工业控制计算机
工業控製計算機
공업공제계산궤
Industrial Control Computer
2015年
10期
49-51,53
,共4页
改进BP神经网络%刀具磨损%图像处理%参数融合%状态识别
改進BP神經網絡%刀具磨損%圖像處理%參數融閤%狀態識彆
개진BP신경망락%도구마손%도상처리%삼수융합%상태식별
improved BP neural networks%tool wear%image processing%parameter integration%state recognition
运用带动量因子的最速下降BP神经网络实现对刀具磨损状态识别.首先标定相机并采集刀具图像;经过倾斜校正、边缘定位、坐标变换和曲线拟合等技术处理后,提取刀尖后角、刀尖夹角、刀尖半径和刀尖高度4种特征参数并融合构成特征矢量组;最后建立改进BP神经网络模型,对刀具磨损情况进行训练和预测.仿真实验证明该方法具有很好的刀具磨损识别准确度和检测效率.
運用帶動量因子的最速下降BP神經網絡實現對刀具磨損狀態識彆.首先標定相機併採集刀具圖像;經過傾斜校正、邊緣定位、坐標變換和麯線擬閤等技術處理後,提取刀尖後角、刀尖夾角、刀尖半徑和刀尖高度4種特徵參數併融閤構成特徵矢量組;最後建立改進BP神經網絡模型,對刀具磨損情況進行訓練和預測.倣真實驗證明該方法具有很好的刀具磨損識彆準確度和檢測效率.
운용대동량인자적최속하강BP신경망락실현대도구마손상태식별.수선표정상궤병채집도구도상;경과경사교정、변연정위、좌표변환화곡선의합등기술처리후,제취도첨후각、도첨협각、도첨반경화도첨고도4충특정삼수병융합구성특정시량조;최후건립개진BP신경망락모형,대도구마손정황진행훈련화예측.방진실험증명해방법구유흔호적도구마손식별준학도화검측효솔.
The realization of the tool wear state recognition by using BP neural network based on the steepest descent Algorithm with momentum factor.First,camera is calibrated and tool wear images are col ected.After some techniques pro-cesses such as tilt correction,edge location,coordinate transformation and curve fitting,acquirement of 4 kinds of characteris-tic parameters including blade angle,tip angle,nose radius and tip height.A feature vector group is constituted with combina-tion of parameters.Final y the improved BP neural network model is established to train and to predict the tool wear.
