基于平均灰度值与随机森林算法的光伏组件积灰程度预测
PREDICTION OF ASH ACCUMULATION DEGREE OF PHOTOVOLTAIC MODULES BASED ON AVERAGE GRAYSCALE VALUE AND RANDOM FOREST ALGORITHM
陈佳豪 1杨建蒙 1翟永杰 2李斌 1曾侨飞1
作者信息
- 1. 华北电力大学能源动力与机械工程学院,保定 071003
- 2. 华北电力大学控制与计算机工程学院,保定 071003
- 折叠
摘要
通过对光伏组件可见光图像的灰度直方图进行数值化处理,引入平均灰度值的概念,结合学校搭建的光伏实验台,证实平均灰度值与光伏组件积灰密度之间存在对应关系.基于此结论,对内蒙某光伏电站视觉检测平台的图像进行识别与处理,将其与光伏发电数据相对应,发现随着时间的推移,清洁侧与积灰侧平均灰度值差值的变化与发电损失正相关.根据该实验台收集的气象及图像数据,通过随机森林算法建立积灰程度预测模型,在相关气象因素的基础上,结合前一天的积灰板平均灰度值共同作为输入变量,对积灰板平均灰度值进行逐天预测,并通过设置不同的调优参数提高预测算法寻优鲁棒性.
Abstract
This article numerically processes the grayscale histogram of visible light images of photovoltaic modules,introduces the concept of average grayscale value,and combines it with the photovoltaic experimental platform built by the school to confirm the corresponding relationship between the average grayscale value and the ash density of photovoltaic modules.Based on this conclusion,the image of a visual inspection platform for a photovoltaic power station in Inner Mongolia is identified and processed,and it is found that over time,the change in the average gray level difference between the clean side and the accumulated side is positively correlated with power generation loss.Based on the meteorological and image data collected by the experimental platform,a prediction model for the degree of ash accumulation is established using the random forest algorithm.Based on relevant meteorological factors,the average gray level of the ash accumulation board from the previous day is combined as the input variable to predict the average gray level of the ash accumulation board day by day.Different tuning parameters are set to improve the optimization robustness of the prediction algorithm.
关键词
光伏组件/积灰/随机森林/图像识别/平均灰度值Key words
photovoltaic modules/ash deposition/random forest/image recognition/average grayscale value引用本文复制引用
出版年
2024
NETL
NSTL
万方数据