Research On Aerosol Leak Detection System Based On Infrared Absorption Spectroscopy

An aerosol is a preparation prepared by pressurization.Common products include pesticides,fresheners,and the like.Such products often contain flammable and explosive gas components.If the package is not qualified,leakage will accumulate.The boots may be a constant temperature water bath method in the traditional aerosol leak detection method,which has the disadvantages of low sampling rate and water and time consumption.,Unable to meet the increasingly stringent regulatory requirements in the industry.Therefore,the development of aerosol leak detection system based on mid-infrared absorption spectrum is of great significance to the safe production of the aerosol industry.Here,the leakage detection system based on infrared absorption spectrum technology is mainly researched and constructed,as follows:（1）The source of the most significant absorption peak in the spectrum is determined by detecting and analyzing the absorption spectrum of the aerosol product,combined with the chemical properties of the coupling group.Further,in order to solve the interference problem of the background gas,the absorption spectrum of the air in the filling workshop is detected and analyzed.The feasibility of using the absorption characteristics at2967（88）^-1 as the target absorption peak was verified.（2）In order to solve the problem that the direct absorption spectrum needs to be scanned with zero absorption overlap and the accuracy is insufficient,the wavelength measurement technology is used to transfer the measurement quantity to the frequency conversion interval.Aiming at the limitation of the need for fine tuning of the wavelength modulation,a method based on the absorbance curve Loren was proposed in order to solve the complicated calculation of the harmonic curve.An experimental system model based on Simulink was constructed,and the harmonic characteristics were determined through simulation.Finally,in view of the shortcomings of gas concentration inversion based on second harmonics that rely on calibration gas calibration,the second harmonic normalized by the first harmonic is used as the basis for qualitative judgment,and the dependence on standard gases is avoided.（3）Due to different leak thresholds set appropriately for different aerosol products,a classification algorithm based on aerosol images is proposed.In the stage of implementation,a combination of Otsu segmentation and Hough transform line detection in Lab color space was proposed.Aiming at the problem that the product angles in the original image are different and may be incomplete,a Densenet deep learning network is established,which is realized by the classification of two types of aerosol products.Automatic switching of detection thresholds.（4）A gas concentration detection optical module was constructed,and an inter-band cascade laser was picked up as a mid-infrared light source,which was combined with a mercury-mercury mid-infrared detector to collect light intensity.At the same time,in order to increase the accuracy of the cumulative absorption,a long optical path absorption cell is used to reflect the light beam multiple times.In order to meet the needs of mid-infrared optical path adjustment,two sets of switching systems based on space light and optical fiber are designed and constructed,and the structure is also suitable for optical path debugging of maintenance personnel.Further build a system prototype that includes screening air path and product counting function,tested the tank with supplementary water bath method to determine the leak,and achieved a correct rate of not less than 85%,indicating the feasibility of aerosol leak detection based on spectroscopy.