Information Restoration Method Of Nuclear-chemical Monitoring

In order to effectively cope with nuclear and chemical attacks or diversified emergencies,it is urgent to develop high performance,multifunctional and low-cost monitoring instrument technology.This study took nuclear and chemical monitoring process as information process.Theoretical derivation,numerical simulation,experimental measurement and other technical means were used.The technical framework of nuclear and chemical information restoration was preliminarily constructed,and the varying mechanism and processing methods of nuclear and chemical monitoring information in these systems were explored deeply.The purpose is to break hardware constraints,improve performance-cost ratio,and achieve deep integration of nuclear and chemical monitoring systems.The main research work and conclusions are summarized as follows:（1）The information model for nuclear and chemical monitoring was established.The monitoring process was divided into two steps:information recording and information restoration.The mechanism of information loss in information recording step was analyzed.The abstract model for information restoration was set up,which involves four factors:system response,background,noise and information restoration algorithms.In addition,the noise characteristics in analog signal processing were analyzed in detail.（2）The system response,background estimation method and noise model were studied.In order to improve the accuracy of system response parameter estimation,the Gaussian Smoothing Least Squares method based on single data and the average data fitting method based on multiple data were proposed.Numerical simulation verifies the effectiveness of the two algorithms.In order to improve the accuracy of background estimation,4 common algorithms were compared,and the best results came from the ar PLS algorithm.As to noise,a nonstationary model was proposed,and the expression of noise covariance and the corresponding optimal amplitude information extraction operator were derived.The model was applied to scintillation detectors,and its reliability was verified by measured pulses from Na I（Tl）and La Br₃（Ce）detectors.The optimal weighting factors were calculated for the weighted gated integral method to extract amplitude information for pulses of different energies.（3）Information restoration algorithm for nuclear and chemical monitoring was studied.For separating the folded response,the linear transform and deconvolution transform methods were compared.The results showed that the deconvolution method has significant advantages compared with the linear method,but the number of iterations would have opposite effects on the false peak effect and the accuracy of the number of responses.A deconvolution preprocessing method was proposed,in which the response matrix and the measured data were smoothed at the same time,and deconvolution is performed subsequently.The validity of this method was verified with simulated data.A new iterative algorithm construction framework was proposed on the basis of the original proportional iterative algorithm for constrained optimization problems.The common R-L algorithm and MAP algorithm were incorporated into the same optimization model,and the effectiveness of this method was verified by simulated and measured spectral data.In terms of the influence of system response estimation accuracy on deconvolution results,the simulated data was used to study the influence of inaccurate peak width estimation on the solution results of different deconvolution algorithms.The Gaussian convolution non-negative least square algorithm showed the best performance.（4）The methods for information processing under typical complex measurement conditions were explored.As regards the temperature effect of Na I（Tl）detectors,a correction method was adopted which consists of temperature estimation,pulse deconvolution,trapezoidal shaping and amplitude correction.This method could satisfy the requirements of temperature correction with high count rates,with which excellent energy resolution of 6.91%@662 ke V was achieved from measured spectra under temperature ranging from-20 to 50℃.As regards the temperature effect of ion mobility spectrometry systems,compound identification under varying temperature was realized based on establishment of the data base of temperature dependent system response,as well as temperature estimation according to the transition time of reactive ion peak.The difficult problem of nuclide recognition with complex geometry and medium condition was also discussed by using 86 different geometry and medium cases with 14 nuclides.A simulated spectrum training set was constructed,and a neural network classifier was used to verify the feasibility of this method.（5）The conceptual design of the integrated nuclear and chemical monitoring system was proposed.In the information recording step,information integrity was ensured with no distortion and low loss,and in the information restoration step,it focused on system response calibration as accurate as possible,obtaining and making full use of prior knowledge,and selection of information restoration algorithms.The hardware of the system was designed with a kernel FPGA chip and replaceable probes,supporting multi-channel parallel signal processing.The feasibility of this scheme was verified by ion mobility spectrometry data acquired with an oscilloscope.The signal-to-noise ratio of 6.28×10³was achieved with sampling frequency of 1.25 MHz,which was 23.6%higher than that of the original spectrometer.Based on function module reuse,the software architecture of integrated nuclear and chemical monitoring system was constructed.Finally,the overall design of cloud-edge collaborative integrated system for nuclear and chemical monitoring was proposed.In this study,the processes of nuclear and chemical monitoring were unified from the perspective of information restoration,and a nuclear and chemical monitoring information model was constructed.A series of methods were proposed to improve the accuracy of system response estimation and deconvolution results.Typical cases of information restoration were explored under complex conditions.A nuclear and chemical integrated technology architecture were designed based on cloud edge collaboration.This study provides a technical solution for realizing the next generation of nuclear and chemical monitoring of high performance and low cost.