Research On Coal Gangue Sorting Method Based On Delta Parallel Robot

Coal gangue is produced in the process of coal production and coal washing.When sorting coal gangue,my country mostly adopts traditional visual marking and manual sorting method to sort coal gangue,due to the late industrial development.This method not only has the disadvantages of low identification efficiency of coal gangue,complex process,and low generalization ability,but also the toxic gas released by the exposed coal gangue will cause great harm to the health of sorting workers.In view of the above problems,this paper proposes an improved SSD target detection algorithm based on the SSD target detection algorithm,which is used for coal gangue sorting,in order to realize the automatic sorting of coal gangue.The structural characteristics of the Delta parallel robot are studied,the degrees of freedom and kinematics are solved,and the feasibility of the Delta parallel robot sorting coal gangue is analyzed and verified.Aiming at the target positioning method in the coal gangue target detection task,Zhang Zhengyou calibration method and Tais algorithm were used to calibrate the camera and hand-eye of the Delta robot respectively,and the conversion relationship between the manipulator and the camera of the Delta parallel robot was solved.The calibration task of the Delta parallel robot is presented,and the accuracy of the calibration is verified by experiments.For the target recognition of coal gangue,after the comparison and screening of the current advanced target detection networks,the resnet network is used to replace the VGG16 network,and the residual structure is introduced into the shallow network in the SSD target detection network.Gangue target detection belongs to small and intensive target detection.Combining FPN network in SSD network,an improved SSD target detection algorithm suitable for 50-150 mm coal gangue target detection scale is proposed,and the optimization is verified in experiments.Compared with the previous network,the m AP value of the latter network has been greatly improved,and the missed detection rate and false detection rate are also significantly reduced.The experimental platform is built by simulating the actual environment,and the experiment of coal gangue target detection task is carried out on the experimental platform.The experimental results show that the optimized SSD target detection network can accurately identify,locate and classify coal gangue.The correctness of this method is verified.