Photoacoustic spectrometry identification method and device for characteristic gas in transformer oil

Abstract

The invention relates to a photoacoustic spectrum identification method and device for characteristic gas in transformer oil. The method comprises the following steps: acquiring a plurality of photoacoustic spectrums of target characteristic gas under different excitation light sources, different temperatures, different concentrations and different pressures; performing image enhancement on each photoacoustic spectrum, removing spectral lines of which the absorption intensity is lower than a threshold value, performing segmented sampling and smooth filtering on each photoacoustic spectrum in sequence, and constructing a photoacoustic spectrum data set; mapping the temperature, the concentration, the pressure, the shape of an absorption spectral line, the position of the absorption spectralline, the intensity of the spectral line and a peak absorption coefficient of the target characteristic gas into a multi-dimensional vector by utilizing a principal component analysis method when thephotoacoustic effect occurs; and training a convolutional neural network by utilizing the characteristics and the photoacoustic spectrum data set and applying the convolutional neural network to photoacoustic spectrum identification. According to the method, a traditional spectrum processing method and the convolutional neural network are combined to recognize the characteristic information of the characteristic gas of the photoacoustic spectrum, the recognition speed is high, the accuracy is high, and the method can adapt to various working environments.

Description

technical field [0001] The invention belongs to the field of power equipment measurement and deep learning, and in particular relates to a photoacoustic spectrum identification method and device for characteristic gases in transformer oil. Background technique [0002] Photoacoustic spectroscopy technology is a new simple, high detection sensitivity, high selectivity, large dynamic range, strong universality, and no damage to the sample analysis and testing method. The method is different. It directly measures the absorbed energy instead of measuring the projected or reflected light intensity. It is considered to be one of the best tools for detecting trace gases and is widely used in many fields. The basic principle is the photoacoustic effect. The photoacoustic effect is a photoacoustic conversion phenomenon first discovered in solids by Alexander Graham Bell (A.G.Be11), an American scientist and founder of the Bell Telephone Company in 1880. He discovered that when the s...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problems that the existing photoacoustic spectrum detection technology of gas in transformer oil cannot satisfy the problems of high detection accuracy, low failure rate and long service life at the same time, the present invention provides a photoacoustic spectrum identification method for characteristic gases in transformer oil, The method includes the following steps: obtaining multiple photoacoustic spectra of the target characteristic gas at different excitation light sources, different temperatures, different concentrations and different pressures; each band of the photoacoustic spectrum includes a single absorption band of the target characteristic gas and multiple gases. Absorption band; image enhancement is performed on each of the photoacoustic spectra and spectral lines whose absorption intensity is lower than the threshold are removed, and then the photoacoustic spectra are sequentially segmented and sampled, smoothed and filtered to construct a photoacoustic spectrum data set; from the photoacoustic spectra The shape of the absorption line, the position of the absorption line, the intensity of the line and the peak absorption coefficient of the target characteristic gas are extracted from the acoustic spectrum data set, and the temperature, concentration, pressure, The shape of the absorption line, the position of the absorption line, the intensity of the line and the peak absorption coefficient are mapped to a multidimensional vector; the photoacoustic spectrum in the photoacoustic spectrum data set is used as a sample, and the multidimensional vector is used as a label to construct a sample data set; The sample data set trains the convolutional neural network until its error is lower than the threshold and tends to be stable, and the trained convolutional neural network is obtained; the photoacoustic spectrum to be identified is input into the trained convolutional neural network, and the obtained convolutional neural network is obtained. The identification information of the target characteristic gas in the photoacoustic spectrum; the identification information includes the concentration of the target characteristic gas, the shape of the absorption line, the position of the absorption line, and the intensity of the line

Images