Research On The Static And Dynamic Performance Calibration System For Robot's Multi-Axis Wrist Force Sensor

A static and dynamic performance calibration system for robot's multi-axis wrist force sensor is lucubrated in this dissertation. The main research works are as follows. The calibration method is studied on. The performance characteristics and the calibration routes are determined. It is studied that the correlation wavelet transfer method used in wrist force sensor's dynamic performance calibration based on step response. The method of wavelet denoising is used in preprocessing data of dynamic performance calibration. The way of loading on the sensor is also studied. The calibration experimental-bench is designed, manufactured and debugged. The data acquisition system is developed. The error of the calibration system is analyzed. The virtual instrument of performance calibration on LABVIEW is also developed.This dissertation is organized as:1. In chapter 1, it is introduced that the robot's multi-axis wrists force sensors' development, various configurations and applications. It is summarized that the current research works about static and dynamic performance calibration of the robot's multi-axis wrist force sensor. The insufficiencies of the research works are also generalized. Then the background, purpose and signification of the research work in this dissertation are put forward. The dissertation's task is also definite.2. In chapter 2, the calibration method of the robot's multi-axis wrist force sensor is lucubrated. It is put forward that the high calibration precision lies on the measuring precision of each channel and loading precision. The matrix of calibration loading should be orthogonal. It is studied that the wrist sensor's static characteristics and its calibration method.It is put forward that the step response method should be applied in the dynamic performance calibration by comparing with other sensor calibration experimental methods. The problems are point out that the mean power spectrum method used to obtain the wrist force sensor's frequency response function in the step response experiment. It is researched that graphic method to identify the wrist force sensor's dynamic characteristics based on step response. It is point out that the graphic method is simple and convenient, but the identification result is easy to be influenced by the noise and the identification precision is low. It is also researched that the finite difference method used to obtain the wrist force sensor's pulse response function. This method is efficient because the step response signal is a two-value signal but the identification precision is easy to be influenced by the noise also. At last, a correlation wavelet transfer method is put forward to be used in the wrist force sensor calibration based on the step response. It is used to obtain the wrist sensor's pulse response function. The correlation wavelet transfer method analyzes the signal by decomposing it into two-dimension space of time-frequency. It is good for non-stationary and transient dynamic signal. In the on-line calibration, it is easy to create step exciting by the magic hand releasing work-piece while the robot is working. So the study of correlation wavelet transfer method based on step response being used to obtain the wrist force sensor's dynamic characteristics will also be very useful in the on-line calibration.The repeatability and linearity of the dynamic performance calibration based on step response are also studied by a large quantity of experiments.In this chapter, the wavelet threshold denoising method is researched. Decomposing the noise signal by the wavelet, depressing the wavelet coefficients of the frequency band involved noise, reconstructing the signal by depressed wavelet coefficients. Then the signal is denoised and the ideal signal is obtained. This wavelet threshold denoising method will be very useful in on-line calibration.3. In chapter 3, the way of loading on the wrist force sensor is studied. The calibration experimental-bench is designed, manufactured and debugged. The orthogonal calibration force/moment could be obtained on this experimental-bench. By difference accessories, the static and dynamic performance calibration could be done on the same experimental-bench. A piezoelectricity force sensor is used when the dynamic performance of the robot's multi-axis wrist force sensor is calibrated. By the signal of the piezoelectricity force sensor, the "soft-trigger" can be set up to caught the step change time of the wrist force sensor exactly. Then the signal of the wrist force sensor's exciting and response will be intercept by the step change time for the performance evaluation. At last, the method of measuring the value of exciting force/moment by the piezoelectricity force sensor's output is researched.4. In chapter 4, the data acquisition system used in the calibration of robot's multi-axis wrist force sensor is developed. Its data acquisition breadboard can be infixed in the ISA bus slot of the computer. It can sample the signal of 8 channels synchronously. The sampling rate is very fast. The data acquisition breadboard has a double-port RAM. It can communicate with the other computer in real time. So it has the functions of data transferring, modifying model, etc. In the process of calibration, the sampled data are saved in data files prepared for being used to analysis the wrist force sensor's performance.5. In chapter 5, the error of the calibration system is analyzed. The experimental-bench's error is analyzed and calculated. The error of the force acting on the sensor is E_I≤0.1271%. The error caused by the interference force is E_II < 0.00667% . Method of reducing the error of the experimental-bench is also put forward. The errors of data acquisition system are also measured. The signal-to-noise rate of each channel is greater than 65.32dB. The interference between channels is greater than 20.0dB. The error of transfer function's amplitude frequency characteristics between channels is less than 1%. The error of transfer function's phase frequency characteristics between channels is less than 0.125°. The quantization error of ADC is calculated. It is less than 0.0061%. The algorithm error of the data process and performance analysis system is very small and could be ignored. Above all, it is concluded that the performance calibration system's error is very small. Its precision is high.6. In chapter 6, the present research condition and development trend of the virtual instrument is summarized. The constitution principle and data stream of robot's multi-axis wrist force sensor calibration system is studied. The virtual instrument on LABVIEW is developed for the sensor's calibration. The virtual instrument calls MATLAB programs by MATLAB Script to calculate. The virtual instrument consists of data display module, static calibration module and dynamic calibration module. The functions of all modules are finished. The man-machine interactive interfaces of the virtual instrument are friendly and convenient.7. It is roundly generalized that the originality innovations and the deficiencies of the dissertation. Then it is pointed out that the keystones, research directions and the problems to be further studied on.