基于混合神经网络的光伏组件输出特性数据驱动建模方法

doi:10.11959/j.issn.2096-6652.202019

摘要/Abstract

摘要：

针对传统物理机理建模方法不适用于复杂光照条件下光伏组件建模的问题，提出一种基于混合结构神经网络的光伏组件输出特性数据驱动建模方法。在深入分析光伏组件物理机理及输出特性的基础上，提出利用卷积神经网络和径向基函数神经网络对不均匀光照条件、温度、湿度等环境因素进行特征提取，并对光伏组件的输出特性进行仿真拟合。为提高模型的拟合效果，提出针对不均匀光照条件的阴影形态等效分析方法，同时采用改进型的遗传编码方案对网络参数进行优化，最后利用实际运行数据对模型效果进行分析验证。结果表明，该模型对不均匀光照条件具有一定的泛化跟踪能力，同时仿真结果平均误差保持在7%以内。

关键词: 数据驱动建模, 卷积神经网络, 径向基函数神经网络, 遗传算法优化

Abstract:

A data-driven modeling approach of photovoltaic modules based on hybrid structure neural network was proposed to solve the issue that traditional physical mechanism modeling methods are difficult to be applied under some complex lighting conditions.Based on an in-depth analysis of the physical mechanism and output characteristics of photovoltaic modules,the convolutional neural network and radial basis function network were used to extract features of several environmental factors such as uneven lighting conditions,temperature and humidity.In order to improve the fitting effect of the hybrid network model,an equivalent analysis method of shadow morphology for uneven lighting conditions was proposed,a genetic coding scheme was used to optimize network parameters.Model effect was verified by the actual operation data.The model had generalized tracking capacity under various uneven lighting conditions,and the average error of simulation results was kept within 7%.

Key words: data-driven modeling, convolutional neural network, radial basis function neural network

中图分类号:

TP183

张国宾,王新迎. 基于混合神经网络的光伏组件输出特性数据驱动建模方法[J]. 智能科学与技术学报, 2020, 2(2): 169-178.

Guobin ZHANG,Xinying WANG. Research on data-driven modeling for photovoltaic characteristics based on hybrid neural network[J]. Chinese Journal of Intelligent Science and Technology, 2020, 2(2): 169-178.

图/表 18

图1

表1

表2

图2

表3

图3

图4

图5

图6

图7

图8

图9

图10

图11

图12

表4

图13

图14

参考文献 31

[1]	陈思远, 白昱阳, 张俊 ,等. 基于区块链和通证经济的跨国跨洲电力市场机制设计[J]. 智能科学与技术学报, 2019,1(1): 96-105.
	CHEN S Y , BAI Y Y , ZHANG J ,et al. Design of transnational and intercontinental electricity market with blockchain and token economics[J]. Chinese Journal of Intelligent Science and Technology, 2019,1(1): 96-105.
[2]	郭创新, 王惠如, 张伊宁 ,等. 面向区域能源互联网的“源-网-荷”协同规划综述[J]. 电网技术, 2019,43(9): 3071-3080.
	GUO C X , WANG H R , ZHANG Y N ,et al. Review of“source-grid-load” co-planning orienting to regional energy Internet[J]. Power System Technology, 2019,43(9): 3071-3080.
[3]	郑玉平, 王丹, 万灿 ,等. 面向新型城镇的能源互联网关键技术及应用[J]. 电力系统自动化, 2019,43(14): 2-15.
	ZHENG Y P , WANG D , WAN C ,et al. Key technologies and applications of energy Internet[J]. Automation of Electric Power Systems, 2019,43(14): 2-15.
[4]	MIRANGEIGI M , IMANEINI H . Hybrid modulation technique for grid-connected cascaded photovoltaic systems[J]. IEEE Transactions on Industrial Electronics, 2016,63(12): 7843-7853.
[5]	侯国莲, 弓林娟, 苏烨 ,等. 基于 ACP 的平行发电控制系统[J]. 智能科学与技术学报, 2019,1(3): 269-279.
	HOU G L , GONG L J , SU Y ,et al. ACP based parallel power generation control system[J]. Chinese Journal of Intelligent Science and Technology, 2019,1(3): 269-279.
[6]	孙秋野, 胡旌伟, 张化光 . 能源互联网中自能源的建模与应用[J]. 中国科学:信息科学, 2018,48(10): 1409-1429.
	SUN Q Y , HU J W , ZHANG H G . Modeling and application of we-energy in energy Internet[J]. Scientia Sinica (Informationis), 2018,48(10): 1409-1429.
[7]	周林, 郭珂, 刘强 ,等. 光伏电池工程用数学模型研究[J]. 电工技术学报, 2011,26(10): 211-216.
	ZHOU L , GUO K , LIU Q ,et al. Research on engineering analytical model of solar cells[J]. Transactions of China Electrotechnical Society, 2011,26(10): 211-216.
[8]	师楠, 周苏荃, 李一丹 ,等. 基于 Bezier 函数的光伏电池建模[J]. 电网技术, 2015,39(8): 2195-2200.
	SHI N , ZHOU S Q , LI Y D ,et al. PV cell modeling based on Bezier function[J]. Power System Technology, 2015,39(8): 2195-2200.
[9]	徐岩, 高兆, 朱晓荣 . 基于混合蛙跳算法的光伏阵列参数辨识方法[J]. 太阳能学报, 2019,40(7): 1903-1911.
	XU Y , GAO Z , ZHU X R . Parameter identification method of phootovoltaic array based on shuffled frog leaping algorithm[J]. Acta Enerciae Solaris Sinica, 2019,40(7): 1903-1911.
[10]	CHATTERJEE A , KEYHANI A , KAPOOR D . Identification of photovoltaic source models[J]. IEEE Transactions on Energy Conversion, 2011,26(3): 883-889.
[11]	JING J S , LOW K S . Photovoltaic model identification using particle swarm optimization with inverse barrier constraint[J]. IEEE Transactions on Power Electronics, 2012,27(9): 3975-3983.
[12]	程泽, 董梦男, 杨添剀 ,等. 基于自适应混沌粒子群算法的光伏电池模型参数辨识[J]. 电工技术学报, 2014,29(9): 245-252.
	CHENG Z , DONG M N , YANG T K ,et al. Extraction of solar cell based on self-adaptive chaos particle swarm optimization algorithm[J]. Transactions of China Electrotechnical Society, 2014,29(9): 245-252.
[13]	沈赋, 尹斌, 孙维广 . 基于单神经元自适应 PID 的光伏发电MPPT[J]. 电力系统及其自动化学报, 2017,29(2): 89-95.
	SHEN F , YIN B , SUN W G . Single-neuron adaptive PID control in MPPT of photovoltaic generation[J]. Proceedings of the CSUEPSA, 2017,29(2): 89-95.
[14]	胡桂廷, 仲程超, 张伟君 ,等. 基于模型辨识的BP神经网络在光伏系统MPPT中的应用[J]. 计算机测量与控制, 2017,25(10): 213-216.
	HU G T , ZHONG C C , ZHANG W J ,et al. Application of BP neural network based on model identification in photovoltaic system MPPT[J]. Computer Measurement ＆ Control, 2017,25(10): 213-216.
[15]	杨东海, 刘洋, 王毅 ,等. 基于二进制蚁群模糊神经网络的光伏系统MPPT控制算法研究[J]. 电气工程学报, 2017,12(6): 41-46.
	YANG D H , LIU Y , WANG Y ,et al. Research on MPPT control algorithm of photovoltaic system by binary ant colony algorithm and fuzzy neural network[J]. Journal of Electrical Engineering, 2017,12(6): 41-46.
[16]	张晓强, 刘宜罡, 邹应全 ,等. 基于自适应神经网络控制的光伏MPPT算法改进[J]. 太阳能学报, 2019,40(11): 3091-3102.
	ZHANG X Q , LIU Y G , ZOU Y Q ,et al. An enchanced photovoltaic MPPT approach based on adaptive neural network control[J]. Journal of Electrical Engineering, 2017,40(11): 3091-3102.
[17]	WANG H Z , YI H Y , PENG J C ,et al. Deterministic and probabilistic forecasting of photovoltaic power based on deep convolutional neural network[J]. Energy Conversion and Management, 2017(153): 409-422.
[18]	赵康宁, 蒲天骄, 王新迎 ,等. 基于改进贝叶斯神经网络的光伏出力概率预测[J]. 电网技术, 2019,43(12): 4377-4386.
	ZHAO K N , PU T J , WANG X Y ,et al. Probabilistic forecasting for photovoltaic power based on improved bayesian neural network[J]. Power System Technology, 2019,43(12): 4377-4386.
[19]	CHOU K Y , YANG S T , CHEN A Y . Maximum power point tracking of photovoltaic system based on reinforcement learning[J]. Sensors (Basel), 2019,19(22):5054.
[20]	ZHANG X S , LI S N , HE T Y ,et al. Memetic reinforcement learning based maximum power point[J]. Energy, 2019,174(C): 1079-1090.
[21]	杨元培, 杨奕, 王建山 ,等. 光伏发电传输最大功率储能优化建模仿真[J]. 计算机仿真, 2017,34(9): 103-108.
	YANG Y P , YANG Y , WANG J S ,et al. Model and simulation of the in photovoltaic power energy storage optimization power generation transmission[J]. Computer Simulation, 2017,34(9): 103-108.
[22]	陈如亮, 崔岩, 李大勇 ,等. 光照不均匀情况下光伏组件仿真模型的研究[J]. 系统仿真学报, 2008,20(7): 1681-1690.
	CHEN R L , CUI Y , LI D Y ,et al. Study on simulation model of PV module under non-uniform insolation[J]. Journal of System Simulation, 2008,20(7): 1681-1690.
[23]	吴登盛, 王立地, 刘通 ,等. 基于神经网络的光伏阵列多峰 MPPT的研究[J]. 电测与仪表, 2019,56(7): 69-74.
	WU D S , WANG L D , LIU T ,et al. Research on multi-peak MPPT of PV array based on neural network[J]. Electrical Measurement ＆ Instrumentation, 2019,56(7): 69-74.
[24]	王云平, 阮新波, 李颖 . 不均匀光照光伏单元串联电路快速 MPPT方法[J]. 中国电机工程学报, 2015,35(19): 4870-4878.
	WANG Y P , RUAN X B , LI Y . A rapid hacking method of maximum power point for solar units in series under uneven solar irradiance[J]. Proceedings of the CSEE, 2015,35(19): 4870-4878.
[25]	乐全明, 赵健, 郭力 . 不均匀光照条件下太阳能电池串联电路特性分析及GMPPT控制[J]. 光子学报, 2017,46(6): 214-224.
	YUE Q M , ZHAO J , GUO L . Analysis of dynamic characteristic for solar arrays in series and GMPPT control[J]. Acta Photonica Sinica, 2017,46(6): 214-224.
[26]	REHMAN S , TU S S , WAQAS M ,et al. Unsupervised pre-trained filter learning approach for efficient convolution neural network[J]. Neurocomputing, 2019(365): 171-190.
[27]	ZHANG J M , WU Y . Complex-valued unsupervised convolutional neural networks for sleep stage classification[J]. Computer Methods and Programs in Biomedicine, 2018(164): 181-191.
[28]	AMORIMA W P , TETILAB E C , PISTORIB H ,et al. Semi-supervised learning with convolutional neural networks for UAV images automatic recognition[J]. Computers and Electronics in Agriculture, 2019,(164):104932.
[29]	AHN E , KUMAR A , FULHAM M ,et al. Convolutional sparse kernel network for unsupervised medical image analysis[J]. Medical Image Analysis, 2019(56): 140-151.
[30]	文常保, 马文博, 刘鹏里 . 基于改进遗传算法的RBF神经网络结构优化研究[J]. 计算机工程与科学, 2019,43(5): 917-923.
	WEN C B , MA W B , LIU P L . Research on structure optimization of RBF neural network based on improved genetic algorithm[J]. Computer Engineering ＆ Science, 2019,43(5): 917-923.
[31]	JIA W K , ZHAO D , DING L . An optimized RBF neural network algorithm based on partial least squares and genetic algorithm for classification of small sample[J]. Applied Soft Computing, 2016(48): 373-184.

类型	材料	优势	劣势
无机材料	单晶硅	原料易获得，易加工，寿命长	转化效率依赖于制作技术
	铜铟硒	转化效率在20%以上，优于单晶硅	制作技术存在门槛
	砷化镓	转化效率可达50%，极端环境下性能稳定	造价昂贵
	碲化镉	成本效益高	转化效率低，在20%以下
有机材料	单质结、异质结、染料敏化材料	易合成，成本低廉	转化效率一般低于10%

参数	含义	单位	标准值
R_s	等效串联电阻	?	5×10^-5
R _sh	等效并联电阻	?	5×10⁵
I_{D 0_st}	标准运行条件下的二级反向饱和电流	A	19.969 3×10^-6
T_r	参考温度	K	301.08
k_i	温度系数	A/K	0.001 7
I_{sc _s}	标准运行条件下的等效短路电流	A	2.3
E_g	半导体材料禁带宽度	eV	1.1
A	品质因数	/	1.72
K	玻尔兹常数	J/K	1.38×10^-23
G_r	标准光照强度	W/m²	1 000

运行场景	光伏板遮挡位置	环境温度/℃	光照强度/W/m²
情景1	（1,1）、（1,2）、（2,3）	20	1 000
	（1,3）、（2,1）、（2,2）、（3,1）、（3,2）、（3,3）	15	500
情景2	（1,2）、（2,1）、（2,3）、（3,2）	20	800
	（1,1）、（1,3）、（2,2）	15	500
	（3,1）、（3,3）	15	300

阴影类型编号	遮挡位置
1	（1,2）、（2,2）
2	（1,2）、（1,3）、（1,4）
3	（1,1）、（1,2）、（2,2）、（2,4）