基于可见/近红外光谱技术的橘小实蝇侵染柑橘检测研究

目的 使用可见/近红外光谱技术实施橘小实蝇侵染柑橘不同时期的无损检测。方法 研究选取人工制备的不同侵染时期的柑橘样本作为研究对象,利用搭建的可见/近红外光谱系统测量的光谱信息结合人工标定的侵染时期,对原始光谱进行了5种预处理,采用竞争性自适应重加权算法(competitive adaptive reweighted sampling, CARS)、连续投影算法(successive projections algorithm, SPA)两类方法提取反映侵染柑橘时期变化的光谱特征波长,应用偏最小二乘判别分析(partial least squares discriminant analysis,PLS-DA)建立基于特征波长光谱的柑橘侵染时期分类模型,对比分析不同光谱预处理方法的模型分类效果。结果 原始光谱经多元散射校正(multiplicative scatter correction, MSC)预处理的模型分类效果最佳,分别经CARS方法和SPA方法优选出了34和16个光谱特征波长。采用MSC-CARS-PLS-DA方法构建的模型分类效果最好,总准确率、假阳率分别为96.8%和0.0...     

商用车气压制动系统快放阀优化设计

针对商用车气压制动系统快放阀排气气压延迟问题,对快放阀进行优化设计以降低延迟气压与时间。应用系统仿真软件AMESim对快放阀进行模拟仿真,获得对排气延迟影响较大的变量,利用正交试验法研究了快放阀膜片厚度、排气口流通面积、排气口气隙宽度对快放阀排气延迟的影响。结果表明:最优快放阀结构的排气口气隙宽度为3 mm,膜片厚度为2 mm,下阀口支撑筋面积为70 mm²;通过快放阀结构的参数优化,可有效降低排气延迟气压与时间,压差同比降低约110 kPa,时间缩短0.65 s,同时提升了制动系统的稳定性。     

基于嗅觉可视化技术的食用植物油分类识别

为实现山茶油与3种常见食用植物油(菜籽油、大豆油和玉米油)的区分,制备可视化传感器阵列,采用嗅觉可视化技术对4种不同种类的食用植物油进行分类识别。采用主成分分析(PCA)对4种油样的特征数据进行降维,然后将降维后的数据导入K近邻(KNN)、极限学习机(ELM)、支持向量机(SVM) 3种分类模型中进行模型参数优化,对比了3种分类模型的分类结果。结果表明：建立的SVM分类模型性能最优,当输入主成分向量数为7、c=1.741 1、g=4.549 8时,SVM分类模型的测试集分类识别准确率为95.8%,五折交叉验证准确率为89.6%。制得的可视化传感器阵列可以实现4种食用植物油的分类识别,嗅觉可视化技术用于食用植物油检测是可行的。     

The influence of grain geometry and wear conditions on the material removal mechanism in silicon carbide grinding with single grain

High efficiency and precision grinding of brittle materials is challenging due to material physical and chemical properties. To understand the effect of grain geometry and wear conditions on the material removal mechanism in brittle material precision grinding, a single diamond grain grinding experiment was conducted on Silicon Carbide (SiC). The cutting edge radius and deflection angle were measured by confocal scanning. Under six different cutting edge radius and three maximum undeformed chip thickness, grinding force and ground surface were measured. Diamond grain wear was investigated by observing the grain morphology, wear rate, grinding force, and ground surface change over accumulative material removal volume. The result showed the existence of a critical cutting edge radius for improving SiC ground surface quality.. Normal grinding force increased with the cutting edge radius increase. Tangential grinding force increased with the cutting edge radius increase and reached the peak value at the critical cutting edge radius. Flank wear was the major wear mode in precision SiC grinding. The grain wear was associated with the grinding force and ground surface.     

Broad spectral response photosensitive organic field-effect transistors realized by the hybrid planar-bulk heterojunction composed of three molecules with complementary optical absorption

Photodetectors have a growing number of diverse applications in both military and civilian use, and improving their response bandwidth is of great importance. Broad spectral response photosensitive organic field-effect transistors (PhOFETs) are realized by adopting hybrid planar-bulk heterojunction (HPBHJ) as the active structure that containing three molecules with complementary optical absorption. PhOFETs based on the HPBHJ composed of 3,4,9,10-perylene tetracarboxylic dianhydride (PTCDA), chloroaluminum phthalocyanine (AlClPc) and C₆₀, and SiO₂ as the gate dielectric exhibit photo response in the broadband wavelength range of 300–850 nm. At an incident optical intensity of 0.07 mW/cm², the mean photoresponsivity and uniformity factor were 0.40 A/W and 0.79, respectively. Furthermore, by replacing SiO₂ gate dielectric with polyvinyl alcohol (PVA), the device performance could be further improved, the mean photoresponsivity was increased to 2.44 A/W. The physical origin of the performance improvement is discussed.     

基于注意力机制的交通标志识别

针对实际场景中的交通标志大多小而密集,导致小目标交通标志识别准确度较低的问题,提出一种改进YOLOv5算法。首先将CBAM同时嵌入YOLOv5网络的Backbone和Head部分,以提升网络特征提取能力。其次为解决GIoU Loss可能造成的模型收敛速度较慢问题,改用DIoU Loss作为网络回归损失函数。实验结果表明,改进后的算法对于交通标志图像的识别平均准确率达到96.40%,相较于原算法有了6.83%的提升。最后为验证模型的实时可行性,在TX2嵌入式系统中利用本文改进YOLOv5算法对实景视频中的交通标志进行识别,结果表明本文改进算法能在嵌入式系统中流畅运行。     

铁路卧铺客车人体振动舒适性建模与仿真

铁路卧铺客车乘客在卧姿状态下所承受的全身振动是影响其乘用舒适度的主要因素。研究了卧姿人体垂直振动模型,在铁道车辆二系悬挂动力学模型基础上,考虑卧铺的隔振作用与卧姿人体的垂直振动响应特性,建立了14自由度"人-铺-车辆"振动系统空间垂向耦合动力学模型,研究了在轨道随机不平顺激励下的卧姿人体垂直振动响应。应用卧姿人体全身振动舒适性评价标准,建立了铁路卧铺客车人体振动舒适性仿真流程。通过对人体头-臀二部位加速度均方根值先后进行部位计权和频率计权,得到卧铺客车卧姿人体振动舒适性指标。以M atlab为工具编制了卧铺客车人体振动舒适性仿真软件,交互输入车辆与人体的结构和动力学参数后,自动完成卧姿人体振动舒适性仿真计算,进而为铁路卧铺客车人体振动舒适性分析及车辆悬架参数优化提供了有效分析手段。     

Synthesis and photoluminescence properties of Sr3(PO4)2:Re3+, Li+ (Re = Eu,Sm) red phosphors for white light-emitting diodes

Sr₃(PO₄)₂:Re³⁺, Li⁺ (Re = Eu, Sm) red phosphors were prepared via a high temperature solid state reaction, and their structure and luminescence properties were investigated. X-ray diffraction patterns indicate that the phase of as-prepared samples is in good agreement with standard Sr₃(PO₄)₂ structure. Under 395 nm excitation, the emission of Sr₃(PO₄)₂:Eu³⁺ consists of a strong peak centered at 622 nm and two weak peaks centered at 598 nm and 660 nm, which correspond to ⁵D₀→⁷F₂, ⁵D₀→⁷F₁ and ⁵D₀→⁷F₃ transitions, respectively. Also, the emission spectrum of Sr₃(PO₄)₂:Sm³⁺ shows three main peaks at 568 nm, 603 nm and 651 nm, which are attributed to ⁴G_5/2→⁶H_I/2 (I = 5, 7, 9) transitions of Sm³⁺. Furthermore, luminescence properties of Sr₃(PO₄)₂:Re³⁺, Li⁺ (Re = Eu, Sm) samples are enhanced significantly by Li⁺ ions doping as charge compensator. Results indicate that as-prepared Sr₃(PO₄)₂:Re³⁺, Li⁺ (Re = Eu, Sm) could be the potential red phosphors used in white light-emitting diodes.     

油茶果采摘机阀控液压马达模糊神经网络PID控制

推摇式油茶果采摘机在作业机构作业时,需要保证振动液压马达恒定转速输出,以保证油茶果能够顺利通过推摇振动从树枝脱落,对此,推导了推摇式油茶果采摘机阀控振动液压马达系统的状态空间方程,并在传统增量式PID控制原理的基础上设计了模糊径向基函数(Radial Basis Function,RBF)神经网络PID控制方法。采用MATLAB/Simulink仿真软件对液压系统在空载和5 s带载工况进行仿真,并与传统PID控制和模糊PID控制方法进行比较和分析。仿真结果显示,传统PID控制和模糊PID控制响应速度较慢、鲁棒性较差;而采用模糊RBF神经网络PID控制方法响应速度快、鲁棒性强,能够很好地满足振动液压马达恒定转速输出的要求,并且能够灵活地在线调整PID的3个参数,控制精度较高。     

A new image encryption scheme based on a chaotic function

In recent years, several methods of secure image encryption were studied and developed through chaotic processes or functions. In this paper, a new image encryption scheme based on a coupling of chaotic function and xor operator is presented. The main advantages of such a method are the abilities to produce a large key space to resist brute-force attacks, and to encrypt securely images with any entropy structure assuring indistinguishability, confusion and diffusion properties in the corresponding cipher-images. The results of several statistical analysis about randomness, sensitivity and correlation of the cipher-images show that the proposed cryptosystem is efficient and secure enough to be used for the image encryption and transmission. Moreover, the implementation of the corresponding algorithm is easy and only integers are used.