Environment Detection And Control Method Of Inner-surface Grinding Robot For Rotary Shells

Grinding robot of inner-surface of rotary shells is widely used.Such as solid rocket engine,it is the most common source of tactical missile.In order to ensure the stable propulsion of a missile during flight,the thermal insulation layer of its engine must be ground.At present,most of the grinding process relies on manual labor.Due to the structure requirements of the engine combustion chamber and requirements of propulsion performance of the engine,inner-surface of the engine shell is mostly unstructured environment.Therefore,automatic grinding of the inner-surface of the thermal insulation layer of engines is necessary.Taking the automatic grinding of the inner-surface of thermal insulation layer of solid rocket engines as research background,a 4-DOF grinding robot system is established as a typical research object.Automatic detection and grinding method of unstructured inner-surface of rotary shells are key researching contents of this paper.Related numerical simulation and experimental study are carried out on the basis of the theoretical research.The main research contents and results are as follows:(1)Based on the background of the automatic grinding of the thermal insulation layer in the solid rocket motor,the grinding process of the inner-surface of the rotary shell,which is long-narrow and has small opening,were analyzed.The basic structure and control system of the grinding robot are studied.The tasks of the grinding robot are decomposed and planed.The basic model of the grinding robot,which is adapted to the small opening and long-narrow rotary shell,is established based on the kinematics and dynamics analysis.The mechanical body and open control system of the 4-DOF grinding robot are built.The tasks of the grinding robot are decomposed and planed.The basic model of the grinding robot is established based on the kinematics and dynamics analysis.(2)The detection and modeling method of the unknown unstructured environment is established based on laser sensor.Sizes of the inner-surface of different types of solid rocket engine shells are different and not completely consistent with its design size cause of technical reasons.Before grinding the inner-surface should be detected and recognized.The Translation Method and the Walking Method for detection of uncertain environment are proposed in this paper.By measuring longitudinal generatrix of the inner-surface of the engine shell,a series of detection point which reflect features of the surface will be obtained.And a series of the longitudinal generatrix,which is constituted by the detection points,will construct the model of the inner-surface of the rotary shell.The affection of the key parameters of the detection for the detection accuracy is studied by means of numerical simulation.Different fitting method is used to construct the model of the inner-surface.And the accuracy of the constructed model with different fitting method is analyzed.(3)Considering the problem of repeated oscillation between the end of the manipulator and the inner-surface of the engine shell caused by the model error,a criterion for judging whether the manipulator is oscillating is proposed based on hybrid position/force control.Control process of the position and force is studied.The relationship between the controller parameters and the stress response is studied.As there are errors in the surface model,the influence of size and frequency of the environment model error,which mainly depends on the resultant velocity of translational speed of the manipulator end-effector and rotating speed of the engine,is analyzed.The boundary conditions,which make the manipulator stick to the surface,of the control parameters based on the existent model error are put forward.The criterion for judging whether the manipulator is oscillating with the surface is summarized.The simulation results show that when the controller parameters meet the criteria,the robot manipulator will stick to the surface,and oscillating between the manipulator and the surface will be avoided.(4)The influence of model error on grinding force control is analyzed.A fuzzy prediction based on force and position feedback is introduced in the traditional impedance control model.A criterion for judging whether the manipulator is oscillating is proposed based on the proposed impedance control.Combining impedance force control and intelligent control method,the core content is adjusting of the reference trajectory based on the model of the inner-surface obtained by laser detection and force/position feedback of the manipulator.An adjusting factor for fuzzy prediction of the reference trajectory is introduced in to adjust the reference trajectory during the current control cycle.Thus the manipulator can adapt to the model error of the inner-surface and unknown stiffness changing during the grinding process.The influence of model error and change of the stiffness of the surface on the movement of the manipulator is analyzed.A criterion for judging whether the manipulator sticks to the surface is obtained.The validity of the fuzzy impedance control method and correctness are of the criterion is verified by simulation.(5)Numerous of experiments are carried out according to the proposed methods.The inner-surface of the engine shell is detected and modeled according to the proposed rapid detection and modeling method.Force control experiments are carried out based on the established model with existence of several errors in the model.The test results show that,when the parameters of the controller meet the criterion,the force/position control method and impedance control method can adapt to the model error,and the manipulator will stick to the surface.Grinding force is controlled to be stable and nearby the expected value with average error within 20%.Force response curves of two kinds of force control method and different force responses based on different models with dififerent accuracy are analyzed.The influence of model error on grinding force control is studied.The effectiveness of the proposed control algorithm and criterion are verified.