Modeling and Prediction of Broadband Signals Using Symmetric Acausal Impulse Responses

In modern power dynamic system, the identification and simulation of signals from various sources is one of the important ways to analyze the model features. The current methods are not efficient due to a large number of coefficients, and have limitations on modeling signals which have energy distributed over a large frequency range. In this study, a new method of estimating and predicting broadband signal models from symmetric acausal impulse responses is proposed. The procedure involves using least squares algorithm to calculate parameters and refining the model using prediction error methods. The detailed steps of the algorithm were described and derived. A spectrally equivalent causal infinite impulse response filter is constructed to preserve the power spectral density of a signal which is generated from acausal impulse responses. The designed filter is able to follow the characteristics of signals in frequency domain. Experiment results of a broadband audio signal from measurement are presented to illustrate the performance of the methodology.