Research Of Keytechnologies On Nuclear Signal Digital Processing

The application of nuclear technology has penetrated into many fields of social development,nuclear science and technology brings a lot of benefits,but human beings are often threatened by its application.It is important to collect and analyze nuclear information accurately and effectively,which has been the precondition of the application of nuclear technology.The digital processing of nuclear spectrometer mainly includes the digital collection of the nuclear signal,the digital processing of the nuclear information and the analysis of the spectrum.Traditional signal acquisition and processing often build analog circuits to achieve the functions of filtering,pulse shaping,baseline recovery and peak detection.Limited by the bad processing power of analog circuits,it can't meet the actual needs in many aspects of signal processing(such as filter shaping).Therefore,the improvement of nuclear signal processing has been an important research topic in the field of nuclear measurement.With the rapid development of high speed,high resolution ADC and digital programmable devices(FPGA,DSP),digital nuclear spectrum measurement system has become the trend of the new nuclear instrument.The digital processing of nuclear signal is the key to the realization of digital nuclear spectrum measurement system.Compared with analog shaping,digital pulse shaping has the characteristics of simple and fast algorithm processing,which can reduce the impact of electronic interference and ballistic loss,so as to effectively provide the energy and time resolution of digital nuclear spectrometer.The selection of shaping parameters and the cascading of filter structure have great influence on the forming results.How to get better filtering performance is a difficult problem to be solved in digital processing.The traditional energy spectrum analysis mainly uses the semi empirical mathematical function to fit spectrum,the mathematical formula is complex and the meaning of physical parameters is fuzzy,which is difficult to solve the complex nonlinear problem.How to effectively solve the complex nonlinear problem and to realize the decomposition of heavy peak is also the direction of the research.In this paper,on the basis of the existing research on digital nuclear signal in the process of digital nuclear signal acquisition,processing and spectrum analysis technology.We did the research of nuclear data acquisition based on FPGA,digital signal processing of pulse shaping and energy spectrum analysis based on BP neural networkits.The main contents are as follows:(1)Complete the hardware design of the digital acquisition circuit of the nuclear signal based on FPGA.Pulse directly connected to the high speed ADC to finish sampling work after pre signal conditioning circuit.The FPGA chip of Xilinx is the controller to complete nuclear signal acquisition and digital signal processing.(2)Complete the design and implementation of the algorithm in the FPGA chip for nuclear signal processing.By means of the system modeling tool of system generator,we finished the two level S-K Gauss filter design,trapezoidal(triangle)shaping filter design and CIC digital filter design.In addition,the algorithm and the logic design of the control module was realized in FPGA,Include ADC control module,digital filter module,pulse shaping module,pulse peak-seeking module,USB transmission module and clock configuration module design.(3)The BP neural network model was built to extract the relevant spectrum characteristic parameters,which can be applicatedin spectral line fitting and element judgment,and verified the feasibility of artificial neural network in spectrum analysisThrough the research on the key technology of the nuclear signal digital processing,the main achievements are as follows?(1)In the process of digital signal processing,digital Gauss filter and trapezoidal(triangle)filter are used in pulse shaping.In order to solve undershoot of one stage S-K filter in the processing of nuclear pulse Gaussian shaping.A two stage series S-K filter simulation model is established based on the mathematical expression between input and output,and the optimal shaping factors were confirmed by the influence they have on shaped waveforms.In order to reduce the design complexity and to prevent overflow data,trapezoidal shaping filter is decomposed into four sub modules,by adjusting the different shaping parameters values,we can achieve corresponding trapezoid triangle shaping waveform.(2)In order to solve the nonlinear problem of the complex spectrum,the nonlinear spectrum analysis method was established based on BP neural network.It can effectively solve the problem of fuzzy parameters on empirical function fitting.At the same time,it is proved that the BP network is feasible to distinguish the type of spectrum elements.