Polishing Robot Force Control Technology And Application Based On Big Data

Installing a six-axis force/torque sensor on industrial robot is a commonly used control strategy on robot constant force.However,due to the closedness of the robot control system and the high price of the force sensor,this method is not convenient for industrial mass production applications.Utilizing big data advanced technology,we could establish a data model for the complex non-linear relationships of the force at the end of the robot.Using the feedback by data model instead of a force sensor,we achieve the effect of sensorless force control of the grinding robot,and this can be extended to large-scale industrial production.According to the force condition of the grinding robot and the magnitude and direction of the force applied on the work-piece during the grinding process,this paper preliminary set of the collected data being the torque of each axis and normal grinding force during the operation of the grinding robot.In this paper,the data acquisition method and data preprocessing are analyzed and explained.The method of gravitational robot tool gravity compensation between the collected force information and the real contact force information is analyzed.So far this article has obtained the relevant available data.This paper used neural network regression to establish a data model which is the relationship between the torque of each axis of the grinding robot and the end normal grinding force and used the genetic algorithm in the neural network training process to optimize the selection of the initial parameters in the neural network so that this paper overcomes the defect that the neural network training can easily fall into the local optimal value.This paper established the relationship between the normal grinding force and the normal displacement.Then running the polished initial trajectory which generated by the robot off-line programming software,this paper obtained the torque of each axis of the grinding robot.According to the data model of the previous training,this paper predicted the grinding force of the end of the grinding robot.Then this paper compared the predicted grinding force with the previously set grinding force value,and adjusted the initial grinding path according to the normal grinding force difference so that the final trajectory meet the grinding robot constant force grinding effect.Based on the above research,this paper have developed data acquisition and display modules and intelligent optimization modules on offline programming software which could facilitate data acquisition and analysis and polishing robot path optimization between PC and robot controller.By grinding robot constant force grinding experiment,the effectiveness and practicability of this method are verified.