社会快速发展伴随着能源使用需求的提升，为响应绿色能源发展战略的号召，光伏发电得到了大力扶持与发展。光伏发电主要依靠光伏组件对太阳光照能量进行转化，由于天气环境的变化导致光伏组件输出功率存在一定波动性与不稳定性，并网时会对电网的平稳运行造成冲击，所以需要对光伏功率进行预测来调整并网运行的结构与策略。传统光伏功率预测主要以温湿度、辐照度等因素作为输入，但实际上长期运行的光伏组件表面一般会有积灰附着，积灰会影响光照的透过率，进而减少太阳辐照度造成功率下降。因此，本文针对积灰影响下的光伏组件输出功率进行预测研究。

首先论文对传统光伏功率预测研究现状进行论述，同时介绍积灰对光伏组件输出功率的影响，综合以上本文对考虑积灰影响的光伏功率预测进行研究。为了得到积灰影响下的光伏组件发电数据，本文以小功率光伏组件为实验对象，自行设计数据采集平台，使用两块相同的光伏组件进行积灰与洁净的对照实验。采集温湿度、太阳辐照度、积灰密度以及两块组件的发电数据。根据采集数据分析能够看出在无降雨冲洗情况下，西安市夏季天气光伏组件积灰密度与积灰天数基本成正比。组件积灰7天后平均输出功率下降4.82%，积灰15天后平均输出功率下降10.54%，对比每日组件积灰密度和输出功率衰减率曲线，能够看出二者呈现极强的正相关性，说明当其他条件基本一致的情况下，积灰的存在会对组件输出功率造成较大影响。

其次选定支持向量机（Support Vector Machine，SVM）作为功率预测的模型，同时为了提高模型的精度提出采用智能优化算法秃鹰算法（Bald Eagle Search，BES）对SVM进行参数优化。并针对原始BES存在的固定取参局限性与迭代后期种群多样性下降等问题，引入Tent混沌映射、自适应变化参数与高斯变异来对BES算法进行改进。为了验证改进算法（IBES）的性能，采用6种函数进行算法求解测试，并与粒子群算法（Particle Swarm Optimization，PSO）、灰狼算法（Grey Wolf Optimizer，GWO）以及原始秃鹰算法进行对比。测试结果表明，IBES算法在迭代速度和寻优精度上均高于其他3种算法，验证了改进算法的优越性。

最后对积灰光伏组件进行功率预测。为提高预测精准度，将采集数据通过灰色关联分析法分为晴天与多云分别进行预测。预测结果表明，引入积灰密度后在晴朗天气下预测结果精度均优于常规预测结果，且IBES-SVM效果最好，其平均绝对误差、均方根误差、绝对平均百分比误差分别为0.7587、1.2318、0.1515。在多云天气预测结果中，PSO-SVM在引入积灰密度后预测误差出现增长，其余3种算法优化模型均在引入积灰密度后实现不同程度的预测精度上升，其中IBES-SVM效果仍为最佳，三种误差评价指标分别为0.7765、1.0735、0.2861。综合各项预测结果表明，在光伏功率预测中考虑积灰因素的想法是可行的，且IBES-SVM具有较好的预测效果。

The rapid development of society is accompanied by the improvement of energy use demand. In response to the call of green energy development strategy, photovoltaic power generation has been vigorously supported and developed. Photovoltaic power generation mainly relies on photovoltaic modules to convert solar light energy. Due to changes in the weather environment, the output power of photovoltaic modules has certain volatility and instability. When the grid is connected, it will impact the smooth operation of the grid. Therefore, it is necessary to predict the photovoltaic power to adjust the structure and strategy of grid-connected operation. Traditional photovoltaic power prediction mainly takes temperature, humidity, irradiance and other factors as input, but in fact, the surface of long-running photovoltaic modules generally has ash deposition, which will affect the transmittance of light, thereby reducing solar irradiance and causing power decline. Therefore, this paper predicts the output power of photovoltaic modules under the influence of ash deposition.

Firstly, the paper discusses the research status of traditional photovoltaic power prediction, and introduces the influence of ash deposition on the output power of photovoltaic modules. Based on the above, this paper studies the photovoltaic power prediction considering the influence of ash deposition. In order to obtain the power generation data of photovoltaic modules under the influence of ash deposition, this paper takes low-power photovoltaic modules as the experimental object, designs a data acquisition platform, and uses two identical photovoltaic modules for ash deposition and clean control experiments. Collect temperature and humidity, solar irradiance, ash density and power generation data of the two components. According to the analysis of the collected data, it can be seen that in the absence of rainfall washing, the ash density of photovoltaic modules in summer weather in Xi 'an is basically proportional to the number of ash days. The average output power of the module decreased by 4.82 % after 7 days of ash deposition, and the average output power decreased by 10.54 % after 15 days of ash deposition. Compared with the daily ash deposition density and output power decay rate curve, it can be seen that the two show a strong positive correlation, indicating that when other conditions are basically the same, the presence of ash deposition will have a greater impact on the output power of the module.

Secondly, Support Vector Machine (SVM) is selected as the model of power prediction. At the same time, in order to improve the accuracy of the model, Bald Eagle Search (BES) is proposed to optimize the parameters of SVM. Aiming at the limitations of fixed parameters and the decline of population diversity in the later iteration of the original BES, Tent chaotic mapping, adaptive change parameters and Gaussian mutation are introduced to improve the BES algorithm. In order to verify the performance of the improved algorithm (IBES), six functions are used to test the algorithm, and compared with particle swarm optimization (PSO), grey wolf algorithm (GWO) and original vulture algorithm. The test results show that the IBES algorithm is higher than the other three algorithms in iteration speed and optimization accuracy, which verifies the superiority of the improved algorithm.

Finally, the power prediction of the ash photovoltaic module is carried out. In order to improve the accuracy of prediction, the collected data are divided into sunny and cloudy days for prediction by grey correlation analysis. The prediction results show that the accuracy of the prediction results after introducing the ash density is better than that of the conventional prediction results in sunny weather, and the IBES-SVM has the best effect. The average absolute error, root mean square error and absolute mean percentage error are 0.7682,1.2318 and 0.1515, respectively. In the cloudy weather prediction results, the prediction error of PSO-SVM increases after the introduction of ash density, and the other three algorithm optimization models all achieve different degrees of prediction accuracy increase after the introduction of ash density. Among them, IBES-SVM still has the best effect, and the three error evaluation indexes are 0.8516, 1.0735 and 0.2861 respectively. The prediction results show that it is feasible to consider the ash deposition factor in photovoltaic power prediction, and IBES-SVM has a good prediction effect.

[1] 魏景东,赵增海,郭雁珩, 等. 2021年中国光伏发电发展现状与展望[J]. 水力发电,2022,48(10):4-8.

[2] 国务院. 国务院关于促进光伏产业健康发展的若干意见：发〔2013〕24号［A/OL］．（2013-7-4）.

[3] Masa-B D, Castillo-C M, Matallanas E, et al. Improving photovoltaics grid integration through short time forecasting and self-consumption[J]. Applied Energy, 2014, 125: 103-113.

[4] Pan M, Li C, Gao R, et al. Photovoltaic power forecasting based on a support vector machine with improved ant colony optimization[J]. Journal of Cleaner Production, 2020, 277: 123-138.

[5] Kim S K , Jeon J H , Cho C H , et al. Modeling and simulation of a grid-connected PV generation system for electromagnetic transient analysis[J]. Solar Energy, 2009, 83(5):664-678.

[6] Almeida M P , Perpinan O , Narvarte L . PV power forecast using a nonparametric PV model[J]. Solar Energy, 2015, 115:354-368.

[7] Haixiang, Zang, Lilin, et al. Hybrid method for short-term photovoltaic power forecasting based on deep convolutional neural network[J]. IET Generation, Transmission, Distribution, 2018, 12(20):4557-4567.

[8] Li L L , Wen S Y , Tseng M L , et al. Renewable energy prediction: A novel short-term prediction model of photovoltaic output power[J]. Journal of cleaner production, 2019(Aug.10):228.

[9] Mellit A, Pavan A M, Lughi V. Deep learning neural networks for short-term photovoltaic power forecasting[J]. Renewable Energy, 2021, 172: 276-288.

[10] 时珉, 许可, 王珏, 等. 基于灰色关联分析和 GeoMAN 模型的光伏发电功率短期预测[J]. 电工技术学报, 2021, 36(11): 2298-2305.

[11] 刘卫亮, 刘长良, 林永君, 等. 计及雾霾影响因素的光伏发电超短期功率预测[J]. 中国电机工程学报, 2018, 38(14): 4086-4095.

[12] 王福忠, 王帅峰, 张丽. 基于VMD-LSTM与误差补偿的光伏发电超短期功率预测[J]. 太阳能学报,2022,43(08):96-103.

[13] Yan J, Hu L, Zhen Z, et al. Frequency-Domain decomposition and deep learning based solar PV power ultra-short-term forecasting model[J]. IEEE Transactions on Industry Applications, 2021, 57(4): 3282-3295.

[14] 赫卫国,郝向军,郭雅娟, 等.基于ARIMA和SVR的光伏电站超短期功率预测[J].广东电力,2017,30(08):32-37.

[15] Han Y , Wang N , Ma M , et al. A PV power interval forecasting based on seasonal model and nonparametric estimation algorithm[J]. Solar Energy, 2019, 184(MAY):515-526.

[16] 常东峰, 南新元.基于改进麻雀算法的深度信念网络短期光伏功率预测[J].现代电子技术,2022,45(17):135-140.

[17] 钱子伟,孙毅超,王琦, 等.基于OS-ELM的光伏发电中长期功率预测[J].南京师范大学学报(工程技术版),2020,20(01):8-14.

[18] Han S, Qiao Y, Yan J, et al. Mid-to-long term wind and photovoltaic power generation prediction based on copula function and long short term memory network[J]. Applied energy, 2019, 239: 181-191.

[19] 翟载腾.任意条件下光伏阵列的输出性能预测[D]. 中国科学技术大学, 2008.

[20] Dolara A, Leva S, Manzolini G. Comparison of different physical models for PV power output prediction[J]. Solar energy, 2015, 119: 83-99.

[21] Ma T, Yang H, Lu L. Solar photovoltaic system modeling and performance prediction[J]. Renewable and Sustainable Energy Reviews, 2014, 36: 304-315.

[22] 杨德州, 尹立夫, 王洲, 等. 基于改进广义神经网络的光伏阵列短期功率预测[J]. 电子器件,2022,45(03):722-726.

[23] 汤航, 李萍, 王宽, 等.基于改进蜂群算法优化BP神经网络的短期光伏功率预测[J].电气应用,2022,41(05):18-22.

[24] 仇琦, 杨兰, 丁旭, 等.基于改进EMD-ARIMA的光伏发电系统短期功率预测[J].电力科学与工程,2020,36(08):42-50.

[25] 郭婕, 厉虹, 王丽婕.基于样本双重选择的马尔科夫链光伏功率预测[J].北京信息科技大学学报(自然科学版),2019,34(06):14-18.

[26] 葛浩然, 夏宇, 邹文进, 等.基于RF-XGBoost的光伏发电功率预测[J].电气自动化,2022,44(05):12-15.

[27] 王粟, 隗磊锋, 曾亮.基于GWO-SVM与随机森林的组合光伏功率预测模型[J].昆明理工大学学报(自然科学版),2021,46(05):82-88.

[28] Semaoui S, Arab A H, Boudjelthia E K, et al. Dust effect on optical transmittance of photovoltaic module glazing in a desert region[C]// Elsevier Ltd. Elsevier Ltd, 2015:1347 – 1357.

[29] Said S, Walwil H M. Fundamental studies on dust fouling effects on PV module performance[J]. Solar Energy, 2014, 107(sep.):328-337.

[30] Vivar M, Herrero R, Anton I, et al. Effect of soiling in CPV systems[J]. Solar Energy, 2010, 84( 7):1327-1335.

[31] Al-Hasan A Y, Ghoneim A A . A new correlation between photovoltaic panel's efficiency and amount of sand dust accumulated on their surface[J]. International Journal of Solar Energy, 2005, 24(4):187-197.

[32] Rao A, Pillai R, Mani M, et al. Influence of Dust Deposition on Photovoltaic Panel Performance[J]. Energy Procedia, 2014, 54.

[33] Guo B, Javed W, Figgis B W, et al. Effect of dust and weather conditions on photovoltaic performance in Doha, Qatar[C]// 2015 First Workshop on Smart Grid and Renewable Energy (SGRE). IEEE, 2015.

[34] 林志鸿, 曾飞, 王哲, 等. 灰尘对光伏电站发电性能的影响[J]. 科技创新与应用, 2022, 12(12): 90-93.

[35] 陈东兵, 李达新, 时剑等. 光伏组件表面积尘及立杆阴影对电站发电功率影响的测试分析[J]. 太阳能, 2011(9):39-41.

[36] Hassan A H, Rohoma U A, Elminir H K, et al. Effect of Airborne dust concentration on the performance of PV modules[J].Journal of Astronomical Society Egypt, 2005,13(1):24-38．

[37] Khatib T, Kazem H A, Sopian K, et al. Effect of dust deposition on the performance of multi-crystalline photovoltaic modules based on experimentalmeasurements[J]. International Journal of Renewable Energy Research, 2013,3(01): 850-853.

[38] Li X, Niu K . Effectively predict the solar radiation transmittance of dusty photovoltaic panels through Lambert-Beer law[J]. Renewable Energy, 2018, 123(AUG.):634-638.

[39] 居发礼. 积灰对光伏发电工程的影响研究[D]. 重庆：重庆大学，2010.

[40] 张羽翔, 魏霞.光伏发电厂清灰模式仿真研究[J].计算机仿真，2017, 34(04): 157-161.

[41] Mekhilef. S, Saidur. R, Kamalisarvestani. M . Effect of dust, humidity and air velocity on efficiency of photovoltaic cells[J]. Renewable and Sustainable Energy Reviews, 2012, 16(5):2920-2925..

[42] 李智,白悄,宗谨, 等.光伏组件数学模型研究与分析[J]. 华北电力技术,2013, 12(06):1-4.

[43] 豆岁阳. 新型纳米结构太阳能电池的制备及性能表征[D]. 北京:北京交通大学，2014.

[44] 高宏玲, 赵翼. 并网光伏发电系统设计中影响系统发电效率的因素分析[J].工业技术创新, 2014, 14(02): 148-154.

[45] 魏晨晨, 王超, 李想, 等. 光伏组件发电效率影响因素分析[J]. 建设科技, 2023, No.466(01):79-81.

[46] 姜栋潇. 基于数据挖掘的光伏阵列故障诊断研究[D]. 北京: 华北电力大学, 2016.

[47] 田冰涛, 刘耀昌, 马明乐, 等. 灰尘对太阳能电池组件特性参数的影响[J]. 物理通报, 2013, 11(05): 115-118.

[48] 马小龙. 光伏面板积尘特性及高效除尘方法研究[D]. 杭州: 浙江工业大学,2015.

[49] 乔永力. 积尘对太阳能光伏板性能影响及清洁装置的研究[D].中国计量大学,2018.

[50] Ramli M A M, Prasetyono E, Wicaksana R W, et al. On the investigation of photovoltaic output power reduction due to dust accumulation and weather conditions[J]. Renewable Energy, 2016, 99: 836-844.

[51] 吕玉坤, 杨宇星, 赵伟萍. 低风速环境下光伏组件积灰特性模拟研究[J]. 电源技术,2021,45(06):801-804+827.

[52] 李玉娥,高万泉. 保定市日照气候变化特征及成因分析[A]. 中国气象学会.第27届中国气象学会年会城市气象，让生活更美好分会场论文集[C].中国气象学会:中国气象学会,2010:373-380.

[53] Quan Y Y, Zhang L-Z, Qi R-H, et al. Self-cleaning of Surfaces: the Role of Surface Wettability and Dust Types[J]. Scientific Reports, 2011, 6(1): 38239.

[54] 白田田. 多能源互补的分布式冷热电联供系统的优化运行研究[D]. 北京:华北电力大学(北京), 2016.

[55] Jaszczur W. An investigation of the dust accumulation on photovoltaic panels[J]. Mathematical research letters: MRL, 2020, 27(2).

[56] Paudyal B R, Shakya S R . Dust accumulation effects on efficiency of solar PV modules for off grid purpose: A case study of Kathmandu[J]. Solar Energy, 2016, 135(2016):103-110.

[57] 宁会峰, 程荣展, 王伟志, 等. 积灰对光伏发电的影响及除尘效果实验研究[J]. 太阳能学报,2020,41(11):120-125.

[58] Vapnik V N , Chervonenkis A . A note on one class of perceptrons[J]. Automation and Remote Control, 1964, 25(1).

[59] 邓乃扬, 田英杰. 支持向量机：理论、算法与拓展[M]. 北京: 科学出版社, 2009.

[60] Kennedy J, Eberhart R. Particle Swarm Optimization[C]// Icnn95-international Conference on Neural Networks. IEEE, 1995.

[61] Mirjalili S, Mirjalili S M, Lewis A . Grey Wolf Optimizer[J]. Advances in Engineering Software, 2014, 69(3):46–61.

[62] Alsattar H A, Zaidan A A, Zaidan B B. Novel meta-heuristic bald eagle search optimisation algorithm[J]. Artificial Intelligence Review, 2020, 53(6).

[63] 张燕, 陈慧丹, 周颖, 等. 基于改进GWO-LSSVM磨矿粒度软测量模型[J]. 计算机仿真, 2020, 37(06): 298-304.

[64] 贾鹤鸣, 姜子超, 彭晓旭, 等．基于改进鬣狗优化算法的多阈值彩色图像分割[J]．计算机应用与软件，2020,37(05)：261-267．

[65] Iacomi M, Cascio D, Fauci F, et al. Mammographic imagessegmentation based on chaotic map clustering algorithm[J]. BMC Medical Imaging，2014，14( 1) : 12

[66] 王涛，Chellali R. 非线性权重和收敛因子的鲸鱼算法[J]. 微电子学与计算机，2019，36( 1) : 11 － 15.

[67] 魏昕, 冯锋. 基于高斯-柯西变异的帝国竞争算法优化[J].计算机科学, 2021, 48(S2): 142-146.

[68] 李兴莘, 张靖, 何宇, 等.基于改进粒子群算法的微电网多目标优化调度[J]. 电力科学与工程, 2021, 37(03): 1-7.

[69] 王继东, 宋智林, 冉冉. 基于改进支持向量机算法的光伏发电短期功率滚动预测[J]. 电力统及其自动化学报, 2016, 28(11): 9-13.

[70] 何锋, 章义军, 章建华, 等. 基于相似日和分位数回归森林的光伏发电功率概率密度预测[J].热力发电, 2019,(7):64-69.

[71] 殷豪, 陈云龙, 孟安波, 等. 基于二次自适应支持向量机的光伏输出功率预测[J]. 太阳能学报, 2019,40(7):1866-1873.

[72] 卢艳云. 基于支持向量机和人工蜂群的光伏发电功率预测研究[D].合肥工业大学, 2020.