The applications of NARMAX nonlinear identification and Markov parameter linear identification technique

• Main Authors: kung, I-Chung, 龔一中
• Other Authors: Liu, Jui-Jung
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/92537607427613677617

博士 === 中正理工學院 === 國防科學研究所 === 88 === The extraction of information from observations of the surrounding world is fundamental to most scientific studies. This information is often used to develop theories for describing and understanding situations more accurately. The process of observation and modeling is the key to the technique of identification, which has found wide application in almost every branch of science and engineering. The application of identification ideas went back to the time when ancestors used observations of the stars to predict eclipses and follow the seasons. These days the importance of modeling has increased considerably and has become the cornerstone of much research. Models can be qualitative and quantitative. The selection of one plausible model out of a class of possible models and the determination of the unknown model parameters using a limited amount of information are the aim of system identification. In other words, system identification is the process of developing or improving a mathematical representation of a physical system using experimental data. The task of system identification for engineering system is to provide effective and accurate analytical methods, to develop computational procedures, and to use the appropriate tools to identify a model or a class of models which best captures the information of the system. Once a mathematical model of the system is obtained, the model can be used for analysis, prediction, or control of the physical system. The process of system identification generally consists of different procedures such as experiment design, model representation, parameter estimation, and model validation. In practice, these identification procedures have an interactive relationship with each other. Here a linear and a non-linear algorithm of identification will be discussed. The main purpose is that the algorithm can assist the analysis, estimation, and control of the system. The applications include that the force estimation of a cantilever beam and a cantilever plate, the validation of the vibration force estimation to a rotating machine, the river-flow prediction, speed estimation of the induction motor and possible controller design in the future.