Spectrum estimation by using of neural network method

Spectral estimation has been applied broadly in the signal-processing domain. Modem spectral analyzing method has been paid more attentions for its advantages such as higher ability in spectral analyzing and shorter length of data series to be need than the traditional method. In modem spectral estimation method, the model parameters will be obtained by solving the Yule-Walker equations, such as ARMA model spectral estimation, it is in order to obtain the parameters of ARMA model. There are several common disadvantages such as complicated processing steps and heavy calculation load in all the optimized algorithms. In this paper, two methods by new using of neural network are discussed. In the first one, the solving of the equations set is avoided. hi the second one, the solving of the equations set is simplified by using a simple BP neural network. Simulations show that the spectral analyzing effect is better than some other methods.