Mining electric shovel bucket health monitoring method based on deep learning and machine vision

• Main Authors: WANG ZHIQIANG, XIAO HAIBO, YAN DONG, XUE YINBO, LI XIAOLIANG, XIA TIEFENG, ZHAI LEI, YAO JIANG, WANG KAIFU
• Format: Patent
• Language: Chinese; English
• Published: 19.08.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; DREDGING; FIXED CONSTRUCTIONS; FOUNDATIONS; HANDLING RECORD CARRIERS; HYDRAULIC ENGINEERING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS; SOIL SHIFTING

The invention belongs to the technical field of strip mine electric shovel bucket health monitoring, and particularly relates to a mining electric shovel bucket health monitoring method based on deep learning and machine vision. Comprising the following steps: 1, image acquisition: acquiring a working video of a bucket from a camera, and storing visible images of all parts of the bucket as an image file; 2, preparing an image, manually marking each part to obtain an XML file, and dividing the XML file into a training set and a verification set; 3, training preprocessing is carried out, the XML file is converted into a TFRecord format file, and finally a PBTXT file is made; 4, training is implemented, a pipeline configuration file is set, three results are obtained through training, and a frozen reasoning graph file is generated; 5, bucket health monitoring: collecting a bucket working video, inputting the bucket working video into a LabVIEW development environment, and carrying out real-time bucket health mon