The Research On Some Key Technical Problems Of Heavy Load Robot CMOR For Maintenance In Fusion Reactor Vacuum

CFETR(China Fusion Engineering Test Reactor)is a major scientific project inde-pendently designed and developed by China and jointly cooperated with other countries.The inner components of fusion reactors should be maintained because it would be ru-ined by the tremendous and complicated thermal load,electromagnetic load and neutron radiation,which influences the normal function.Because of the high radiation of main-tenance surroundings,maintaining tokamak fusion cores needs a remote handling robot.Currently,the tokamak of the world mainly adopts a remote manual control strategy,which is low efficient and flexible.This mode cannot demand the future meets of in-telligent maintenance of fusion devices.Therefore,the Remote Handling team at the Institute of Plasma Physics Chinese Academy of Sciences has developed a new genera-tion of intelligent maintenance system,CMOR(CFETR Multipurpose Overload Robot)for CFETR,which is used for maintenance of inner components of the fusion device,such as maintaining the first wall of divertor,monitoring the inner components,clean-ing the tritium dust in a vacuum and so on.Due to the huge maintenance space,the complexity of the mechanism and the delicate maintenance tasks of the fusion reac-tor,the CMOR was designed as a master/slave multi-DOF arm.The master robot,a transporting robot,is responsible for locating and transporting the slave robot,which is redundant,flexible and heavy.The slave robot,a dual arm robot,is responsible for compensating the locating error and maintaining delicately,which is a rigid robot with high accuracy.This dissertation focuses on some key problems of kinematics.The main content is as follow:Develop a collision free kinematic algorithms for the multi-DOF robot in compli-cated surroundings of CFETR’s vacuum.Through optimizing ant colony optimization with dynamic accuracy by the evolution theory,kinematics of the multi-DOF robot based on dynamic ant colony optimization is studied,which solves the multi-DOF robot’s nonlinear problems.As for probable collision between the master robot and CFETR’s vacuum,a collision detection algorithm based on the gray value of section plane figures is designed,which ensuring the safety of the robot.Study the kinematic parameters calibration of dual arm robot at the end effector.The accuracy of the dual arm robot would influence the quality of robot delicate mainte-nance.By designing a specific trajectory of the robot and fitting its curves,a kinematic calibration algorithm is designed by geometric features.The high accuracy and unique-ness of the algorithm are guaranteed by using the least square method and the 3σ rule in fitting features.Design the collision perceptron of the dual arm robot.As for probable collision between the dual arm robot and fusion device,a hybrid collision perceptron based on primary component analysis is studied,which ensures the safety of the robot motion.This algorithm avoids solving the robot dynamic model and uses primary component analysis to achieve the data’s dimension reduction such as position,velocity and accel-eration/current information.The algorithm with the help of a perceptron,a supervised machine learning,detects collision efficiently.