Compensation Method Of Industrial Robot Accuracy And Experimental Research For Aircraft Automated Assembly

The production efficiency and quality reliability of aircraft can be greatly improved whenindustrial robots are introduced for flexible automatic assembling and component machining inaviation manufacturing. Positioning accuracy is one of the important index of the robot performance,which depends directly on differences between the actual value and the nominal value of robotkinematic parameters. The industrial robots usually have a high repeat positioning accuracy, but theabsolute positioning accuracy is still far from being able to meet the precision requirements of aircraftassembly. In order to use industrial robots extensively in the aviation industry, all need to do first is toimprove its absolute positioning accuracy. This paper proposes a compensation method based onweight measured error similarity on the basis of establishing the robot positioning error model.Considering the impact of ambient temperature changes to the robot positioning accuracy, anintegrated compensation method based on particle swarm optimized BP neural network is alsoproposed, and the proposed methods are verified by experiments. The main research contents of thispaper are as follows:（1）The robot kinematic analysis and positioning error modeling. The KUKAKR150-2robot’skinematic model is established by using the classic D-H modeling method, and the forwardkinematics and inverse kinematics are analyzed, on which the robot positioning error model isestablished based on the idea of differential transformation. Taking account of the error introduced inestablishing the robot coordinate system, he positioning error model is corrected.（2）A compensation method based on weight measured error similarity is proposed. Theconcept of positioning error similarity is proposed by studying robot positioning error model, onwhich the compensation method based on weight measured error similarity is proposed. By giving themaximum allowed positioning error, the sufficient conditions for determining the error similarity isdeduced. The concept of optimum grid step is also proposed. A KUKA KR150-2industrial robot isstudied to determine its optimal grid step through experiment.（3）An integrated compensation method based on particle swarm optimized BP neural networkis proposed. The principle of BP neural network based on particle swarm optimization algorithm isstudied on the basis of BP neural network and particle swarm optimization algorithm. An integratedcompensation method based on particle swarm optimized BP neural network is also proposed byconsidering the impact of ambient temperature changes to the robot positioning accuracy, combining proposed compensation method based on weight measured error similarity. The cross-validationmethod used to verify the stability and adaptability of neural network model is studied, some newevaluation indexs are added to the common ones under the actual subject situation.（4）Robot accuracy compensation experiment is studied. For the traditional calibration methodof robot tool parameters cannot meet the requirements of high-precision machining, a calibrationmethod based on indirect measurement of tool parameters is proposed, meanwhile coordinate systemsrefered in the robot automatic drilling system are studies.The compensation method based on weightmeasured error similarity is verified respectively in a large area and a small range of specified areawhen the robot’s load is empty. Then the integrated compensation method based on particle swarmoptimized BP neural network is also verified by drilling on a plate when the robot mounted an endeffectors under the situation that the ambient temperature changes.The result of the paper can not only effectively promote absolute positioning accuracy of robots,but also can further promote the application of industrial robots in the field of aviation manufacturing.