粒子探测器的高速数字化脉冲形状甄别电路及基于FPGA逻辑的实时甄别算法研究

Pluse shape discrimination (PSD) has been widely used in radiation detection systems, including nuclear equipments, facilities of paricle physics experiments, medical diagnostic imaging cameras, and so on. The function of PSD is to differentiate signals based on different pulses shapes, which depends on the time response of the detector or the characteristics of the incident paticle. It can be used to distinguish the species of incident particles, or to obtain the spatial/time information of a radiation event...Many analog-based traditional methods have been invented, which can be generally summarized to two major categories: the double integration method and the shaping-discriminator method. The performance of both methods are restricted by the long deadtime, low discrimination reliability, etc.. The digital PSD method based on waveform sampling technique, by contrast, can maxiumly extract the original information from the detector signals, which can utilize many flexible algorithms and results in a better discrimination effect...Although digital PSD has become a new trend in nuclear/particle experiments, it still needs some improvments. Firstly the sampling rates of exsisting digital PSD are not adequate, which degrade the discrimination effect for rather high speed signals (e. g. with a transient time of about several nanoseconds); secondly the discrimination processes are mainly rely on programs running in DSP (Digital Signal Processor) or personal computer, which limits the discrimination throughput...In this project, a new digital PSD method is proposed, which combining the achievements of high speed analog-to-digital conversion (with 1 GSPS sampling rate and 12-bit resolution) and the advantages of high performance FPGA (Field Programmable Gate Array). Utilizing the DWT (Discrete Wavelet Transform), the digital double integration, the weighting integration, FFT (Fast Fourier Transform) and other signal proscess algorithms in communication engineering and multi-media area, real-time PSD for ultra-fast pulses will be excecuted by large amount of parallel logic cells in FPGA, which helps achieve a greatest throughput(above 100 KHz).

脉冲形状甄别（PSD）技术在核与粒子物理、核医学等领域应用广泛，常被用来区分探测器或入射粒子的性质、及获得粒子与探测器相互作用的空间/时间信息等。. 传统PSD主要采用双积分和成形甄别等模拟路线实现，具有死时间长、甄别效率低等缺点。基于波形采样技术的数字化PSD避免了模拟积分的死时间,直接利用原始波形中的信息，更灵活、高效。但现有的数字化PSD一方面采样率和集成度还有待进一步提高，另一方面处理算法大多依赖于PC机或DSP程序串行实现，难以适应纳秒级快脉冲信号的数字化和高计数率情况下的实时甄别。. 本课题拟将高速波形采样技术（采样率1 GS/s，分辨率12位）和高性能FPGA的并行处理技术相结合，进行快脉冲的PSD读出电路方案研究。并借鉴已在通信及多媒体领域获得成功应用的离散小波变换等多种信号处理技术，开展基于数字逻辑的高速、高计数率数字化脉冲形状甄别算法研究。

脉冲形状甄别（Pulse Shape Discrimination,简称PSD）是核电子学领域的一项传统技术，被广泛应用到核物理、粒子物理和影像医学等诸多领域。.传统的PSD 主要采用双积分或成形甄别等模拟的技术路线，但具有死时间长、甄别效率低、难以适应高计数率等缺点。相比之下，数字化PSD既充分利用探测器原始信号波形中所包含的物理信息，提高了甄别效率，同时减小了死时间、避免了信号堆积，近年来获得了迅速发展。.但近年来的数字化PSD应用采样率达1GS/s量级的尚不多见，无法适应纳秒级快脉冲的波形采样要求。其次即使有个别高采样率的应用，也主要是用示波器或商业AD采集卡将原始波形采样并存储到计算机中，然后利用CPU或DSP进行计算分析，在高计数率的情况下难以胜任实时处理，同时给数据传输和存储也带来很大压力。.针对这一现状，本项目着眼于核探测器的高速数字化脉冲形状甄别（PSD）方法的创新研究。通过利用当前高速采样技术的新成果，及大规模FPGA 逻辑的高速并行处理优势，并借鉴吸收已有的多种处理算法，将其利用FPGA实现，以突破传统PSD 路线的性能瓶颈，使之适应于高速、实时、高计数率的应用。.本项目主要完成了计划中的三方面的研究内容：.一、.基于高速波形采样技术的新型数字化PSD 读出电路研究.针对多通道、高集成度、但计数率稍低的应用方向，本项目采用了瑞士PSI（Paul Scherrer Institute）设计的开关电容阵列芯片DRS4来实现高速波形采集卡，实现低计数率情况下的2 GS/s快脉冲波形采样。.针对高计数率、但通道数较少的应用，本项目实现了基于高速ADC的1 GS/s 12bit波形采集卡，并在FPGA中实现了实时处理。.二、.新型数字化PSD算法研究和对比.本项目中，对数值双积分，以及面积-峰值法、峰值-梯度法、部分面积法、离散小波变换（DWT）、人工神经网络等几种算法进行了研究和比较。.三、.高速并行数字PSD 算法的FPGA 实现.在算法研究的基础上，本项目将多种效果显著、且适合于数字逻辑实现的算法集成到了FPGA中。本项目专门搭建了一个BaF2闪烁体探测器，并利用BaF2晶体对α、γ的响应脉冲波形不同，证明了甄别方法的有效性。.本项目现已发表会议论文1篇，已正式接收的核心期刊论文1篇。另有多篇论文正在撰写和提交，将于2016年陆续发表。