Conceptual design of a 714-MHz RFQ for compact proton injectors and development of a new tuning algorithm on its aluminium prototype

Radio frequency quadrupoles(RFQs),which are crucial components of proton injectors,significantly affect the performance of proton accelerator facilities.An RFQ with a high frequency of 714 MHz dedicated to compact proton injectors for medi-cal applications is designed in this study.The RFQ is designed to accelerate proton beams from 50 keV to 4 MeV within a short length of 2 m and can be matched closely with the downstream drift tube linac to capture more particles through a preliminary optimization.To develop an advanced RFQ,challenging techniques,including fabrication and tuning method,must be evaluated and verified using a prototype.An aluminium prototype is derived from the conceptual design of the RFQ and then redesigned to confirm the radio frequency performance,fabrication procedure,and feasibility of the tuning algorithm.Eventually,a new tuning algorithm based on the response matrix and least-squares method is developed,which yields favorable results based on the prototype,i.e.,the errors of the dipole and quadrupole components reduced to a low level after several tuning iterations.Benefiting from the conceptual design and techniques obtained from the prototype,the formal mechanical design of the 2-m RFQ is ready for the next manufacturing step.

基于prompt tuning的中文文本多领域情感分析研究

不同领域的情感文本表达方式不一样,通常需要为各个领域训练相应的情感分析模型。针对无法用一个模型进行高效多领域情感分析的问题,提出了基于提示微调(prompt tuning)的多领域文本情感分析方法MSAPT。借助hard prompt,指示情感文本的所属领域和待选的情感标签,调动不同领域情感分析相关的知识,再为情感分析预训练一个统一的“通才模型”,在下游的各领域文本学习中,保持模型冻结,通过prompt tuning使模型学习到下游各领域情感文本的特征。MSAPT仅需保存一个模型和一些参数量远远小于模型的prompt,实现了多领域情感分析。在多个属于不同领域的情感文本数据集上进行实验,结果表明仅进行prompt tuning时,MSAPT效果优于模型微调(model tuning)的。最后,分别对适应特定领域的prompt tuning、hard prompt、soft prompt的长度和中间训练数据集的大小进行消融实验,从证明其对情感分析效果的影响。

A highly sensitive LITES sensor based on a multi-pass cell with dense spot pattern and a novel quartz tuning fork with low frequency

A highly sensitive light-induced thermoelectric spectroscopy(LITES)sensor based on a multi-pass cell(MPC)with dense spot pattern and a novel quartz tuning fork(QTF)with low resonance frequency is reported in this manuscript.An erbi-um-doped fiber amplifier(EDFA)was employed to amplify the output optical power so that the signal level was further enhanced.The optical path length(OPL)and the ratio of optical path length to volume(RLV)of the MPC is 37.7 m and 13.8 cm^(-2),respectively.A commercial QTF and a self-designed trapezoidal-tip QTF with low frequency of 9461.83 Hz were used as the detectors of the sensor,respectively.The target gas selected to test the performance of the system was acetylene(C2H2).When the optical power was constant at 1000 mW,the minimum detection limit(MDL)of the C2H2-LITES sensor can be achieved 48.3 ppb when using the commercial QTF and 24.6 ppb when using the trapezoid-al-tip QTF.An improvement of the detection performance by a factor of 1.96 was achieved after replacing the commer-cial QTF with the trapezoidal-tip QTF.

Vibration attenuation performance of wind turbine tower using a prestressed tuned mass damper under seismic excitation

With the rapid development of large megawatt wind turbines,the operation environment of wind turbine towers(WTTs)has become increasingly complex.In particular,seismic excitation can create a resonance response and cause excessive vibration of the WTT.To investigate the vibration attenuation performance of the WTT under seismic excitations,a novel passive vibration control device,called a prestressed tuned mass damper(PS-TMD),is presented in this study.First,a mathematical model is established based on structural dynamics under seismic excitation.Then,the mathematical analytical expression of the dynamic coefficient is deduced,and the parameter design method is obtained by system tuning optimization.Next,based on a theoretical analysis and parameter design,the numerical results showed that the PS-TMD was able to effectively mitigate the resonance under the harmonic basal acceleration.Finally,the time-history analysis method is used to verify the effectiveness of the traditional pendulum tuned mass damper(PTMD)and the novel PS-TMD device,and the results indicate that the vibration attenuation performance of the PS-TMD is better than the PTMD.In addition,the PS-TMD avoids the nonlinear effect due to the large oscillation angle,and has the potential to dissipate hysteretic energy under seismic excitation.

Multiscale and Auto-Tuned Semi-Supervised Deep Subspace Clustering and Its Application in Brain Tumor Clustering

In this paper,we introduce a novel Multi-scale and Auto-tuned Semi-supervised Deep Subspace Clustering(MAS-DSC)algorithm,aimed at addressing the challenges of deep subspace clustering in high-dimensional real-world data,particularly in the field of medical imaging.Traditional deep subspace clustering algorithms,which are mostly unsupervised,are limited in their ability to effectively utilize the inherent prior knowledge in medical images.Our MAS-DSC algorithm incorporates a semi-supervised learning framework that uses a small amount of labeled data to guide the clustering process,thereby enhancing the discriminative power of the feature representations.Additionally,the multi-scale feature extraction mechanism is designed to adapt to the complexity of medical imaging data,resulting in more accurate clustering performance.To address the difficulty of hyperparameter selection in deep subspace clustering,this paper employs a Bayesian optimization algorithm for adaptive tuning of hyperparameters related to subspace clustering,prior knowledge constraints,and model loss weights.Extensive experiments on standard clustering datasets,including ORL,Coil20,and Coil100,validate the effectiveness of the MAS-DSC algorithm.The results show that with its multi-scale network structure and Bayesian hyperparameter optimization,MAS-DSC achieves excellent clustering results on these datasets.Furthermore,tests on a brain tumor dataset demonstrate the robustness of the algorithm and its ability to leverage prior knowledge for efficient feature extraction and enhanced clustering performance within a semi-supervised learning framework.

Parameters Optimization and Performance Evaluation of the Tuned Inerter Damper for the Seismic Protection of Adjacent Building Structures

In order to improve the seismic performance of adjacent buildings,two types of tuned inerter damper(TID)damping systems for adjacent buildings are proposed,which are composed of springs,inerter devices and dampers in serial or in parallel.The dynamic equations of TID adjacent building damping systems were derived,and the H2 norm criterion was used to optimize and adjust them,so that the system had the optimum damping performance under white noise random excitation.Taking TID frequency ratio and damping ratio as optimization parameters,the optimum analytical solutions of the displacement frequency response of the undamped structure under white noise excitation were obtained.The results showed that compared with the classic TMD,TID could obtain a better damping effect in the adjacent buildings.Comparing the TIDs composed of serial or parallel,it was found that the parallel TIDs had more significant advantages in controlling the peak displacement frequency response,while the H2 norm of the displacement frequency response of the damping system under the coupling of serial TID was smaller.Taking the adjacent building composed of two ten-story frame structures as an example,the displacement and energy collection time history analysis of the adjacent building coupled with the optimum design parameter TIDs were carried out.It was found that TID had a better damping effect in the full-time range compared with the classic TMD.This paper also studied the potential power of TID in adjacent buildings,which can be converted into available power resources during earthquakes.

Morphological disruption and visual tuning alterations in the primary visual cortex in glaucoma(DBA/2J)mice

Glaucoma is a leading cause of irreve rsible blindness wo rldwide,and previous studies have shown that,in addition to affecting the eyes,it also causes abnormalities in the brain.However,it is not yet clear how the primary visual cortex(V1)is altered in glaucoma.This study used DBA/2J mice as a model for spontaneous secondary glaucoma.The aim of the study was to compare the electrophysiological and histomorphological chara cteristics of neurons in the V1between 9-month-old DBA/2J mice and age-matched C57BL/6J mice.We conducted single-unit recordings in the V1 of light-anesthetized mice to measure the visually induced responses,including single-unit spiking and gamma band oscillations.The morphology of layerⅡ/Ⅲneurons was determined by neuronal nuclear antigen staining and Nissl staining of brain tissue sections.Eighty-seven neurons from eight DBA/2J mice and eighty-one neurons from eight C57BL/6J mice were examined.Compared with the C57BL/6J group,V1 neurons in the DBA/2J group exhibited weaker visual tuning and impaired spatial summation.Moreove r,fewer neuro ns were observed in the V1 of DBA/2J mice compared with C57BL/6J mice.These findings suggest that DBA/2J mice have fewer neurons in the VI compared with C57BL/6J mice,and that these neurons have impaired visual tuning.Our findings provide a better understanding of the pathological changes that occur in V1 neuron function and morphology in the DBA/2J mouse model.This study might offer some innovative perspectives regarding the treatment of glaucoma.

Classification of Conversational Sentences Using an Ensemble Pre-Trained Language Model with the Fine-Tuned Parameter

Sentence classification is the process of categorizing a sentence based on the context of the sentence.Sentence categorization requires more semantic highlights than other tasks,such as dependence parsing,which requires more syntactic elements.Most existing strategies focus on the general semantics of a conversation without involving the context of the sentence,recognizing the progress and comparing impacts.An ensemble pre-trained language model was taken up here to classify the conversation sentences from the conversation corpus.The conversational sentences are classified into four categories:information,question,directive,and commission.These classification label sequences are for analyzing the conversation progress and predicting the pecking order of the conversation.Ensemble of Bidirectional Encoder for Representation of Transformer(BERT),Robustly Optimized BERT pretraining Approach(RoBERTa),Generative Pre-Trained Transformer(GPT),DistilBERT and Generalized Autoregressive Pretraining for Language Understanding(XLNet)models are trained on conversation corpus with hyperparameters.Hyperparameter tuning approach is carried out for better performance on sentence classification.This Ensemble of Pre-trained Language Models with a Hyperparameter Tuning(EPLM-HT)system is trained on an annotated conversation dataset.The proposed approach outperformed compared to the base BERT,GPT,DistilBERT and XLNet transformer models.The proposed ensemble model with the fine-tuned parameters achieved an F1_score of 0.88.

Optimizing Enterprise Conversational AI: Accelerating Response Accuracy with Custom Dataset Fine-Tuning

As the realm of enterprise-level conversational AI continues to evolve, it becomes evident that while generalized Large Language Models (LLMs) like GPT-3.5 bring remarkable capabilities, they also bring forth formidable challenges. These models, honed on vast and diverse datasets, have undoubtedly pushed the boundaries of natural language understanding and generation. However, they often stumble when faced with the intricate demands of nuanced enterprise applications. This research advocates for a strategic paradigm shift, urging enterprises to embrace a fine-tuning approach as a means to optimize conversational AI. While generalized LLMs are linguistic marvels, their inability to cater to the specific needs of businesses across various industries poses a critical challenge. This strategic shift involves empowering enterprises to seamlessly integrate their own datasets into LLMs, a process that extends beyond linguistic enhancement. The core concept of this approach centers on customization, enabling businesses to fine-tune the AI’s functionality to fit precisely within their unique business landscapes. By immersing the LLM in industry-specific documents, customer interaction records, internal reports, and regulatory guidelines, the AI transcends its generic capabilities to become a sophisticated conversational partner aligned with the intricacies of the enterprise’s domain. The transformative potential of this fine-tuning approach cannot be overstated. It enables a transition from a universal AI solution to a highly customizable tool. The AI evolves from being a linguistic powerhouse to a contextually aware, industry-savvy assistant. As a result, it not only responds with linguistic accuracy but also with depth, relevance, and resonance, significantly elevating user experiences and operational efficiency. In the subsequent sections, this paper delves into the intricacies of fine-tuning, exploring the multifaceted challenges and abundant opportunities it presents. It addresses the technical intricacies of data integration, ethical considerations surrounding data usage, and the broader implications for the future of enterprise AI. The journey embarked upon in this research holds the potential to redefine the role of conversational AI in enterprises, ushering in an era where AI becomes a dynamic, deeply relevant, and highly effective tool, empowering businesses to excel in an ever-evolving digital landscape.

Improvement of main drift chamber Monte-Carlo tuning model at BESⅢ

Based on real data, a new parameterized model of the main drift chamber response is proposed. In this model, we tune the ratio of good hits and the residual distribution separately. By data quality checking, the diference between simulation and data in track reconstruction efciency reduces from 1% to 0.5% averagely for the pion in J/ψ→π＋π π0, and the momentum resolution agreement improves significantly for the proton in J/ψ→ppˉ.

Fe-Cr二元合金微观组织演化的质量密度场耦合动力学Monte-Carlo模拟研究

本文建立了一种全新的将动力学Monte-Carlo粒子模拟与基于归一化Gauss函数基组的质量密度场空间粗粒化模型耦合的杂化模拟算法.采用该杂化模拟算法,系统对比研究了4种Cr原子含量分别为12.8%,20.0%,30.0%和40.0%的Fe-Cr合金中Cr相在温度为673 K下的时效析出动力学机制,及其时效不同阶段微观组织形貌的演变规律.研究得出Fe-Cr(12.8%)合金富Cr相时效组织形貌呈现孤立颗粒状空间分布形态,时效机制属于形核-长大(NG)机制;对于Fe-Cr(30.0%)和Fe-Cr(40.0%),富Cr相时效形貌在形核-生长及熟化阶段均呈现为三维蠕虫状空间分布特征,时效机制属于条幅分解(SD)机制;对于Fe-Cr(20.0%)合金,其富Cr相组织演化特征介于NG和SD机制之间.研究进一步发现Cr原子短程序参量可用来分析富Cr相形核-生长阶段Fe-Cr合金原子尺度结构的演变,但对于时效熟化阶段微观结构组织变化不敏感.基于空间粗粒化后Fe-Cr合金微观组织形貌,进一步分析了4种Cr原子含量下Fe-Cr合金相变动力学参数如富Cr相体积分数、平均粒径及相颗粒数密度随时效时间演变.本文建立的质量密度场耦合动力学Monte-Carlo模拟方法,为开发多尺度算法模拟合金时效动力学机制及微观组织形貌演变提供了新的思路和研究基础.

基于Monte-Carlo迭代求解策略的局部社区发现算法

针对现有的局部社区发现算法因采用贪心策略进行社区扩张而导致的过早收敛和查全率低的问题,提出一种基于Monte-Carlo迭代求解策略的局部社区发现算法。首先,在每轮迭代的社区扩张阶段,根据节点对社区紧密度增益的贡献比例为所有邻接候选节点赋予选择概率,并结合此概率,再随机选择一个节点加入社区。然后,为避免随机选择导致扩张方向偏离目标社区,根据社区质量变化情况判断本轮迭代中是否触发节点淘汰机制。若触发,计算各个已加入社区节点与社区内其他节点的相似度和,根据相似度和的倒数赋予淘汰概率,并结合此概率,再随机淘汰一个节点。最后,在给定数量的最近迭代轮次中,根据社区规模是否增加判断是否继续迭代。在三个真实的网络数据集上进行实验,相较于局部紧密度扩展(LTE)算法、Clauset算法、加权共同邻居节点(CNWNN)算法和模糊相似关系(FSR)算法,所提算法的局部社区发现结果的F-score值分别提升了32.75、17.31、20.66和25.51个百分点,且能够有效避免查询节点在社区中所处位置对局部社区发现结果的影响。

具有下界约束混料试验的Monte-Carlo渐近最优设计

在具有下界约束的混料试验设计中,由于难以求得含有未知变量的信息矩阵的行列式的显式表达,所以难以得到各类最优准则下的精确最优设计解。文章使用Monte-Carlo方法构造随机试验点的样本,进而可以通过模拟得出D-准则和MV-准则下的渐近最优设计。

Application of artificial neural networks in optimal tuning of tuned mass dampers implemented in high-rise buildings subjected to wind load

High-rise buildings are usually considered as flexible structures with low inherent damping. Therefore, these kinds of buildings are susceptible to wind-induced vibration. Tuned Mass Damper（TMD） can be used as an effective device to mitigate excessive vibrations. In this study, Artificial Neural Networks is used to find optimal mechanical properties of TMD for high-rise buildings subjected to wind load. The patterns obtained from structural analysis of different multi degree of freedom（MDF） systems are used for training neural networks. In order to obtain these patterns, structural models of some systems with 10 to 80 degrees-of-freedoms are built in MATLAB/SIMULINK program. Finally, the optimal properties of TMD are determined based on the objective of maximum displacement response reduction. The Auto-Regressive model is used to simulate the wind load. In this way, the uncertainties related to wind loading can be taken into account in neural network’s outputs. After training the neural network, it becomes possible to set the frequency and TMD mass ratio as inputs and get the optimal TMD frequency and damping ratio as outputs. As a case study, a benchmark 76-story office building is considered and the presented procedure is used to obtain optimal characteristics of the TMD for the building.

基于Monte-Carlo模拟的小样本下齿轮疲劳极限计算方法及软件开发

齿轮疲劳极限评估Dixon-Mood(D-M)法的试验样本需求量较大,标准差估计值存在较大偏差,因此基于Monte-Carlo模拟提出小样本的齿轮疲劳极限分析方法(CQUboot),并结合大量弯曲疲劳极限试验对D-M法和CQUboot法进行了对比分析。与GB/T 14230—2021《齿轮弯曲疲劳强度试验方法》推荐的20~22个样本计算结果相比,样本量降至12时,D-M法的最大误差为15.76%~24.99%,CQUboot法为8.02%~12.98%;以12.66%为疲劳极限预估允许的最大误差时,D-M法至少需要10~18个样本点,CQUboot法只需8个样本点。

超高层Tuned Mass Damper防震支撑系统技术研究应用

超高层建筑是现代城市建设的重要标志之一,其高度已经超过了传统建筑的极限。然而,随着建筑高度不断增加,地震的破坏力也越来越强,超高层建筑面临着更加严峻的安全挑战。因此,研究超高层建筑防震支撑系统技术非常重要,Tuned Mass Damper(TMD)是一种被广泛研究和应用的超高层建筑防震支撑系统技术,TMD最初是在20世纪60年代提出的,最早应用于桥梁上,后来,TMD被引入到建筑领域,并得到广泛的应用。通过精确调节质量、阻尼和弹性等参数来削弱地震引起的建筑物减震效应,从而减少了建筑物因地震造成的损害和崩塌的风险.

Axonal tuning by GABA_(A) receptor unveils novel tricks from an old dog

In the last years,axonal conductance of action potential trains became a novel subject of study,changing the view of axons,from a static cable-like compartment to a more complex and dynamic system(Debanne et al.,2011).Axonal computation,indeed,is canonically constituted by the action of voltage-gated ion channels,such as the classic Na+and K+channels,but recent studies demonstrated that it can be modulated by the action of other ion channel pumps,and metabolic factors(Byczkowicz et al.,2019;Zang and Marder.

Development of multi-group Monte-Carlo transport and depletion coupling calculation method and verification with metal-fueled fast reactor

The accurate modeling of depletion,intricately tied to the solution of the neutron transport equation,is crucial for the design,analysis,and licensing of nuclear reactors and their fuel cycles.This paper introduces a novel multi-group Monte-Carlo depletion calculation approach.Multi-group cross-sections(MGXS)are derived from both 3D whole-core model and 2D fuel subassembly model using the continuous-energy Monte-Carlo method.Core calculations employ the multi-group Monte-Carlo method,accommodating both homogeneous and specific local heterogeneous geometries.The proposed method has been validated against the MET-1000 metal-fueled fast reactors,using both the OECD/NEA benchmark and a new refueling benchmark introduced in this paper.Our findings suggest that microscopic MGXS,produced via the Monte-Carlo method,are viable for fast reactor depletion analyses.Furthermore,the locally heterogeneous model with angular-dependent MGXS offers robust predictions for core reactivity,control rod value,sodium void value,Doppler constants,power distribution,and concentration levels.

Adapter与Prompt Tuning微调方法研究综述

文本挖掘是数据挖掘的一个分支学科,涵盖多种技术,其中自然语言处理技术是文本挖掘的核心工具之一,旨在帮助用户从海量数据中获取有用的信息。近年来,预训练模型对自然语言处理的研究和发展有重要的推动作用,预训练模型的微调方法也成为重要的研究领域。根据近年来预训练模型微调方法的相关文献,选择目前主流的Adapter与Prompt微调方法进行介绍。对自然语言处理的发展脉络进行简要梳理,分析目前预训练模型微调存在的问题与不足;介绍Adapter与Prompt两类微调方法,对两个研究方向中经典方法进行介绍,并从优缺点和性能等方面进行详细分析;进行总结归纳,阐述目前预训练模型的微调方法存在的局限性并讨论未来发展方向。

Monte-Carlo模拟在土壤重金属生态风险评价中的应用

离散检测数据的重金属生态风险评估往往受采样数据空间覆盖度有限、检测数据随机波动等问题影响,使评价结果具有较大的不确定性.相对而言,基于地方区域的重金属概率统计以概率的形式解释污染等级的占比,能更有效指导风控方案的制定、污染的修复和场地的管控.本文应用Monte-Carlo模拟结合Hakanson潜在生态风险指数法,对云南省内某小流域土壤中的8种重金属(Cd、Hg、As、Pb、Cr、Cu、Ni和Zn)展开调查,分别从单种金属的污染等级、风险等级,以及整个区域的生态风险三个维度阐述当地的污染状况,系统性评估了当地的生态风险.研究表明Cd是主要污染贡献因子,其生态风险应引起重点关注.本研究证实了Monte-Carlo模拟可有效平衡Hakanson潜在生态风险指数法的随机性,将二者结合可更加全面精确地反映重金属的生态风险.