Implementation of the frequency domain LMS adaptive filter in MATLAB for active vibration control application

摘要

The complex LMS adaptive algorithm has been implemented for the single channel active vibration control system in the frequency domain. The Fourier transformation of the input signal and the independent weighting of the contents of each frequency bin can accomplish adaptive filtering in the frequency domain. Block processing of the data samples has been made use of. The overlap save technique has been incorporated for the sectioning of the input. This will significantly reduce the computational complexity of adaptive filters The computational complexity of the'adaptive filter has been reduced significantly through elegant parallel processing of the data samples. It should be noted that the sharing of the processing time among the samples in each block is the key to achieving high computational efficiency. Frequency domain filter performs similarly to a conventional adaptive transversal filter but promises a significant reduction in the computation when the filter length equals or exceeds 64 sample points. It is apparent that in many practical case the savings resulting from the use of the frequency domain technique are substantial so much so that even when one takes into account the additional computing of three FFT's per data block it pays to use this technique and to recover the filtered signal from its transform. The output can be accumulated in a buffer memory and " stitched together" creating a continuous though delayed flow of output data. In this way any large adaptive filter can be efficiently realized by a combination of the block LMS and FFT algorithms.