Feature Extraction Of ?-ray Fingerprints Based On Adaptive Network-based Fuzzy Inference System

As a kind of radiation fingerprints,?-ray spectra can be used to identify the type or individual of nuclear material.Therefore,?-ray fingerprint identification technology gets more and more extensive application in the field of nuclear security,especially playing an important role in the verification of nuclear weapons.The extraction of ?-ray fingerprint feature is a significant technology part in process of ?-ray fingerprint identification,and at the same time,it is a crucial technical method to shield the relevant sensitive information.So studying on the extraction of ?-ray fingerprint characteristic possess vital practical significance.In this thesis,the method of taking Adaptive Network-based Fuzzy Inference System(ANFIS)to extract ?-ray fingerprint feature for the second time has been proposed.Firstly,set up first class ANFIS and simulate ?-ray fingerprint.Extract consequent parameters of ANFIS to establish the training sample set.Then train the second class ANFIS by the consequent parameter training sample set of first class ANFIS,and extract the consequent parameters of second class ANFIS as the identification features of ?-ray fingerprint.In order to verify the validity of the method,the following specific research has been done in this thesis:(1)simulate and collect the nuclear material of different type and different individual of same type by241Am?60Co?137Cs?22Na and152 Eu radioactive source and their combination and collect their ?-ray spectra.(2)Extract ?-ray fingerprint features of simulation nuclear materials mentioned above.(3)Identify nuclear material above by means of minimum membership degree principle,including type identification,individual identification and ?-ray fingerprint identification of little difference.The results show that it is feasible to extract the feature of ?-ray fingerprint for second time by two class ANFIS.The identification of type,individual and little difference ?-ray fingerprint for nuclear material are all effective.Meanwhile,it has a good shielding effect on sensitive information in ?-ray fingerprint.In addition,due to the de-noising effect of ANFIS in the process of simulation,it is not necessary to perform the de-noising pretreatment for?-ray fingerprint utilizing ANFIS to extract feature.