摘要
为了提升新能源汽车太阳能电池表面缺陷检测的效果,采用粒子群权重多级加权算法。首先在基本粒子群算法基础上对惯性权重进行3级加权,第1级权值表示为缺陷区域内部像素的分布特性,第2级权值表示为缺陷区域位置特性,第3级权值表示为缺陷区域分布特性;接着对数据集基于欧氏距离通过k-均值聚类进行优化;然后双阈值进行缺陷定位,缺陷特征量提取;最后给出了算法流程。实验仿真显示,算法对新能源汽车太阳能电池表面缺陷检测结果清晰,对各类缺陷的正确识别率比较高。
In order to improve the effect of surface defect detection of solar panels of new energy automobiles,a weighted multi-level particle swarm optimization algorithm was proposed.Firstly,the inertia weights were weighted by three levels based on the basic particle swarm optimization algorithm,the first level weight was expressed distribution characteristics of the internal pixel of the defect region,the second level weight was expressed defect location characteristics,and the tertiary weight was expressed defect area distribution characteristics.Secondly,data set was optimized based on Euclidean distance by k-means clustering.Thirdly,double thresholds were located defects,defect feature was extracted.Finally,the process was given.The experimental simulation shows that the weighted multi-level particle swarm optimization is very clear on the surface defects of new energy automobile solar panels,the correct recognition rate of various types of defect is higher than other algorithms.
作者
刘云潺
毕立恒
LIU Yunchan;BI Liheng(Yellow River Conservancy Technical Institute, Kaifeng 475004, Henan, Chin)
出处
《实验室研究与探索》
CAS
北大核心
2018年第2期62-65,79,共5页
Research and Exploration In Laboratory
基金
中国国家专利(公开号CN202189701U)
关键词
新能源汽车
太阳能电池
表面缺陷
多级加权
阈值
new energy automobile
solar panels
surface defect
multi-level weighted
thresholds