摘要
本文对梯度正则化方法(GradientRegularizationMethod)作了进一步的研究,给出一种建立了梯度正则化迭代算法和选择正则参数的简明实用方法。文中椭圆算子方程参数识别算例不仅说明了GR法具有广泛的适用性和一定的抗噪音能力,而且收敛速度较快,具有较大的收敛范围。
This paper makes further researches on the Gradient Regularization method, and not only establishes a fresh iterative gradient regularization algorithm in a brief way but also gives a practical way to choose regularization parameters. These numerical examples which identify parameters of elliptic operator equations not only show that GR method has a wide scope of application and poses good capability of noise resistance, but also illustrate that GR method poses higher convergence speed and larger convergence scope.
出处
《计算力学学报》
CAS
CSCD
2000年第1期69-75,共7页
Chinese Journal of Computational Mechanics
基金
国家自然科学基金!项目编号 :5 97790 0 3
关键词
梯度正则化方法
参数识别
材料
物性
gradient regularization method
Frechet derivative
regularization parameters
parameter identification
ill posedness
elliptic operator equation
