基于形变长短期记忆网络的换道意图预测

混行交通下的自动驾驶车辆需具备换道意图预测能力来保障行驶安全。为尽早预测车辆换道意图,提出一种基于形变长短期记忆(mogrifier long short-term memory,M-LSTM)网络的换道意图预测模型。首先采用S-G(Savitzky-Golay)滤波器对自然驾驶数据集NGSIM(next generation simulation)进行降噪筛选,按向左换道、向右换道、直线行驶对不同时间长度的轨迹序列标注,选取车辆运动信息与环境信息输入模型,最后采用softmax函数进行意图分类。试验结果表明,在不同预判时间下,模型准确率均高于支持向量机(support vector machine,SVM)、LSTM模型,且越接近换道点预测准确率越高,在1.0、2.5 s时预测准确率分别为93.83%与81.30%。提出的模型具有良好的准确性与预判性,能为自动驾驶车辆尽早识别换道意图提供技术支持。

演示记忆合金弹簧的形状记忆效应的实验设计

设计了测试记忆合金弹簧的形状记忆效应的物理实验.描述了实验装置的设计和结构,给出了实验内容的设计和实验方法,包括观察测试记忆合金弹簧的温度-形变记忆效应实验和电流-形变记忆效应实验.根据测量数据,描绘了用该仪器得到的温度-形变曲线和电流-形变曲线.该实验可作为设计性实验在大学物理实验课中开出.

Crystallization of amorphous NiTi shape memory alloy fabricated by severe plastic deformation

Based on the local canning compression,severe plastic deformation（SPD） is able to lead to the almost complete amorphous nickel-titanium shape memory alloy（NiTi SMA）,in which a small amount of retained nanocrystalline phase is embedded in the amorphous matrix.Crystallization of amorphous NiTi alloy annealed at 573,723 and 873 K was investigated,respectively.The crystallization kinetics of the amorphous NiTi alloy can be mathematically described by the Johnson-MehlAvrami-Kolmogorov（JMAK） equation.NiTi SMA with a complete nanocrystalline phase is obtained in the case of annealing at 573 K and 723 K,where martensite phase transformation is suppressed due to the constraint of the grain boundaries.Crystallization of amorphous NiTi alloy at 873 K leads to the coarse-grained NiTi sample,where（001） martensite compound twin is observed at room temperature.It can be found that the martensitic twins preferentially nucleate at the grain boundary and they grow up towards the two different grains.SPD based on the local canning compression and subsequent annealing provides a new approach to obtain the nanocrystalline NiTi SMA.

可作为肝脏损伤能量评估模型的新型塑性复合材料的制备研究

在(95±5)℃的条件下,采用不同比例的硬脂酸钾(PS)、硬脂酸钠(SS)、液体石蜡(PL)和聚乙烯醇(PVA)共混搅拌制备出可作为肝脏损伤能量评估模型的塑性材料。采用Instron 8871动态疲劳材料拉伸压缩试验机测定不同比例的塑性材料的力学性能,结果表明,PS/SS/PL/PVA质量比为35/35/28/2时,变形抗力为0.056MPa,极限压缩应变为10.673%,塑性材料具有良好的塑性变形性能;扫描电子显微镜(SEM)观察结果与力学性能测试结果吻合。

Ni_(56)Mn_(25-x)Cr_xGa_(19)(x=0,2,4,6) high temperature shape memory alloys

The microstructure,martensitic transformation behavior,mechanical and shape memory properties of Ni56Mn25-xCrxGa19（x=0,2,4,6） alloys were investigated.Single phase of martensite with tetragonal structure is present for x=0,and dual-phase containing martensite and γ phase is observed for x≥2.The martensitic transformation peak temperatures decrease monotonically from 401 ℃ for x=0 to 197 ℃ for x=6.The introduction of γ phase by Cr addition is proved to be effective in improving the workability and ductility.The tensile stress and strain are 497 MPa and 8 % for x=4,and 454 MPa and 5.5 % for x=6,respectively.The shape memory strain values are 2.7 % under a residual strain of 4.5 % for x=4,and 1.9 % under a residual strain of 3.5 % for x=6,respectively.

Influence of cooling rate on phase transformation and microstructure of Ti-50.9%Ni shape memory alloy

Heat treatment of Ti-50.9%Ni （mole fraction） alloy was studied by differential scanning calorimetry, X-ray diffraction, scanning electron microscopey and energy dispersive X-ray analysis to investigate the influence of cooling rate on transformation behavior and microstructures of NiTi shape memory alloy. The experimental results show that three-stage phase transformation can be induced at a very low cooling rate such as cooling in furnace. The cooling rate also has a great influence on the phase transformation temperatures. Both martensitic start transformation temperature （Ms） and martensitic finish transformation temperature （Mf） decrease with the decrease of the cooling rate, and decreasing the cooling rate contributes to enhancing the M→A austenite transformation temperature. The phase transformation hysteresis （Af-Mf） increases with the decrease of the cooling rate. Heat treatment is unable to eliminate the textures formed in hot working of NiTi sample, but can weaken the intensity of them. The cooling rate has little influence on the grain size.

膨胀堵漏材料及堵漏工艺研究进展

在新一轮战略找矿行动中,为了更好地完成大量的钻探工作,减少冲洗液的漏失,降低钻探施工成本,应加强膨胀堵漏材料的研究和应用。本文综述了吸水膨胀材料、吸油膨胀材料、形状记忆形变堵漏材料的合成方法;为适应钻孔堵漏的需求,提高膨胀材料的强度、抗盐性以及延迟膨胀技术;采用膨胀堵漏材料进行堵漏时的施工工艺。为膨胀堵漏材料的推广应用提供指导。

融合注意力机制SimAM的CNN-MogrifierLSTM航班延误波及预测

连续航班中的延误波及往往会引起大规模的航班延误产生,提前对航班延误波及问题进行预测可以为民航部门提供有效参考,减少相关的经济损失。本文首先对航班数据进行清洗与数据融合,针对空管部门实际航班运行情况提出强空间航班链数据集与强时序航班链数据集两种不同的构造方法;然后根据航班延误波及传播的空时特性提出融合注意力机制SimAM的CNN-MogrifierLSTM网络模型,先使用卷积神经网络(Convolutional Neural Network,CNN)结合注意力机制SimAM模块对空间特征进行提取,再用形变的长短时记忆网络(Mogrifier Recurrent Neural Network,MogrifierLSTM)对时序信息进行学习;最后使用Softmax分类器对延误等级进行分类预测。本文提出的预测方法,在航班延误波及进行预测的实验中取得了93.16%的准确率,相比单独使用CNN或LSTM大有提升,加上SimAM注意力机制后相比CNN-MogrifierLSTM网络在不同数据集上准确率也提升了0.6%左右。

Influence of recovery heating rate on shape memory effect in up-quenched Cu-Al-Mn alloy

The effect of recovery heating rate on shape memory effect of the up-quenched Cu-8.88Al-10.27Mn（mass fraction, %） alloy was investigated by optical microscopy, electron transmission microscopy（TEM） and electrical resistivity measurement. It is found that the shape recovery rate decreases as the heating rate decreases. It can reach 75% when the heating rate is 20 ℃/min, while it is only 8% when the heating rate is 1 ℃/min. In situ microstructure observation indicates that the dependence of shape memory effect on recovery heating rate is caused by the stabilization of twinned martensite induced by deformation. The analysis of electrical resistivity shows that the stabilization of twinned martensite may be ascribed to formation of compound defects of vacancies and dislocations at the boundaries of twinned martensite during the slow heating. The compound defects prevent the reverse transformation of twinned martensite.

基于CNN-Mogrifier LSTM的人体运动模式识别算法

随着传感器、微电子等技术的发展,通过可穿戴式传感器对人体的运动模式进行识别,具有广泛的应用价值,如何提高识别的准确率,具有重要研究意义。考虑到人体下肢运动的特点,提出了一种基于CNN和Mogrifier LSTM的人体运动模式识别算法,先利用CNN提取原始数据的局部相关特征,再使用Mogrifier LSTM代替全连接层,挖掘局部相关特征的前后依赖关系,对行走、跑步、上楼梯、下楼梯、上坡和下坡6种常见的运动模式进行识别。实验结果表明,相比于传统LSTM算法,Mogrifier LSTM的准确率提升了1.03%,将CNN和Mogrifier LSTM相结合后,准确率进一步提升了1.17%,达到了98.18%,证明了算法的优越性。

Superelastic properties of nanocrystalline NiTi shape memory alloy produced by thermomechanical processing

Effects of thermomechanical treatment of cold rolling followed by annealing on microstructure and superelastic behavior of the Ni50Ti50 shape memory alloy were studied.Several specimens were produced by copper boat vacuum induction melting.The homogenized specimens were hot rolled and annealed at 900°C.Thereafter,annealed specimens were subjected to cold rolling with different thickness reductions up to 70%.Transmission electron microscopy revealed that the severe cold rolling led to the formation of a mixed microstructure consisting of nanocrystalline and amorphous phases in Ni50Ti50 alloy.After annealing at 400°C for 1 h,the amorphous phase formed in the cold-rolled specimens was crystallized and a nanocrystalline structure formed.Results showed that with increasing thickness reduction during cold rolling,the recoverable strain of Ni50Ti50 alloy was increased during superelastic experiments such that the 70%cold rolled-annealed specimen exhibited about 12%of recoverable strain.Moreover,with increasing thickness reduction,the critical stress for stress-induced martensitic transformation was increased.It is noteworthy that in the 70%cold rolled-annealed specimen,the damping capacity was measured to be 28 J/cm3 that is significantly higher than that of commercial NiTi alloys.

Constitutive equation and processing map of equiatomic NiTi shape memory alloy under hot plastic deformation

In order to describe the deformation behavior and the hot workability of equiatomic NiTi shape memory alloy (SMA) during hot deformation, Arrhenius-type constitutive equation and hot processing map of the alloy were developed by hot compression tests at temperatures ranging from 500 to 1100 °C and strain rates ranging from 0.0005 to 0.5 s?1. The results show that the instability region of the hot processing map increases with the increase of deformation extent. The instability occurs in the low and high temperature regions. The instability region presents the adiabatic shear bands at low temperatures, but it exhibits the abnormal growth of the grains at high temperatures. Consequently, it is necessary to avoid processing the equiatomic NiTi SMA in these regions. It is preferable to process the NiTi SMA at the temperatures ranging from 750 to 900 °C.

Effects of Co and Al addition on martensitic transformation and microstructure in ZrCu-based shape memory alloys

The effect of ternary alloying element Al and quaternary alloying element Co on the martensitic transformation of ZrCu-based shape memory alloy was investigated. The results show that the addition of Al and Co in ZrCu alloy decreases both the martensitic transformation temperature and the martensitic transformation temperature hysteresis. Transmission electron microscope (TEM) observations reveal that theCm martensite structure is the preferential formation phase. The intervariant structures in ZrCuAlCo alloy are (021) type I twins, while the dominant substructures inside the martensite variant are the (001) compound twins. With the increase of Co content, tensile fracture strength and strain are improved obviously.

Thermo-mechanical Behaviors of Functionally Graded Shape Memory Alloy Timoshenko Composite Beams

This paper focuses on the thermo-mechanical behaviors of functionally graded(FG)shape memory alloy(SMA)composite beams based on Timoshenko beam theory.The volume fraction of SMA fiber is graded in the thickness of beam according to a power-law function and the equivalent parameters are formulated.The governing differential equations,which can be solved by direct integration,are established by employing the composite laminated plate theory.The influences of FG parameter,ambient temperature and SMA fiber laying angle on the thermo-mechanical behaviors are numerically simulated and discussed under different boundary conditions.Results indicate that the neutral plane does not coincide with the middle plane of the composite beam and the distribution of martensite is asymmetric along the thickness.Both the increments of the functionally graded parameter and ambient temperature make the composite beam become stiffer.However,the influence of the SMA fiber laying angle can be negligent.This work can provide the theoretical basis for the design and application of FG SMA structures.

Microstructure evolution and deformation behavior of as-cast NiTi shape memory alloy under compression

The as-cast ingot of equiatomic nickel-titanium shape memory alloy （SMA） was prepared via vacuum consumable arc melting. Differential scanning calorimetry （DSC）, X-ray diffraction （XRD） analysis, energy dispersive X-ray （EDX） measurement and microanalysis were adopted in order to understand the microstructure evolution and deformation behavior of the as-cast NiTi SMA under compression at various strain rates and temperatures. The microstructures of as-cast NiTi SMA composed of dendritic grains and equiaxed grains are inhomogeneous and show segregation. The as-cast NiTi SMA consists of B19′ martensite, B2 austenite and Ti2Ni phase simultaneously at room temperature. The as-cast NiTi SMA is sensitive to strain rate under compression at high temperature, at which NiTi SMA is characterized by dynamic recrystallization at strain rates of 0.1 and 0.01 s-1, but by dynamic recovery at strain rate of 0.001s-1. The strain rates have little influence on the true stress—strain curves as well as microstructure of NiTi SMA at room temperature and -100 °C.

Microstructure, martensitic transformation and mechanical properties of Ni-Mn-Sn alloys by substituting Fe for Ni

The effects of partial substitution of Fe element for Ni element on the structure,martensitic transformation and mechanicalproperties of Ni50-xFexMn38Sn12(x=0and3%,molar fraction)ferromagnetic shape memory alloys were investigated.Experimentalresults indicate that by substitution of Fe for Ni,the microstructure and crystal structure of the alloys change at room temperature.Compared with Ni50Mn38Sn12alloy,the martensitic transformation starting temperature of Ni47Fe3Mn38Sn12alloy is decreased by32.5K.It is also found that martensitic transformation occurs over a broad temperature window from288.9to352.2K.It is found that themechanical properties of Ni-Mn-Sn alloy can be significantly improved by Fe addition.The Ni47Fe3Mn38Sn12alloy achieves amaximum compressive strength of855MPa with a fracture strain of11%.Moreover,the mechanism of the mechanical propertyimprovement is clarified.Fe doping changes the fracture type from intergranular fracture of Ni50Mn38Sn12alloy to transgranularcleavage fracture of Ni47Fe3Mn38Sn12alloys.

Thermo-electric behaviour of NiTi shape memory alloy

The thermo-electric behavior of shape memory alloy （SMA） wire was presented. When the wire was electrically heated above its transformation temperature by current, a large mechanical force is exerted due to transformation in its phases. In order to make use of SMA wire as an actuator, different parameters and their relationships were investigated. These parameters are recoverable strain （displacement）, temperature hysteresis and electrical resistance variation under different stress levels. Optimum safe heating current was assessed and phase transformation temperatures were estimated by heat transfer model. The wire was heated and cooled by 680 mA of current for 796 s under natural air convection. The strain recovered is 4.33% and corresponding change in resistance is 11.2% at 43 MPa of stress. The resistance variation shows linearly with displacement and current during heating and cooling cycle respectively. This study will be useful in precisely controlling of SMA wire actuator with and without external sensor feedback.

基于改进的Mogrifier LSTM算法水质浊度预测模型

水环境资源保护作为当下最重要的工作之一,为了提高水质模型预测精度并制定更加完善的水资源管理策略,现提出一种基于改进的Mogrifier LSTM算法水质浊度数据预测模型,实现对水质数据的精准预测。该模型首先利用CNN卷积神经网络对复杂的水质数据进行特征提取,有效解决了水质数据非线性不稳定的特点,并对传统Mogrifier机制进行优化,引入扩展系数,利用PSO粒子群优化算法对超参数进行寻优操作,通过改造后的Mogrifier机制对LSTM模型中不同时刻的上下文信息进行融合,增强了水质数据的信息的交互。将预测结果与众多传统模型进行对比,结果表明CNN-改进Mogrifier LSTM模型具有更好的预测效果。

A molecular dynamics investigation of the deformation mechanism and shape memory effect of epoxy shape memory polymers

Following deformation, thermally induced shape memory polymers(SMPs) have the ability to recover their original shape with a change in temperature. In this work, the thermomechanical properties and shape memory behaviors of three types of epoxy SMPs with varying curing agent contents were investigated using a molecular dynamics(MD) method. The mechanical properties under uniaxial tension at different temperatures were obtained, and the simulation results compared reasonably with experimental data. In addition, in a thermomechanical cycle, ideal shape memory effects for the three types of SMPs were revealed through the shape frozen and shape recovery responses at low and high temperatures, respectively, indicating that the recovery time is strongly influenced by the ratio of E-51 to 4,4'-Methylenedianiline.

Softening micromechanical constitutive model of stress induced martensite transformation for NiTi single crystal shape memory alloy

Based on the assumption of laminated microstructure, a micromechanical model of stress induced martensite transformation for NiTi shape memory alloys single crystal is proposed. Elasticity anisotropy and different proper-ties for two phases are considered. Martensite volume fraction is chosen as the internal variable that controls the phase transformation quantitatively. An effective macroscopic elasticity matrix based on the different elasticity characteristics of each phase and the martensite volume fraction are obtained with the help of the perfect interfa-cial relationships. A phase transformation driving force is derived to construct the transformation criterion. The model corresponds to a non-convexity free energy function during phase transformation, so softening behavior can be well simulated by the model. A numerical simulation is implemented for the uniaxial loading of NiTi single crystal alloy according to the model, and simulation results are proved by experimental results of polycrystal with strong {111} texture. Superelasticity, Anisotropy, the evolution of microstructure and softening behavior can be well simulated.