Research on Improvement of Model Identification Technology for Unmanned Vehicle System

• Main Authors: WANG, CHAO-HSIEN, 王肇賢
• Other Authors: HAO, SHU-SHENG
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/2he897

碩士 === 國防大學理工學院 === 電子工程碩士班 === 108 === In view of the current difficulties of power system modeling and model errors, the purpose of this thesis is to apply existing system identification software and establish a method to improve the identification accuracy. The research results are also verified by the actual filed tests of the unmanned ground vehicle (UGV). The mainexperimental vehicle of this thesis is Tamiya TXT-1 UGV. Firstly, we set up sensor module to help collect the actual dynamic information of the UGV. Afterwards, we applied Matlab/SITB to process the dynamic information for system identification and buildthe mathematical model. To improve the accuracy of system identification, we used the recursive method to estimate the model, especially considering the noise distribution of the measurement signal and the filter design of the model when estimating the system parametersin order to obtain the best system identification scheme. Different from the general system identification research that generally lacks the hardware verification,after establishing the software mathematical model, we applied the Tamiya TXT-1 for actual hardware tests. Finally, after comparing with the actual dynamic information of UGV and the identification model of this study,it’s confirmed that the identification improvement scheme proposed by this research has the extremely high precision, and it can provide a considerable application in future system modeling and subsequent integration, and it can also be used for the reference by subsequent researchers.