The rapid growth of neutron flux has driven the development of^(3)He-free neutron detectors to satisfy the requirements of the neutron scattering instruments under construction or planned at the China Spallation Neutr...The rapid growth of neutron flux has driven the development of^(3)He-free neutron detectors to satisfy the requirements of the neutron scattering instruments under construction or planned at the China Spallation Neutron Source(CSNS).Position-sensitive neutron detectors with a high counting rate and large area play an important role in the instruments performing neutron measurements in or close to the direct beam.The ceramic gas-electron-multiplier(GEM)detector serves as a promising solution,and considerable work has been done using the small-area GEM neutron detectors.In this article,we designed and constructed a detector prototype utilizing ceramic GEM foils with an effective area of about307 mm×307 mm.To evaluate and investigate their basic characteristics,the Monte Carlo(MC)tool FLUKA was employed and several neutron beam tests were conducted at CSNS.The simulated spatial resolution was basically in agreement with the measured value of 2.50±0.01 mm(FWHM).The wavelength spectra measurement was verified through comparisons with a commercial beam monitor.In addition,a detection efficiency of 4.7±0.1%was achieved for monoenergetic neutrons of 1.59 A wavelength.This is consistent with the simulated result.The results indicate that the large-area ceramic GEM detector is a good candidate to implement neutron beam measurements.Its efficiency can be improved in a cascading manner to approach that reached by traditional^(3)He detectors.展开更多
全聚焦算法依靠信号的幅度信息进行延迟叠加(delay and sum,DAS)成像,实际应用中信号并非总能满足相干叠加这一前提,而非相干信号的叠加导致噪声和伪影。文章提出一种循环相干因子(circular coherence factor,CCF)加权的延迟乘和(delay ...全聚焦算法依靠信号的幅度信息进行延迟叠加(delay and sum,DAS)成像,实际应用中信号并非总能满足相干叠加这一前提,而非相干信号的叠加导致噪声和伪影。文章提出一种循环相干因子(circular coherence factor,CCF)加权的延迟乘和(delay multiply and sum,DMAS)CCF-DMAS优化算法,实现薄板中缺陷的兰姆波全聚焦成像。该方法考虑接收阵元间的空间相干性,对接收信号进行相乘耦合,利用数据中的相位信息计算相干因子实现自适应加权,以扩大相干和非相干信号间的差异,从而达到缩窄主瓣,减少旁瓣,提高成像分辨率的效果。建立超声阵列发射、接收实验系统,通过楔块耦合,在含通孔缺陷的锆合金薄板上激发S_(0)模态兰姆波,捕获全矩阵数据;通过CCF-DMAS算法对采集的数据相位加权,生成新的频率分量;利用带通滤波保留二次谐波分量进行全聚焦成像。实验结果表明:与DAS和DMAS全聚焦成像算法相比,CCF-DMAS全聚焦优化算法能够有效抑制噪声和伪影,信噪比提高约39%和22%,阵列性能指数提高约86%和69%,为薄板无损检测的后处理提供了一种有效的改进方案。展开更多
交替方向乘子法(Alternating Direction Method of Multiplier,ADMM)因具有线性规划(Linear Programming,LP)译码条件约束的几何结构,同时利用了消息传递机制,被认为是一种第5代移动通信技术(5th Generation Mobile Communication Techn...交替方向乘子法(Alternating Direction Method of Multiplier,ADMM)因具有线性规划(Linear Programming,LP)译码条件约束的几何结构,同时利用了消息传递机制,被认为是一种第5代移动通信技术(5th Generation Mobile Communication Technology,5G)低密度校验(Low Density Parity Check,LDPC)码新型优化译码算法。通过在LP译码模型的目标函数中引入惩罚项,基于ADMM的变量节点惩罚译码有效地减轻了非积分解,从而提高了误帧率(Frame Error Rate,FER)性能。尽管ADMM在许多实际应用中表现出色,其收敛速度较慢以及对初始条件和参数设置敏感的问题仍然限制了其在高维、实时性要求高的场景中的进一步应用。特别是在LDPC线性规划译码过程中,ADMM的交替更新机制容易导致优化路径振荡,且在处理非精确约束时表现不佳。针对ADMM算法收敛速度慢的问题,我们提出了一种新的优化算法,该算法将Nesterov动量加速方法与ADMM相结合,以解决ADMM对LDPC译码器错误修正能力和收敛效率的影响。算法通过动量项减少迭代次数将一个Nesterov加速格式从无约束复合优化问题推广到ADMM惩罚函数模型,利用ADMM算法将原问题的约束条件有效转化为目标函数的一部分,从而构造出无约束优化子问题;在此基础上,进一步采用Nesterov加速技术对梯度下降迭代过程进行改进,以提高收敛速度和求解精度。仿真实验使用了三种不同码率的5G LDPC短码。结果表明,相对于现有ADMM惩罚译码算法,所提出的基于动量加速的ADMM译码算法不仅有大约0.2 dB的信噪比增益,而且平均迭代次数也降低了20%左右,加快了收敛速度。展开更多
基金Project supported by the National Key R&D Program of China(Grant No.2023YFC2206502)the National Natural Science Foundation of China(Grant Nos.12175254 and 12227810)+1 种基金Guangdong Major Project of Basic and Applied Basic Research(Grant No.2023B0303000003)Guangdong Provincial Key Laboratory of Advanced Particle Detection Technology(Grant No.2024B1212010005)。
文摘The rapid growth of neutron flux has driven the development of^(3)He-free neutron detectors to satisfy the requirements of the neutron scattering instruments under construction or planned at the China Spallation Neutron Source(CSNS).Position-sensitive neutron detectors with a high counting rate and large area play an important role in the instruments performing neutron measurements in or close to the direct beam.The ceramic gas-electron-multiplier(GEM)detector serves as a promising solution,and considerable work has been done using the small-area GEM neutron detectors.In this article,we designed and constructed a detector prototype utilizing ceramic GEM foils with an effective area of about307 mm×307 mm.To evaluate and investigate their basic characteristics,the Monte Carlo(MC)tool FLUKA was employed and several neutron beam tests were conducted at CSNS.The simulated spatial resolution was basically in agreement with the measured value of 2.50±0.01 mm(FWHM).The wavelength spectra measurement was verified through comparisons with a commercial beam monitor.In addition,a detection efficiency of 4.7±0.1%was achieved for monoenergetic neutrons of 1.59 A wavelength.This is consistent with the simulated result.The results indicate that the large-area ceramic GEM detector is a good candidate to implement neutron beam measurements.Its efficiency can be improved in a cascading manner to approach that reached by traditional^(3)He detectors.
文摘全聚焦算法依靠信号的幅度信息进行延迟叠加(delay and sum,DAS)成像,实际应用中信号并非总能满足相干叠加这一前提,而非相干信号的叠加导致噪声和伪影。文章提出一种循环相干因子(circular coherence factor,CCF)加权的延迟乘和(delay multiply and sum,DMAS)CCF-DMAS优化算法,实现薄板中缺陷的兰姆波全聚焦成像。该方法考虑接收阵元间的空间相干性,对接收信号进行相乘耦合,利用数据中的相位信息计算相干因子实现自适应加权,以扩大相干和非相干信号间的差异,从而达到缩窄主瓣,减少旁瓣,提高成像分辨率的效果。建立超声阵列发射、接收实验系统,通过楔块耦合,在含通孔缺陷的锆合金薄板上激发S_(0)模态兰姆波,捕获全矩阵数据;通过CCF-DMAS算法对采集的数据相位加权,生成新的频率分量;利用带通滤波保留二次谐波分量进行全聚焦成像。实验结果表明:与DAS和DMAS全聚焦成像算法相比,CCF-DMAS全聚焦优化算法能够有效抑制噪声和伪影,信噪比提高约39%和22%,阵列性能指数提高约86%和69%,为薄板无损检测的后处理提供了一种有效的改进方案。
文摘交替方向乘子法(Alternating Direction Method of Multiplier,ADMM)因具有线性规划(Linear Programming,LP)译码条件约束的几何结构,同时利用了消息传递机制,被认为是一种第5代移动通信技术(5th Generation Mobile Communication Technology,5G)低密度校验(Low Density Parity Check,LDPC)码新型优化译码算法。通过在LP译码模型的目标函数中引入惩罚项,基于ADMM的变量节点惩罚译码有效地减轻了非积分解,从而提高了误帧率(Frame Error Rate,FER)性能。尽管ADMM在许多实际应用中表现出色,其收敛速度较慢以及对初始条件和参数设置敏感的问题仍然限制了其在高维、实时性要求高的场景中的进一步应用。特别是在LDPC线性规划译码过程中,ADMM的交替更新机制容易导致优化路径振荡,且在处理非精确约束时表现不佳。针对ADMM算法收敛速度慢的问题,我们提出了一种新的优化算法,该算法将Nesterov动量加速方法与ADMM相结合,以解决ADMM对LDPC译码器错误修正能力和收敛效率的影响。算法通过动量项减少迭代次数将一个Nesterov加速格式从无约束复合优化问题推广到ADMM惩罚函数模型,利用ADMM算法将原问题的约束条件有效转化为目标函数的一部分,从而构造出无约束优化子问题;在此基础上,进一步采用Nesterov加速技术对梯度下降迭代过程进行改进,以提高收敛速度和求解精度。仿真实验使用了三种不同码率的5G LDPC短码。结果表明,相对于现有ADMM惩罚译码算法,所提出的基于动量加速的ADMM译码算法不仅有大约0.2 dB的信噪比增益,而且平均迭代次数也降低了20%左右,加快了收敛速度。
