Research On Nuclear Signal Digital Processing Method And Application

The superior performance of digital nuclear instruments makes it the main research direction of large number of nuclear electronics scientific researchers.Among them,nuclear signal processing method is one of the key technology in the research of digital nuclear instrument.With the development of microelectronics and computer technology,the digital acquisition of pulse signal is gradually realized by ray energy spectrum measuring instrument.The digital signal processing method and technology development has become a hot technical problem of ray energy spectrum measurement system.The traditional filtering and shaping method are shaping the output pulse signal of the nuclear radiation detector through the analog filter circuit into a Gauss-like waveform or a trapezoidal waveform through a digital circuit or a direct Z transformation.It is mainly constructed by mathematical methods or formed by Ztransform,but the digital construction and Z-transform processing of nuclear signal system do not clearly point out the physical model of signal transformation,which is not conducive to further algorithm optimization.In addition,for complex molding systems,the system does not converge after Z-transform.Based on the analysis of the output signal and noise characteristics of scintillation and semiconductor detector,this paper focuses on a variety of nuclear pulse signal shaping algorithms.The main research contents and results are as follows:First,by comparing the functions in the convolution function table with the signals in nuclear electronics,the digital convolution unit is divided into small units such as impulse signal,step signal and negative exponential signal.It is proposed that when the signal is convoluted with the truncation function,if the delay time of the delayed signal is less than the signal width,the result is a truncated signal,and if the delay time of the delayed signal is greater than or equal to the signal width,the result is a bipolar signal.The relationship between C-R and R-C filter shaping circuits and negative exponential signal is studied.Finally,the convolution transform table is combined with the nuclear electronics model,and the convolution transform table is improved.Secondly,the commonly used trapezoidal shaping in Z-transform shaping is indepth research.According to the different output signals of the detection system,the single exponential trapezoidal shaping and double exponential trapezoidal shaping are studied.It is mainly used to reduce truncation error and avoid the slow processing speed of FPGA caused by floating-point operation.In addition,Z-transform sawtooth shaping(single sawtooth shaping and double sawtooth shaping)is derived,which can effectively suppress baseline drift.Complex models are constructed through each small unit,such as trapezoidal shaping,cusp pulse shaping and flat-top cusp pulse shaping and so on.Thirdly,through convolution method,multiple small units are constructed into complex forming,such as trapezoidal shaping,cusp pulse shaping and flat-top cusp pulse shaping.And in the process of shaping,reverse truncated convolution is proposed.Then these kinds of shaping are combined to form various signals.For example,flat top cusp shaping can be constructed by trapezoidal shaping and flat-top pulse shaping.Z-transform small units such as impulse function,step function and negative exponential function in the pulse forming method.A cascade convolution shaping algorithm based on Z transform is constructed.The physical model of the algorithm is clearer,and the single exponential impulse shaping,double exponential impulse shaping and single exponential trapezoidal shaping are deduced through the algorithm.Finally,the above method is compared and verified by three different detectors:sodium iodide,fast SDD and HPGe.The results are as follows:1)The output signals of three different detectors are used to implement cascade trapezoidal shaping and flat top cusp pulse shaping.The shaping results show that at high counting rate,flat cusp pulse shaping has better separation effect on stacked pulses,but the shaped amplitude will be reduced to a certain extent.For the distorted pulse signal generated by the fast SDD detector,after the flat top pulse shaping,the signal baseline after the distorted pulse will be raised.The distorted pulse needs to be discarded or repaired,otherwise measurement errors will be caused.2)The commonly used Z-transform trapezoidal shaping is used to application.Firstly,the single exponential trapezoidal shaping is discussed.After amplification and noise removal,the signals from the detector are basically exponential signals with rising edges.And the pulse width of the exponential signal with rising edge after the trapezoidal pulse shaping increases with the increase of the pulse amplitude.A pulse width discrimination method varying with pulse amplitude is proposed to eliminate the inseparable pile-up pulses.Finally,the method is used to pile-up pulse continuous zone reject at high counting rate.Secondly,for double exponential trapezoidal shaping,when the nuclear signal with rising edge is shaped by double exponential trapezoidal shaping,the pulse width after shaping remains unchanged.Therefore,after using fixed pulse discrimination reject the pile-up pulse continuous zone formed by pulses that cannot be completely separated,and then,aiming at the sum peak formed by the complete pileup pulses at high counting rate,a formula for calculating the probability of characteristic photoelectric peak is proposed.The probability of sum peak generation is calculated by the method of probability theory.Finally,the sum peak in the energy spectrum measured by different detectors is calculated and verified.