Application And Implementation Of Blind Source Separation Algorithm In Flammable Liquid Detection Under Noisy Background

In recent years,with the deepening of economic globalization and the continuous changes in the current international environment and situation,the safety issues in public places have become increasingly prominent.In order to ensure the safety of people's lives and property in public places,the state has invested a lot of manpower and material resources in public safety to prevent the occurrence of public safety accidents.At present,the most commonly used security equipment in public place security inspection is mainly based on the principle of X-ray imaging to obtain information on passengers' carry.The relevant detection technology for flammable liquids is not yet mature,and most security inspection sites still use odor detection and tasting.The inspection method is not only cumbersome but also inconvenient for people to travel.Faced with this situation,if a safe and efficient security inspection method can be proposed for flammable liquid detection,this not only can further protect public safety,but also an important supplement to the existing security inspection technology,so it has high research significance and value.Therefore,from the perspective of signal processing,this paper takes flammable liquid detection as the research object and combines with the blind source separation algorithm,and proposes to apply the blind source separation algorithm to flammable liquid detection under common noise background.It provides new ideas and methods for how to detect flammable liquids safely and efficiently.In order to apply blind source separation algorithm in the detection of flammable liquids in the background of noise,this paper combined with previous studies designed a beam focusing system to collect the S parameters of various flammable liquid samples,and established the corresponding Sample S-parameter database.Then,under the common noise background,the observed signal is processed by the blind source separation algorithm to separate the source signal from the observed signal,and then the separated signal is matched with the sample database to realize the judgment of the type of flammable liquid.The main research contents are as follows:(1)First,the principle of ultra-wideband centimeter wave detection of flammable liquids was studied.A beam focusing system that meets the experimental requirements was designed based on practical and theoretical requirements.The antenna and lens of the beam focusing system were designed,and The focusing performance and electromagnetic parameters of the system are simulated and verified.Afterwards,the S-parameters of gasoline,alcohol and water in the 8GHz-18 GHz frequency band are measured using a beam focusing system and a vector network analyzer,and the collected S-parameter data are sorted and cleaned to establish a corresponding sample S-parameter database.Finally,the common noise and noisy observation signals are analyzed,and it is proposed that the common noisy observation signals can be processed by the blind source separation method of the noiseless model.(2)Secondly,on the basis of the research on the Fast ICA algorithm based on negative entropy,and for the problem of the slow convergence speed of the basic Fast ICA algorithm,this paper adopts a new fifth-order convergence Newton iteration method,and on this basis,the Fast ICA algorithm based on negative entropy is improved to improve its convergence speed.Then by using the collected data for simulation experiments,the simulation results prove that the observation signals containing common noise can be separated by the blind source separation method of the noiseless model,and the effectiveness of the improved algorithm is also verified.(3)Finally,for the traditional blind source separation algorithm,there are the problems of large calculation amount and the restriction of the convergence speed and the separation effect.In this paper,after studying the blind source separation algorithm under the natural gradient criterion,combined with the idea of variable step size,the natural gradient blind source separation algorithm is improved.This method judges the signal separation stage through the correlation coefficient between the separated signals,and The step factor is controlled according to the separation stage,and the momentum term is introduced into the algorithm.Simulation experiment results show that the improved algorithm can separate the noisy observation signal better,and has better convergence speed and separation performance compared with the traditional algorithm.