基于图像识别的输煤皮带纠偏方法

The Deviation Correction Method of Coal Conveying Belt Based on Image Recognition

由于现有纠偏方法跑偏现象明显,研究了基于图像识别的输煤皮带纠偏方法。首先,通过计算输煤皮带的边缘坐标信息,对跑偏故障进行划分。其次,运动状态下,计算输煤皮带的受力情况,并根据输煤皮带和滚筒的速度滑差率值进行警告处理,再将软注意力机制用于图像分类,将处理后得到的特征图经卷积操作重建正常样本。再次,对异常样本进行判定,打标签制作异常样本,并对图像进行灰度处理和高斯噪声滤波处理,按照像素进行划分,对目标区域进行分割提取跑偏区域特征。最后,通过连通标记处理目标,对目标进行识别和追踪,从而完成纠偏。实验结果表明,5个小组标签显示0,均未出现跑偏现象,能够较为精准检测皮带的跑偏问题,实现基于图像识别的输煤皮带纠偏方法的良好应用。

Due to the obvious deviation of the existing correction method,the coal belt correction method based on image recognition is studied.Firstly,by calculating the edge coordinate information of the coal conveying belt,the deviation fault is divided.Secondly,in the motion state,the force situation of the coal conveying belt is calculated.And according to the speed slip value of coal belt and drum.The soft attention mechanism is used in image classification,and the feature map obtained after processing are convolved to reconstruct normal samples.Thirdly,abnormal samples were judged,and the abnormal samples were labeled.The gray scale and Gaussian noise filtering were processed,and the target area was segmented to extract the features of the deviation region.Finally,by connecting the target,the target is identified and tracked to complete the correction.The experimental results show that the labels of the five groups show O,none of which has any deviation phenomenon,which can accurately detect the deviation of the belt,and realize a good application of the correction method of coal conveying belt based on image recognition.