A study of the Inverse Numerical Method for the One-Dimensional Heat Conduction Problem

• Main Authors: Ru-Fen Lin, 林如芬
• Other Authors: Wu-Yao Chiou
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/r5m67n

碩士 === 國立高雄應用科技大學 === 模具工程系 === 97 === A numerical analysis of one-dimensional inverse problem with half infinity boundary is presented in the paper. In engineering utilization, such as electrical discharge machining, welding…and so on, belongs to this kind of question category. First, the numerical analysis method uses exact solution to regard as measured value. To solve thermal properties of material are estimated by means of finite difference method in conjunction with conjugate gradient method. The method divides the optimal computational procedure into three problems as (1) direct problem, (2) sensitivity problem, (3) gradient equation. To estimate thermal conductivity, heat capacity and thermal diffusivity of material in the time after iteration operation. Results show that the main factors in the analysis are: the sensor locations, the number of sensors, the time intervals, the initial estimate value. The numerical analysis results show the errors will be decreased with the increase of the number of sensors with the position is in the center of region location, and the more measuring of time the errors tend to more stable. The initial estimated value is closer to the exact solution, the numerical errors are more stable with low iteration times, and accelerated the numerical convergence.