基于传播算子的间谐波参数智能估计算法

为了降低多重信号分类(MUSIC)法频率估计的计算复杂度,提出基于传播算子的间谐波频率估计方法.通过传播算子可以得到噪声子空间,不需要估计协方差矩阵和进行特征分解,并且不需要间谐波个数的先验知识,基于传播算子MUSIC算法的频率估计性能与MUSIC算法几乎相同.构造复数域自适应线性神经网络模型来估计谐波和间谐波的幅值和相位.该模型的输入变量和权值仅为实数域自适应线性神经网络的一半,简化了网络结构；采用Levenberg Marquardt(LM)算法对网络进行学习,大大减少了学习次数.仿真结果表明,该算法无需同步采样,能够快速准确地估计间谐波的频率、幅值和相位.

Interharmonic parameter intelligent estimation algorithm based on propagator method

An interharmonic frequency estimation algorithm based on propagator method (PM) was proposed in order to reduce the computational complexity of multiple signal classification (MUSIC). The propagator can be used to construct the noise subspace. The PM didn’t involve covariance matrix and eigenvalue decomposition and the priori knowledge of interharmonic number, and had the approximation performance compared with MUSIC. A complex Adaline neural network was employed to obtain amplitudes and phases of harmonics and interharmonics. The proposed complex Adaline structure was based on Levenberg Marquardt (LM) rule. The algorithm reduced input vectors and weights to half of the number that real Adaline used. These attributes can increase convergence speed. The simulation results show that the algorithm can accurately achieve frequencies, amplitudes and phases of interharmonics without synchronous sampling data.