Research On Electromagnetic Method For Insulation Detection Of Stator Lamination

During the assembly, transport and running process of modern large turbine generator, if any insulation defects or flaws exist, the turbine generator will encounter overheating under normal operating condition, which dose harm to the generator. The foreseeable consequence of overheating is inefficiency, winding insulation accelerated aging, and the service life shorten. But that is only part of the story, the other half is catastrophe, for the stator core will burn, resulting in the breakdown of main insulation, and causing blackouts. Therefore, in the manufacturing, operation and maintenance process of the turbine generator stator, regularly nondestructive testing is very necessary.In view of the present situation of nondestructive testing equipment for stator lamination of large turbine generator by foreign long-term monopoly, this paper completed the following work:Research and development of nondestructive testing instrument were conducted. To reduce the interference, this paper firstly designed a band-pass filter, whose upper and lower cut-off frequency can be accurately and flexibly setting according to field interference. For the fault current is usually very small, at the first stage the amplifying circuit with the amplification flexible adjustment was designed. And the acquisition system with the sampling rate of 1Ms/s and the resolution bit of 16 was designed. The FPGA is utilized to perform the sampling logic control and data storage. An ARM based Processor is employed to accomplish the data interaction between FPGA through high-speed parallel bus. The embedded Linux system carried out data analysis and data processing function. Hardware design and software debugging results show that, the portable testing instrument designed in this paper has the advantage of small background noise and powerful data processing capability.De-noising algorithm was designed and implemented. Due to the hardware band-pass filter can only remove part of the electromagnetic interference noise, so in this paper we use wavelet analysis to denoise noise. Firstly, the theory of wavelet analysis is simply introduced, and then based on the discrete Mallat wavelet, the soft threshold method are presented to filter the noise...