Research On The Noise Analysis And Active Noise Control Of Power Transformers

With the development of national economy and the expansion of the urban power grid, many high voltage transformers are gradually close to the urban area, which does harm to people’s work and life. The noise control of large transformers is badly in need of solution.The traditional measures of transformer noise control are mainly based on the damping of the sound source and the insulation of the sound travel path. The results are based on a lot of engineering tests and difficult to quantitative analysis. With the development of modern control technology and the decline of the control chips cost, active noise control becomes the research hotspot. According to the nature of the noise field, active noise control can design and change various characteristics of the control system, which is effect on low-frequency noise control and reduces the impact on noise sources. According to linear superposition principle, the effect of active noise control depends entirely on the relationship between the amplitudes of the primary source and the secondary source.Based on previous researches, two measurement methods, sound pressure method and sound intensity method, are compared and discussed. According to the measurement, transformer noise energy is concentrated in the low frequency region. At100Hz,200Hz,300Hz and400Hz, the noise is higher than other frequencies. With periodic features, transformer noise waveform is relatively stable, which facilitates the implementation of active noise control.Based on the measurement and analysis, we carried out a study of the transformer active noise control on noise radiation model and parameters optimization of the secondary sources. With consideration of practical requirements, key parameters of the secondary sources are divided into two categories as to realize parameter optimization in terms of source number, location and strength. As interactive impacts resulted from secondary source location and strength present, a scheme to alternately optimize the fixed parameters and the flexible parameters is proposed, and thereby a gradual search approximation strategy based on genetic algorithm is further put forward.