Research On Detecting Technique Of The Gas Dissolved In Transformer Oil Based On The Blind Signal Separation

With the rapid development of computer, network and communication technology, mankind has entered the information and digital age. And the digital signal processing is one of the most crucial technology. Blind Signal Separation(BSS) is main content of the blind signal processing. Because the BSS could separate source signals in each component only based on the source signal statistical characteristics without knowing the source signal and the transmission channel parameters, it has become a hot research content of current signal processing area. At present, BSS has been widely used in speech recognition, mobile communication, image processing, biomedical, economics, sonar, seismic signal processing and other fields. The modern society, electricity has become an essential part of people’s daily life, people also make higher demand for the security and reliability of the power grid. As one of the most important part of the power grid, the large transformer’s stability in a certain extent determines the safety of the power grid.In the article, the BSS algorithm is combined with photoacoustic effect. An experimental platform of gas component in oil detection system based on blind signal separation is designed. It realizes the detection with little time and high precision.In the article, first of all, the detection method of the mixture gases in transformer insulating oil’s research situation are introduced and compared. And the method of the BSS research present situation is introduced. Secondly, summarized the method of BSS, introduced the mathematical foundation of the BSS, discussed the mathematical model, the separability of blind signal and the two uncertainty of sequence and magnitude and the algorithm principle. The Independent Component Analysis method, a classical algorithm of the BSS, is also been made a careful study. Thirdly, an experimental platform, including select the light source, chopper, photoacoustic cell, lock-in amplifier, ARM control system, of gas component in oil detection system based on blind signal separation is designed, including the hardware and the software design, Finally, analyzed the BSS model of the mixture gases and blind separability. Used the FastICA method to separated the C2H6and C2H2which have the similar center frequency. And the result of separation algorithm is evaluated by correlate method.The experimental results and the assessment results show that using the BSS algorithm to separate mixture gases is feasible. The experimental platform and the algorithm have the advantage of fast detection and highly precision.