EMD-TCN模型进行不同时间尺度的太阳辐照度预测实验,并与4种主流深度学习预测算法进行对比分析,结果表明该太阳辐照度预测模型具有更高的预测精度和泛化能力。     

基于多维特征分析的双层协同太阳辐照度预测

为增强逐日太阳辐照度预测的准确性和普适性,提出一种基于多维特征分析的双层协同预测模型。首先,搭建一种双层协同架构,将整个模型分成基准层和提升层两部分,使用分层预测的方式追踪目标对象的多维特征和变化趋势;其次,以数值天气预报（NWP）为输入,采用LightGBM基于特征学习预测方法构建基准预测模型;然后,在前者的基础上,挖掘目标时刻太阳辐照度与历史时序数据之间的关联性,引入改进AdaBoost算法与多隐层极限学习机（MH-ELM）作为提升层主体,提高时序预测的稳定性;最后,选用中国中部地区某光伏电站实测太阳辐照度数据进行算例分析,验证了该模型的合理性和有效性。     

利用蝙蝠算法优化SVR的太阳辐照度预测方法研究

为了提高太阳辐照度的预测精度,提出一种利用蝙蝠算法(BA)优化支持向量回归(SVR)的太阳辐照度预测方法。首先,确定SVR预测器的基本结构,选取环境温度、云量、风速、风向、环境湿度以及年积日等与太阳辐照度关系较为紧密的气象监测数据,构成SVR的输入特征向量,将待预测时段小时平均太阳辐照度作为SVR的输出;然后,以预测精度为判断依据,利用蝙蝠算法对SVR的惩罚因子和RBF核函数方差进行寻优;最后,利用最优参数建立SVR预测模型,并对太阳辐照度进行预测。分析结果表明,相比于无参数优化SVR预测模型和利用粒子群算法优化SVR模型的太阳辐照度预测方法,文章所提出的预测方法具有更高的预测精度。     

基于二阶数据分解算法和蝗虫优化混合核LSSVM的太阳辐照度预测模型研究

针对太阳辐照度时间序列的非线性特点,文章设计了一种新的基于二阶数据分解算法和蝗虫优化混合核LSSVM的太阳辐照度预测模型,并对该模型进行了验证。首先,利用集合经验模态分解(EEMD)算法对原始太阳辐照度时间序列进行分解,得到若干个频率不同的分量;然后,利用变分模态分解(VMD)算法进一步分解频率最高的分量,得到K个相对稳定的分量,其中,K由各分量与利用VMD算法分解得到的残差的相关系数确定;接着,建立基于高斯核和多项式核的混合核最小二乘支持向量机(LSSVM)预测模型,对所有分量进行预测,并利用蝗虫优化算法优化混合核函数的参数;最后,将所有分量的预测结果相加得到原始太阳辐照度时间序列的预测结果。模拟结果表明,与BP神经网络模型、ARIMA模型、LSSVM模型和基于EEMD,LSSVM的预测模型相比,基于二阶数据分解算法和蝗虫优化混合核LSSVM的太阳辐照度预测模型的预测精度更高,能有效反映太阳辐照度的变化规律。     

硅电池辐照计非线性模型与散射模型的修订

文章针对高、低太阳辐照度情况下,硅电池辐照计测量数据失真,以及太阳散射辐照度计算模型中水平面与倾斜面太阳辐照度比值系数较为单一,不能满足多种天气情况下太阳散射辐照度的测量需求,提出了电压与非线性太阳辐照度计算模型的修订方法,以及适应多种天气情况的太阳散射辐照度修正系数的修订方法。实验结果表明,对于修订后的非线性太阳辐照度计算模型以及太阳散射辐照度修正系数得到的实验数据,其中,80%的实验数据的误差在±10%之间,因此,文章的模型和修正系数均比较合理,可适应多种天气情况,并提高了硅电池辐照计测量结果的可靠性。     

基于长短期记忆神经网络的地表太阳辐照度预测

针对地表太阳辐照度(GHI)短期预测问题,提出一种基于长短期记忆神经网络的短期太阳辐照度预测模型。采用递归结构的训练样本,以保证训练样本内部的时间耦合性。为验证所提模型预测GHI的有效性,采用算例与传统人工神经网络模型预测结果进行对比分析。结果表明:基于长短期记忆神经网络预测模型将均方误差降低88.48%,表明所建模型更适用于GHI预测。     

基于Kmeans++-Bi-LSTM的太阳辐照度超短期预测

针对地表太阳辐射的不确定性和随机波动性，进而对大型光伏发电并网对电力系统的稳定性造成冲击，提出一种新的太阳辐照度超短期预测方案。该方案通过使用皮尔逊相关性分析和无监督学习中的Kmeans++算法，对多种气象数据进行筛选，找出关键气象数据并进行划分以及添加标签，接着将带有标签的关键气象数据输入双向长短期记忆网络预测模型中，以达到10 min时间间隔的太阳辐照度超短期预测。结果表明所提预测模型相较于目前常用的模型提高了预测精度。     

基于ICEEMDAN-LSTM和残差注意力的短期太阳辐照度预测

为提升短期太阳辐射预测的准确性，提出一种基于ICEEMDAN-LSTM和残差注意力的短期太阳辐照度预测方法。该方法利用改进的自适应噪声完备集合经验模态分解（ICEEMDAN）将原始辐射序列分解为多尺度模态分量，同时引入残差注意力机制对原始气象特征进行重构，然后利用长短期记忆网络分别提取两部分的时序特征，并融合所得特征输入至多层感知器，进行提前1小时的水平面总辐照度预测。实验结果表明，该方法能捕捉辐射序列的波动和突变，并考虑不同气象特征的重要程度，可有效提高短期太阳辐照度的预测精度。     

基于非线性自回归神经网络的GHI预测

针对水平面总辐照度(global horizontal irradiation,GHI)短期预测问题,提出一种基于非线性自回归神经网络的短期水平面太阳总辐照度预测模型。首先,提出一种并联结构训练样本,以保证训练样本内部的时间耦合性。其次,通过对9项气象参数共511种组合作为输入的模型预测精度进行分析,确定模型最优输入组合。最后,利用4种典型气象条件下GHI时延神经网络预测模型,非线性自回归动态神经网络预测模型预测标准均方根误差均降低。     

基于BP神经网络的PTC工质出口温度研究

在槽式抛物面太阳集热器的热性能研究中,数据往往具有随机性、非线性和不确定性等特点,采用传统建模方法经常做出大量假设,导致仿真精度不高且复杂。以槽式抛物面太阳集热器为研究对象,将传统理论模型与BP人工神经网络相互耦合,通过集热器热性能室外动态试验,建立工质出口温度的神经网络预测校正模型。引入Levenberg-Marquardt(LM)法对BP神经网络的权值及阈值进行优化。分析结果表明,预测校正模型可将绝对误差控制在3.8℃以内,相对误差保持在3.6%以内,可有效提高槽式抛物面太阳集热器热性能的仿真模型计算精度。     

基于特征变权的超短期风电功率预测

针对当前风电功率预测过程中历史信息利用不充分及多维输入权重值固定忽略了不同时间维度的特征重要性的问题,提出一种基于特征变权的风电功率预测模型。该方法利用随机森林（RF）分析不同高度处的风速、风向、温度等气象特征对风电输出功率的影响程度,并利用累积贡献率完成气象特征的提取。对提取的特征及历史功率信息利用奇异谱分析（SSA）去噪,以去噪后的数据作为输入建立级联式FA-CNN-LSTM多变量预测模型对超短期风电功率进行预测。通过在CNN-LSTM网络中增加特征注意力机制（FA）自适应挖掘不同时刻的特征关系,动态调整不同时间维度各输入特征的权重,加强预测时刻关键特征的注意力,从而提升预测性能。基于某风电场实测数据的算例分析表明,所提方法可有效提高超短期风电功率预测精度。     

基于改进递归深度信念网络的CSP电站短期出力预测

以预测CSP电站短期出力为目的,首先引入自适应思想对递归深度信念网络的训练算法进行改进,并建立直接法向辐射的短期预测模型。随后提出一种结合静态模型的CSP电站短期出力预测方法。最后进行性能检验,验证了改进递归深度信念网络的可行性,以及CSP电站短期出力预测方法的有效性。研究结果表明：建立的改进递归深度信念网络可提升预测准确性和收敛速度;提出的CSP电站短期出力预测方法可较为准确地预测其短期出力情况。     

Influence of neural network training parameters on short-term wind forecasting

This paper investigates factors which can affect the accuracy of short-term wind speed prediction when done over long periods spanning different seasons. Two types of neural networks (NNs) are used to forecast power generated via specific horizontal axis wind turbines. Meteorological data used are for a specific Western Australian location. Results reveal that seasonal variations affect the prediction accuracy of the wind resource, but the magnitude of this influence strongly depends on the details of the NN deployed. Factors investigated include the span of the time series needed to initially train the networks, the temporal resolution of these data, the length of training pattern within the overall span which is used to implement the predictions and whether the inclusion of solar irradiance data can appreciably affect wind speed prediction accuracy. There appears to be a relatively complex relationship between these factors and the accuracy of wind speed prediction via NNs. Predicting wind speed based on NNs trained using wind speed and solar irradiance data also increases the prediction accuracy of wind power generated, as can the type of network selected.     

Experimental verification of some clear-sky insolation models

Clear-sky solar irradiance can be predicted when a number of essential atmospheric parameters are known. A number of parameterization methods to predict solar irradiance with various degrees of difficulty are available in the literature. In this study, three models called model A, model B and model C, with medium degree of difficulty, have been examined. In these models, the solar transmittance due to each atmospheric parameter is available in simple algebraic form. Based on these algebraic equations, the direct normal, diffuse, and global horizontal irradiance can be predicted. These models have been compared with measured data from Carpentras, a French radiometric station. At this station, several daily observations of the clear-sky irradiance are carried out. Corresponding instantaneous values of the Ångström turbidity coefficient β and several other necessary surface meteorological observations are also made. For diffuse irradiance, a value of 0.95 is assumed for the single-scattering albedo of the aerosols. Based on the calculation of the mean bias error and root mean square error, model C has the best correspondance with the measurements as for as direct irradiance is concerned. Model B appears to be more accurate for prediction of diffuse and global irradiance.Regression equations are provided to help the user of any one of the three models for better prediction of solar irradiance.     

Study of models for predicting the diffuse irradiance on inclined surfaces

Solar irradiance data on various inclined surfaces at different orientations are important information for active solar-system analyses and passive energy-efficient building designs. In many parts of the world, however, the basic solar irradiance data for the surfaces of interest are not always readily available. Traditionally, different mathematical models have been developed to predict the solar irradiance on various inclined surfaces using “horizontal” data. Alternatively, the diffuse irradiance of a sloping plane can be calculated by integrating the radiance distribution generated with a sky radiance model. This paper presents the evaluation of two slope irradiance models, namely, the Perez point-source model (PEREZSIM) and the Muneer model (MUNEERSIM), and two sky-distribution models, namely, the Perez all-weather model (PEREZSDM) and the Kittler standard-sky model (KITTLERSDM). Three-year (1999–2001) measured average hourly sky radiance and horizontal sky diffuse irradiance data were used for the model assessment. Statistical results showed that all four models can accurately predict the solar irradiance of a 22.3° (latitude angle of Hong Kong) inclined south-oriented surface, indicating the good predictive ability for modelling an inclined surface with a small tilted angle. In general, the KITTLERSDM and PEREZSIM show the best predictions for vertical solar irradiance at this location, followed by the PEREZSDM, then the MUNEERSIM.     

基于风速波动特征提取的超短期风速预测

针对风电场风速预测准确度不高的问题,提出一种基于风速波动特征提取的超短期风速预测方法。首先建立风速-风速变化量联合概率密度模型,分析风速的不确定性特征;根据风速波动特征,应用集合经验模态分解（EEMD）和风速分量样本熵（SampEn）值,将风速分解重组为波动量和趋势量;应用人工鱼群算法（AFSA）优化小波神经网络（WNN）进行趋势量预测;应用改进非线性自回归（INARX）神经网络对风速波动量进行预测,进而得到预测风速。通过实际风电场风速仿真预测,并与多种预测方法对比,表明该预测方法预测结果误差较小,可准确地进行超短期风速预测。     

Predicting solar irradiance on inclined surfaces using sky radiance data

The availability of more comprehensive solar irradiance data is invaluable for the reduction of cooling load in buildings and for the evaluation of the performance of photovoltaic plants. In many parts of the world, however, the basic solar irradiance data are not always readily available. This paper presents an approach to calculate the solar irradiance on sloped planes by integrating the sky radiance distribution. Sky radiance data recorded from January 1999 to December 2001 in Hong Kong were used to estimate the solar irradiance for the horizontal and four principal vertical surfaces (N, E, S and W). The performance of this approach was assessed against data measured in the same period. Statistical results showed that using sky radiance distributions to predict solar irradiance can give reasonably good agreement with measured data for both horizontal and vertical planes. The prediction approach was also employed to compute the solar irradiance of a 22.3° (latitude angle of Hong Kong) inclined south oriented surface. The findings indicated that the method can provide an accurate alternative to determine the amount of solar irradiance on inclined surfaces facing various orientations when sky radiance data are available.     

Solar radiation over Sudan—Comparison of measured and predicted data

Measurements of global solar irradiance on a horizontal surface at 14 meteorological stations in Sudan are compared with predictions made by two independent methods. The first method is based on Angstrom formula which correlates relative global solar irradiance to corresponding relative duration of bright sunshine . Regional regression coefficients are obtained and used for prediction of global solar irradiance. The agreement with measurements is better than 7.5 per cent. In the second method an empirical relation due to Barbaro et al. which uses sunshine duration and minimum air mass as inputs is employed. An appropriate regional parameter is determined and used to predict solar irradiance at all stations with an accuracy better than 8 per cent. A comparison of the two methods is presented. Estimation of diffuse solar irradiance by Page's as well as Liu and Jordan's correlations is also performed and the results are examined.     

基于系统辨识的地表太阳辐射Box-Jenkins模型的建立

利用天文辐射作为输入数据,采用系统辨识的方法得到地表太阳辐射的BJ(Box-Jenkins)模型,并通过残差分析和零极点检验。该方法可用于预测5～15min时间间隔的地表太阳辐射,为太阳能电站的功率输出预测提供太阳能辐射数据。