基于2D-OTSU图像边缘检测的回转窑工况识别方法

Condition recognition method of rotary kiln based on 2D-OTSU image edge detection

回转窑作为水泥窑炉煅烧过程的核心热工设备,其正常运转率与产品产量、质量及能耗紧密相关,由于回转窑内部核心反应区温度高且装置持续旋转,接触式温度传感器无法安装在窑内核心反应区域,而筒扫系统借助红外扫描装置能够实时监测回转窑筒体表面温度并间接反映窑内热工状况.签于此,提出一种新的基于筒扫图像2D-OTSU边缘检测的回转窑异常工况识别方法.该方法首先构建基于灰度梯度和局部灰度标准差信息的融合模型,并利用相对熵概念计算模型权重系数,进而通过1D-OTSU预搜索策略提升识别算法的效率;然后给出一种2D双阈值检测阈值分割策略,以保证边缘的连续性;最后采用工业现场实际的回转窑筒扫图像对所提方法与其他典型检测方法进行比较研究.对比实验结果表明,所提出方法的检测率和单位误报次数均优于其他算法并具备一定鲁棒性,能够有效检测回转窑内的异常工况,达到提高回转窑运转率的目的.

The rotary kiln is the core thermal reaction equipment of cement calcination process, whose operation state is closely related to the yield, the quality the energy consumption and of products. Contact temperature measurement can not be installed in the core area inside the kiln due to high temperature and continuous rotation. The rotary kiln shell scanning system(RKSSS) is available to monitor the temperature of kiln shell and reflect the thermal condition inside the kiln indirectly in real time. Therefore, a new method of abnormal conditions identification based on 2 D-OTSU edge-detection is proposed. The fusion model based on gray gradient and local gray standard deviation information is firstly constructed,and the weight coefficient of the model is calculated using the concept of relative entropy. The 1 D-OTSU pre-searching strategy is then adopted to improve the efficiency of the algorithm. In addition, a 2 D threshold segmentation strategy is proposed to ensure the continuity of the edge. By applying images from the RKSSS, the proposed method is tested and compared with other typical methods. The results demonstrate that the proposed method can make a promotion in detection rate and false alarm probability with robustness, and is available to detect the abnormal condition and to extend operation cycle of the rotary kiln.