Transformer Fault Gas Detection Based On The Broadband Photo-Acoustic Spectroscopy

The concentration of dissolved gas in transformer oil reflects the operation condition of the transformers,the ratio of the gas concentration indicated the possible fault type in transformer,therefore the detection the dissolved gas concentration can provide technical parameters for transformer maintenance,it has important significance to ensure the safety of the power grid operation.Photoacoustic spectroscopy(PAS)has the advantage of zero background,high sensitivity,wide detection dynamic range and the intensity of PA signal does not rely on the light intensity,so it is considered to be an efficient absorption spectroscopy in gas detection.In the thesis a high performance non-resonant photoacoustic cell was designed based on the principle and theory of photoacoustic spectroscopy and a broadband photoacoustic spectroscopy detection system has been developed,which can be used in the detection of fault gas dissolved in transformer oil.Several approaches include the collimation of the light beam,the calibration of optimum modulation frequency and noise suppression and temperature compensation have been used to improve the sensitivity and stability of the system.The broadband PA detection system was calibrated by different concentration of fault gases,the overall performance of the system was studied in this thesis.The different concentration of various fault gases generated mass flow controller were detected by designed broadband PA system,the experiment shows that the system has good linearity and high sensitivity.Then the important parameters of temperature and pressure in PA signal generation were studied.In order to verify the accuracy of the detection system,the same concentration of fault gas and direct absorption detection system was used to contrast detection with PA system.The artificial neural network algorithm was introduced into the multi-component gas concentration detection to reduce the interference between gas absorption signals to reduce detected concentration error.The photoacoustic detection device and degassing device was integrated for transformer oil dissolved gas detection.The main research results and innovation of thesis are as follows:(1)The multi-component gas detection system was developed based on photoacoustic spectroscopy utilize a broadband thermal radiation light source and non-resonant photoacoustic cell,compared with the existing PA cell,the PA cell designed in this thesis,a thermistor was installed to detected the temperature in PA cell synchronously and there is a reflector installed on the output side of PA cell instead of ordinary optical windows,so the incident light will be reflected back,it is equal to enhance the incident light power,so the PA signal was enhanced.In order to let the light beam of broadband light source maximum into PA cell to enhance the utilization ratio of light source,a self-designed reflector was developed for the broadband light beam shaping refer to the design of LED light collimation system.Though the simulation of TracePro software,the designed optical reflection system meet the requirements.(2)Through the response of detection system under different modulation frequencies,the optimum modulation frequency was selected.Utilize the designed broadband PA system for a variety of different concentrations of fault gas detection,the system has high sensitivity and good linearity.Then the important parameters of temperature and pressure in PA signal generation were studied.The experiment illustrated that when the pressure in PAcell near to a atmospheric pressure,the impaction of pressure to PA signal to be very little.The intensity of PA signal under different temperature was investigated,a set of temperature compensation algorithms was designed for the PA signal correction to effectively eliminate the impaction of temperature to PA signal,and the stability of the system was further improved.In order to investigate the detection performance of designed PA system,the same concentration of fault gas was selected to be detected by direct absorption spectroscopy(DAS)based on multi-pass cell and broadband PA system,the experiment illustrate that the concentration acquired by PA detection system has approximate result to DAS,it is proved that the designed broadband PA system has good detection accuracy and it is suitable for transformer fault gas detection.(3)The artificial neural network was used to reduce the cross interference of absorption spectrum lines.Because of the certain bandwidths of narrow band filters,the absorption lines of fault gases will overlap,so the PA signal have interference each other.In order to reduce the interference between gas absorption signals,the artificial neural network algorithm was introduced into the multi-component gas concentration signal processing to reduce the error of different gas concentration detection.(4)3D printed technology in was applied in photoacoustic system.In order to develop the portable,robust and compact PA detection device,a miniaturization resonant PA cell was produced by 3D printed technology,compare to traditional mechanical manufacturing technology,the 3D printed technology used in PA cell process has the advantage of high machining accuracy,easy accessibility and once shaped.