Font Size: a A A

A Decoupling Research Of Multi-axes Force Sensor Based On ICA

Posted on:2013-02-22Degree:MasterType:Thesis
Country:ChinaCandidate:P Y ZhangFull Text:PDF
GTID:2218330374960831Subject:Mechanical and electrical engineering
Abstract/Summary:PDF Full Text Request
With the progress of science and technology, the requirements of the sensors in the engineering and technical field has become higher and hirher, and the traditional one-axis force sensor can't meet the requirements of comprehensive measurement, so the development and research on multi-axis force sensors have become an important issue.Due to the Multi-axis force sensors'integrated elastic structure and the process error in the processing and patching, which makes coupling between conversion channels of different directions, and this seriously affects the sensors' measurement accuracy. Therefore, the decoupling of the multi-axis force sensors become a research focus.This paper presented a new decoupling method which based on the independent component analysis of multi-axis force sensor, and It is a new attempt to decouple the multi-axis force sensor. The main work and innovation of the paper is as follows:Calibrating the static and dynamic characteristics of the Six-axis force sensor developed in this laboratory. We used the method of loading standard weight to do Static calibration of Six-axis force sensor to get the main static characteristics indicators of the sensor in three directions force and torque—non-linearity, sensitivity and return error. In addition, we designed three-axis accurate vibration bench calibration system for dynamic calibration of the frequency response characteristics of the three-axis force sensor, It could get the frequency response curve of the sensor in three directions force and torque and defined the sensor's dynamic range of applications.The paper focused on the theoretical algorithm of independent component analysis, and fulfilled the computer realization of the related algorithm through an example. Firstly, we described the background of the independent component analysis using the "cocktail party problem". Secondly, through the hypothesis of ignoring the impact of noise and the compositors are statistically independent and obey non-Gaussian distribution, we introduced the model of independent component analysis, and defined the independent component analysis from two aspects of the probability density and relevance. Third, we introduced the current main methods of estimating the independent component analysis model through the maximization of non-Gaussian, the minimization of mutual information and the maximum likelihood function estimates (ML).The latter part of the paper proposed decoupling methods of the multi-axis force sensor based on the least squares method. The coupling between dimensions is the main factor restricting the sensor's measurement accuracy. In order to overcome the limitations of the traditional decoupling methods, we introduced decoupling methods of the multi-axis force sensor based on independent component analysis and applied the method to decouple the full shear strain gauge six-axis force sensor. Then we compared it to the linear decoupling method of the least squares method and the results showed that the independent component analysis had higher calibration accuracy than the least squares method.
Keywords/Search Tags:multi-axis force sensor, static and dynamic characteristics, ICA, decoupling
PDF Full Text Request
Related items