基于云层图像的塔式太阳能热发电能量预测方法  

A METHOD OF FORECASTING SOLAR TOWER THERMAL POWER BASED ON CLOUD IMAGES

作　　者：李德文[1] 宓霄凌 黄文君[1] 张鹏[1] 田军 

机构地区：[1]浙江大学智能系统与控制研究所,杭州310027 [2]浙江中控太阳能技术有限公司,杭州310052

出　　处：《太阳能学报》2015年第11期2674-2679,共6页Acta Energiae Solaris Sinica

基　　金：国家自然科学基金(61273100)

摘　　要：提出一种应用于塔式太阳能热发电系统的云层图像识别与运动检测方法,动态预测云层移动对系统热发电能量的影响。该方法利用小波维纳滤波对工业相机实时采集的天空图像进行降噪处理,基于Lab空间边缘检测算法识别天空图像中的云层,结合CSIFT匹配特征点计算得到云层运动速度和方向,并通过云层、太阳、镜场间的相对位置关系实时分析和预测系统热发电能量受云层运动的影响。实验结果表明,该方法的预测实时性和准确性均满足要求,具有较高的工程应用价值。A method of cloud image recognition and motion detection is proposed to dynamically forecast the impact on solar tower power generation caused by cloud movement. It firstly utilizes Wiener filtering in wavelet domain to process sky images collected by industrial camera, and then identifies clouds in these images by edge detection algorithm based on Lab space. After that, it calculates the velocity and direction of moving clouds using CSIFT feature points, and finally analyzes and forecasts the impact on system thermal power generation through the positional relationship of the clouds, the sun and the heliostat field. The experimental results showed that this method meets the requirement of timeliness and accuracy, thus is useful in engineering applications.

关 键 词：云层图像识别 云层运动特性检测 热发电能量预测 小波维纳滤波 Lab空间边缘检测 CSIFT特征 匹配 

分 类 号：TK514[动力工程及工程热物理—热能工程]