文中采用BP神经网络决策树算法实现钢材表面缺陷非人工检测和分类,依据缺陷的特征提取数据,采用BP神经网络决策树算法构多类造分类器,输入标准样本分类器训练后,对钢材在实际生产过程中可能出现的氧化色、飞边、辊印、擦伤、孔洞、压痕等缺陷进行检测和分类,通过对检测结果的分析,缺陷分类准确率高,速度快。%This article uses the BP neural network decision tree algorithm to realize the steel products＇ surface defect and the classification with non-manual examination,extracts data basing on the flaw characteristic,and uses the BP neural network decision tree a
展开▼
