Due to relying on manual inspection of bucket teeth, generally it can only be carried out when the shovel is shut down or during shift handover, and the fault cannot be found in time; Some electric shovels are also equipped with surveillance cameras. Due to the poor working environment of electric shovels, the judgment of operators will be affected by light, sand, dust and the own factors of on-site personnel. Therefore, some domestic and foreign scholars and relevant enterprises have studied and developed the fault monitoring of shovel bucket teeth, and have formed certain achievements. In 2004, Xiujuan Luo et al. Proposed a scheme of bucket teeth detection. The scheme is to reconstruct the bucket teeth in the working state through the laser rangefinder, and then use the normal bucket teeth model as the template to judge whether the bucket teeth fall off or break through the template matching. In this scheme, the parallel computing method is used to synchronously collect and process the data, making the system more real-time. This method is less sensitive to light and environmental changes, The detection effect is good.
In 2011, he Li et al. Proposed a target detection algorithm combined with image processing technology. Based on the image gray gradient histogram and support vector machine, the algorithm uses the position relationship between bucket teeth to add structural feature constraints to the detection results to improve the accuracy of target detection. At the same time, the author trained a support vector machine based on the bottom shape features of the image, It is equivalent to adding another layer of constraint to the potential bucket tooth target, which improves the detection rate of bucket teeth.
In 2016, Ser Nam Lim et al. Studied a set of bucket tooth falling off visual detection system. The scheme first extracts the image samples of the bucket during the movement of the shovel, and uses these sample information to locate the approximate position of the bucket teeth first. Then the frame difference method and optical flow method are used to correct the results. The method of template matching and tooth line fitting is comprehensively used to accurately locate the bucket tooth target. Finally, combined with the relevant gray features of the bucket teeth image, judge whether the bucket teeth are broken or fall off.
In 2017, Tang Heng of Zhongbei University studied the detection algorithm of bucket teeth falling off of electric shovel, learned from the relatively successful gradient direction histogram feature of pedestrian detection, and combined with the target detection framework of support vector machine to detect the bucket teeth in the infrared image. On this basis, by analyzing the similarity relationship of bucket teeth image in shape, the shape features of bucket teeth are extracted by using the shape context algorithm, The shape matching method is used to add shape feature constraints in the target detection process of bucket teeth, so as to improve the accuracy of target detection. Finally, the whole detection algorithm is verified through experiments, the causes of errors are analyzed, and some improvement ideas and methods are put forward.
In 2018, Duan Yuxiu et al. Proposed a detection method of bucket tooth loss of electric shovel based on machine vision. In this method, the infrared thermal imager is used to collect the bucket image, the template matching method is used to locate the target area of the bucket tooth, and the frame difference method is used to detect the moving target of the bucket tooth. In the difference image, the bucket teeth are segmented and extracted by using the located region and combined with the position of the bucket tooth line region. Through the adaptive threshold, the accurate and rapid detection of missing bucket teeth is realized.
In 2018, Wang Ping of China University of mining and technology analyzed the fault condition of bucket teeth. Based on the preprocessing of infrared thermal imaging images, convolution neural network was used to classify the bucket teeth fault. At the same time, the template matching method combined with SIFT operator was compared to realize the counting and positioning of bucket teeth, and the size of bucket teeth area was used as the judgment basis to judge whether the bucket teeth had fault, The fault detection method of bucket teeth is studied.
To sum up, at present, there are relatively few research results on bucket tooth fault detection of electric shovel bucket. Most of the detection methods are based on traditional methods, but using the detection system of machine vision technology to detect bucket tooth fault has become the main research direction.