An algorithm for active control of transformer noise with on-line cancellation path modelling based on the perturbation method

Abstract

Previous work has demonstrated the potential for the active control of transformer noise using a combination of acoustic and vibration actuators and the filtered-x LMS algorithm (FXLMS), the latter being implemented to make the system adaptive. For a large electrical transformer, the number of actuators and error sensors needed to achieve a significant global noise reduction can be up to hundreds, and this makes the convergence of the FXLMS algorithm very slow. The memory requirement for the cancellation path transfer functions (CPTF) and the computation load required to pre-filter the reference signal by all the CPTFs are relatively large. On the other hand, not only the transformer noise but also the CPTF varies considerably from day to day, which makes on-line CPTF modelling very necessary. A new adaptive algorithm based on waveform synthesis is proposed, and the perturbation method is used to obtain the CPTF on-line. A comparison of the performance of the proposed algorithm with the FXLMS algorithm and the H-TAG algorithm shows the feasibility of the algorithm for the control of a slowly time-varying system with just a few fixed frequency components.