摘要
基于微分算子在紧支撑正交小波基下的精确显式表示,给出了一种分布参数系统辨识方法。将微分算子投影到小波空间,得出其矩阵表示形式,从而将分布参数系统转化为集中参数系统,再利用最小二乘参数估计的一次完成算法进行辨识。该方法不需要考虑初始条件和边界条件的影响,降低了计算的复杂程度;基于Daubechies(db1)小波的数值计算表明,在小波分解层数很低的情况下,辨识就具有很高的精度。
Based on the exact and explicit representations of differential operators in orthonormal ba- ses of compactly supported wavelets, this paper presents an identification method for distributed parameter systems (DPS). The matrix representations are obtained by projecting differential operators onto wavelet space. The proposed method translates DPS into lumped parameter systems. Identifica- tion can be made with the algorithm of least square parameter estimation. The complexity of computation decreases because boundary conditions and initial conditions need little consideration. Numerical experiments are performed based on the wavelet of Daubechies (dbl) and the results are accurate even with very low level of wavelet decomposition.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2007年第4期449-452,共4页
Journal of Nanjing University of Science and Technology
基金
国防科技重点实验室基金项目
南京理工大学科研发展基金项目(XKF07010)
关键词
分布参数系统
辨识
微分算子
小波变换
distributed parameter systems
identification
differential operators
wavelet transform