Mixed gas photoacoustic spectrum recognition method and device based on deep learning

Abstract

The invention relates to a mixed gas photoacoustic spectrum recognition method and device based on deep learning, and the method comprises the steps: obtaining the photoacoustic spectrums of a plurality of mixed gases, and recording the photoacoustic spectrums as first photoacoustic spectrums; sequentially performing Fourier deconvolution and bilateral filtering on the plurality of first photoacoustic spectra to obtain a plurality of second photoacoustic spectra; carrying out peak division on the second photoacoustic spectra by utilizing a derivative method, and then extracting waveform characteristics and gas characteristic information of each waveband; constructing a multi-dimensional vector by utilizing the waveform characteristics and the gas information, and constructing a sample dataset; and training a target recognition neural network by using the sample data set, and inputting a photoacoustic spectrum to be recognized into the trained target recognition neural network to obtain recognition information. According to the method, a traditional filtering method and a derivative method are combined to perform peak division on overlapped peaks, then the target recognition neuralnetwork is utilized to recognize the photoacoustic spectrum, the recognition speed is high, the cost is low, and the stability is good.

Description

technical field [0001] The invention belongs to the field of data processing and deep learning of photoacoustic spectroscopy, and in particular relates to a method and device for identifying mixed gas photoacoustic spectroscopy based on deep learning. 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 tha...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problems of the difficulty in identifying overlapping peaks in the photoacoustic spectrum of mixed gases in the existing photoacoustic spectrum detection technology of mixed gases, as well as the high cost and low stability of equipment in the existing identification methods, the present invention provides a method based on deep learning The mixed gas photoacoustic spectrum identification method comprises the steps of: obtaining multiple photoacoustic spectra of the mixed gas, recording the photoacoustic spectrum as the first photoacoustic spectrum; performing Fourier sequentially on the multiple first photoacoustic spectra leaf deconvolution and bilateral filtering to obtain multiple second photoacoustic spectra; determine the order of the derivative according to the number of single peaks contained in the overlapping peaks in each second photoacoustic spectrum, so that each second photoacoustic spectrum The number of overlapping peaks in the derivative photoacoustic spectrum of the photoacoustic spectrum is lower than the threshold; extract the maximum absorption position, absorption depth, symmetry and corresponding Gas information, and map it to a multidimensional vector; the gas information includes the concentration of gas; the first photoacoustic spectrum and the multidimensional vector are respectively used as samples and labels to construct a sample data set; using the sample data set Train the target recognition neural network until its error is lower than the threshold and tends to be stable, and obtain the trained target recognition neural network; input the photoacoustic spectrum to be recognized into the trained target recognition neural network, and obtain the photoacoustic spectrum The identification information; the identification information includes the composition of the mixed gas, the maximum absorption position of the absorption peak, the absorption depth and the degree of symmetry

Images