Research On The Total Error Analysis And Calibration Of A Parallel 6-DOF Earthquake Simulator

Earthquake simulator is often used in the research of earthquake to generate the simulated seismic waves or specified vibration. Building models should be tested under different vibration waves in order to identify the seismic performance, or strengthen the structure design. However, the process of manufacturing and installation of earthquake simulator, will inevitably generate a variety of errors, these errors will affect simulated kinematic accuracy. Therefore, in order to reproduce the seismic wave accurately or generate specified vibration, the total errors need to be analyzed, so the errors could be understood better. Total errors analysis could also guide the design integration. At the same time, the parameters need to be calibrated so that the kinematic accuracy can be improved.In this paper, a new type of parallel redundancy earthquake simulator is introduced. In order to improve the kinematic accuracy, the total error model of earthquake simulator is built, analyzed and optimized, the calibration method is studied as well.First of all, this redundantly actuated branched-chain need to be simplified, and a corresponding D-H co-ordinate system should be built. With the setting of auxiliary coordinate system, this model successfully contains neglected errors in previous publications, such as branched-chain installation location etc., which could ensure the results of error analysis more comprehensive and effective.Second, by means of Monte-Carlo simulation, the total error parameter is analyzed in the whole workspace, study the range of the kinematic error cause by single error and its distribution probability is studied. Based on the results, sensitive error parameters are summarized, and position and posture selection strategy of calibration is proposed. As the parameters and pose is selected more comprehensively, the error could be understood more deeply by the results of analysis.Then, this paper build the calibration model with sensitive error parameters, in order that the calibration model could examine more error parameters. Levenberg-Marquardt algorithm is adopted in parameter identification procedure which based on least squares method. Simulation results show the effectiveness of the algorithm and the integrity of calibration model which based on sensitive error parameters.Finally, in order to reduce complexity of calibration, improve efficiency of calibration, this paper analyzes the relationship between the rank of identification matrix and the number of measurement points, so that the least independent parameters can be selected from the sensitive error parameters. The selection strategy of independent error parameters could not only ensure identification of the system error, but also avoid omission of independent parameters in calibration. According to parameters interchangeability, this strategy provides a very useful reference for the calibration of other parallel mechanism.