输送带钢芯缺陷检测研究

发布时间：2018-09-12 16:49

【摘要】：钢绳芯输送带内嵌钢丝绳芯,外覆胶质带体,广泛地应用于煤矿、港口、码头等领域的物料运输。但由于输送带的应用环境大都较恶劣,加之长期负重运行,输送带内的钢丝绳芯极易发生接头抽动、绳芯断裂等损伤,给工业生产带来严重的事故隐患。因此,对输送带的钢芯状况进行实时检测具有重大意义。目前,常用的输送带实时检测方法包括基于电磁感应原理的漏磁检测法和基于X射线成像的图像分析法。漏磁检测法只能获取钢芯损伤的大概程度和大致位置,检测结果易受干扰;而基于X射线成像的检测方法可直观获得输送带内部的钢绳状况,且便于使用图像处理方法对其进行分析判别,因此该方法正成为目前的研究方向。本文正是基于输送带的X射线图像,对钢绳的缺陷检测方法进行了研究,主要工作内容有:(1)输送带X射线图像增强处理。受工作环境的影响,直接采集到的输送带X射线图像往往具有较多的干扰噪声,且在成像时输送带两侧由于距离X射线源较远而获得较少的射线能量,使得采集到的两侧图像整体较暗,钢芯和胶质的对比度不明显。针对这种情况,本文通过比较各种去噪方法,选取合适的滤波器对图像进行噪声抑制,同时,根据输送带图像的像素分布特征,提出一种改进的Retinex图像增强算法,实现了图像的对比度增强,便于后续的接头定位和接头抽动检测。(2)输送带钢绳接头抽动识别。针对现有的钢绳芯接头抽动检测方法大多是基于接头点对的匹配,通过匹配点对间的垂直距离和接头标准距离或参考图像的点对垂直距离作比较来检测接头抽动的情况,本文将对这种方法存在的不足进行改进,即实现图像反拉伸后的完全抽动检测。实验表明,该方法可实现对所有接头的检测,使检测结果更加全面、准确。(3)输送带钢绳芯断裂识别。对输送带非接头部位的X射线图像进行研究,考察该部位图像的纹理特性,分析各纹理缺陷检测算法对钢绳断裂检测的效果,并在此基础上提出一种基于纹理规则性的钢绳断裂检测方法,实验表明,该方法能够有效地检测出输送带中断裂的钢芯位置,钢芯断裂检测精度达到了98.9%。
[Abstract]:Steel rope core conveyor belt is embedded with steel rope core and covered with colloidal belt body, which is widely used in coal mine, port, wharf and other fields of material transport. However, due to the harsh application environment of the conveyor belt, coupled with long-term load operation, the steel rope core in the conveyor belt is very easy to occur joint twitching, core fracture and other damage, which brings serious industrial production. At present, the commonly used real-time detection methods of conveyor belts include magnetic flux leakage detection based on electromagnetic induction principle and image analysis based on X-ray imaging. Magnetic flux leakage detection method can only obtain the approximate degree and location of steel core damage, and the detection results. The method based on X-ray imaging is easy to be interfered with, and the wire rope condition inside the conveyor belt can be obtained intuitively, and it is easy to be analyzed and discriminated by image processing method, so this method is becoming the current research direction. The main contents are as follows: (1) X-ray image enhancement processing of conveyor belt. Under the influence of working environment, the conveyor belt X-ray images collected directly often have more interference noise, and in imaging, the conveyor belt has less radiation energy because of the distance from the X-ray source on both sides of the conveyor belt. In view of this situation, by comparing various denoising methods, this paper chooses the appropriate filter to suppress the noise of the image. At the same time, according to the pixel distribution characteristics of the conveyor belt image, an improved Retinex image enhancement algorithm is proposed, which realizes the image contrast enhancement and facilitates the subsequent joint location and connection. Twist detection. (2) Twist identification of steel rope joints in conveyor belts. Most of the existing methods of Twist detection of steel rope core joints are based on the matching of joint point pairs. By comparing the vertical distance between matching points and the standard distance between matching points and the vertical distance between matching points or the point pairs of reference images, this method will exist in this paper. The experimental results show that the method can detect all the joints more comprehensively and accurately. (3) Fracture identification of steel rope core of conveyor belt. X-ray image of non-joint part of conveyor belt is studied, texture characteristics of this part of image are investigated, and each part is analyzed. Based on the effect of texture defect detection algorithm on wire rope fracture detection, a wire rope fracture detection method based on texture regularity is proposed. The experimental results show that the method can effectively detect the position of steel core of conveyor belt fracture, and the detection accuracy of steel core fracture reaches 98.9%.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH222;TP391.41

【参考文献】