面向空铁联运的枢纽航线网络优化模型

为减缓枢纽机场的客流量拥堵问题,减少机场运营成本,设计面向空铁联运的枢纽航线网络优化模型.在枢纽航线网络基础上,引入高铁线路建立优化模型;设计高维多角标变量表示方法和禁忌搜索算法的求解方法,减少算法的计算负担,科学规划线路客流量和运输方式.实验结果表明,与严格和非严格的枢纽航线网络相比,该优化模型可以较好减缓枢纽机场的拥堵,降低枢纽机场的负荷率,减少网络成本.该模型求解算法具有收敛速度快且解分布稳定的特点.     

代码搜索方法研究进展

开源软件的成功推动了软件产业的蓬勃发展,大量代码资源为代码搜索创造了条件.如何通过代码搜索技术找到需求代码成为一个重要问题.为了更好地推进后续研究工作,首先对代码搜索相关概念及研究趋势进行介绍和说明;其次对使用不同技术的代码搜索研究工作进行综述,包括基于信息检索、查询和代码描述增强、程序特征分类以及深度学习等方面,并进一步总结归纳不同方法的优缺点;接下来针对代码搜索技术所应用的多个领域进行介绍,包括程序合成、代码推荐与补全和代码风格改善等方面;最后分析现阶段代码搜索面临的主要问题,为未来该方向的发展提供一定的参考与建议.     

基于SSC-BP神经网络的异常检测算法

针对物联网环境下产生的新型网络攻击的数量持续上升和复杂性不断升高,传统的异常检测算法误报率高、检测率低以及数据量大而造成计算困难等问题,提出了一种基于子空间聚类(Subspace Clustering,SSC)和BP神经网络相结合的异常检测算法.首先在网络数据集上通过子空间聚类算法中最常用的CLIQUE算法得到不同的子空间;其次对不同子空间中的数据进行BP神经网络异常检测,计算预测误差值,通过与预先设定好的精度进行比较,来不断更新阈值进行修正,以达到提高识别网络攻击的能力.仿真实验采用NSL-KDD公开数据集和物联网环境下的网络攻击数据集,将NSL-KDD公开数据集分割为4种单一攻击子集和1种混合攻击子集,通过与K-means,DBSCAN,SSC-EA以及k-KNN异常检测模型进行比较,在混合攻击子集中,SSC-BP神经网络模型的检测率比传统的K-means模型的检测率提高了6％,误检率降低了0.2％;而在4种单一攻击子集中,SSC-BP神经网络模型都能以最低的误检率检测出最多的受到攻击的网络.在物联网环境下的网络攻击数据集上,SSC-BP神经网络模型的性能均优于其他几种对比模型.     

基于生活日志的情绪识别

抑郁症日益成为影响现代人生活幸福程度的重要因素。实时有效地识别用户情绪的方法对于抑郁症潜在患者的发现和治疗十分有意义。用户情绪的状态及变化会体现在其生活日志数据上。该文从使用可穿戴设备收集的用户生活日志数据出发,对数据进行了特征方面的分析;进一步地,使用以回归树为弱学习器的集成学习模型,设计了使用全部数据、仅用户自身数据以及仅他人数据进行训练的三组实验构成的对比实验框架,以比较使用不同用户数据对识别结果的影响。实验结果表明,基于生活日志数据的集成学习模型可以有效地识别用户的情绪状态。同时,基于实验结果提出了用户认知不一致的猜想,对于心理学上的抑郁分析也有一定的启发作用。该工作是目前所知第一个利用用户生活日志信息进行情绪识别以及抑郁症患者分析的工作,为后续进一步扩大实验规模和改良实验设计提供了思路。     

配电网自动化DTU终端的103规约的安全性分析

IEC 60870-5-103规约是应用于继电保护设备的信息接口配套标准,传输的主要内容是与继电保护有关的信息.该报文进行的是明文传输,缺乏加密措施和数字签名机制,安全性较低.为了验证以太网传输的103规约存在安全隐患和风险,搭建了主站与配电网自动化DTU终端的通信实验环境.运用ARP欺骗手段对系统进行了中间人攻击测试,实验的结果表明以太网传输的103规约具有中间人攻击的风险.为了提高协议的安全性,提出了一种基于非对称密码算法的双向身份认证机制,并采用对称加密机制、数字签名技术确保传输报文的机密性和完整性,最后通过仿真测试验证该方法的有效性.     

Fault estimation for aero‐engine LPV systems based on LFT

In this paper, a fault estimator with linear fractional transformation (LFT) parameter dependency is designed for the linear parameter‐varying (LPV) system of the aero‐engine with both sensor and actuator faults under disturbances. After an aero‐engine affine parameter‐dependent LPV model is derived by the linear fitting method and matrix pseudo‐inverse method, the LPV model with disturbances and fault signals is transformed into a LFT structure. Based on the full block S‐procedure, the sufficient condition for the existence of the fault estimator is proposed, which can lead to less conservative results. Then the fault estimator design algorithm which can adjust to the current system dynamic adaptively is presented. Finally, a fault estimator is designed for a turbofan aero‐engine under multiple types of faults and disturbances to demonstrate the effectiveness of the proposed method.     

Virtual Model Reduction-based Control Strategy of Planar Three-link Underactuated Manipulator with Middle Passive Joint

This paper presents a position control strategy for a planar active-passive-active (APA) underactuated manipulator with second-order nonholonomic characteristics. According to the structural characteristics of the planar APA system, we divide the system into two parts: a planar virtual Pendubot (PVP) and a planar virtual Acrobot (PVA). For the PVP, we mainly fulfill the target angle of the first link, which is calculated through the geometry method, and make the system stable. In this stage, via keeping the states of the third link being zero, the system is reduced to the PVP. Meanwhile, we design an open-loop control law based on the nilpotent approximation (NA) model of the PVP to make the second link stable and the first link stabilize at its target angle. Then, the planar APA system is reduced to a PVA with all links’ angular velocities being zero. For the PVA, we mainly realize the other two links’ target angles obtained via the particle swarm optimization (PSO) algorithm. Thus, the control objective of the planar APA system is achieved. Finally, above control strategy is verified by simulation results.

     

Chaos-PSO-based Motion Planning and Accurate Tracking for Position-posture Control of a Planar Underactuated Manipulator with Disturbance

Unlike a fully-actuated manipulator, the position-posture control of a planar underactuated manipulator (PUM) is more difficult, but the research on it is significant due to the wide practical applications. The existing control methods consider no external disturbance and are involved in the staged control idea, bringing the problems of nonsmooth control torque and time-consuming. A novel one-stage control approach is proposed in this paper for the position-posture control of a three-link PUM with the first free joint under the external disturbance. By analyzing the coupling relationship between its active joints and free joint, the position-posture control is transformed into the trajectory tracking control. Unlike the general trajectory planning, the trajectories of the active joints are planned to include several parameters. Meanwhile, the parameters are solved using a chaos particle swarm optimization algorithm to guarantee that all joint angles can reach to their desired angles. Then, to obtain the high trajectory tracking accuracy at every moment under the external disturbance, the nonlinear disturbance observer is constructed and a nonlinear fast terminal sliding mode tracking controller is designed. Finally, the feasibility and superiority of this strategy are verified via two simulations.

     

Commutation Torque Ripple Suppression in Three Phase Brushless DC Motor Using Open-end Winding

Three phase brushless DC motor has the characteristics of high power density, simple structure and excellent speed regulation performance, which is widely used in the field of electric drive. Open-end winding brushless DC motor has no neutral point, and each phase winding is electrically isolated, which can realize the independent control of each phase winding. Compared with Y-connected brushless DC motor, the open-end winding brushless DC motor has a wider range of speed regulation under the same voltage, and has certain fault-tolerant performance, which is suitable for the application of low voltage and large current. Compared with the multiphase fault-tolerant motor, open-end winding brushless DC motor has simpler structure and higher winding utilization rate. However, the open-end winding brushless DC motor still has torque ripple, among which the cogging torque is related to the motor body structure, while commutation torque ripple is affected by the mutual inductance of the winding and the current freewheeling loop. The commutation process of the open-winding brushless DC motor is analyzed, and it is concluded that the current change rates of the turn-off phase and turn-on phase are equal under the ideal back electromotive force(back-EMF), which can ensure that the non-commutation phase current and the electromagnetic torque kept constant. The overlapping commutation with phase current closed-loop method is proposed to suppress the commutation torque ripple of open-end winding BLDC motor, and the applied voltage on turn-off phase and delayed time of turn-off phase in overlapping commutation are derived, the sliding mode observer is used to calculate the electromagnetic torque. MATLAB simulation and experimental results verify the effectiveness of this method.

     

基于深度学习的小样本声目标识别方法

声目标分类识别是声源识别领域的核心问题,然而,在应用深层神经网络进行声目标分类识别时,从少量样本中学习(样本复杂度较低)是一个具有挑战性的问题;针对此问题,提出了一种基于深度学习的小样本声目标识别方法,该方法将手工设计特征和对数梅尔声谱特征结合到一起,扩充了深度学习模型的可利用特征量,提高了声信号识别效率和精度;在实验验证中,该方法在测试集上实现了87.6%的识别精度;更进一步地,用较少量的训练样本对该方法和其它几种主流的深度学习模型的性能进行了比较验证,结果表明,该方法只需要更少量的数据即可实现同样的识别精度,在声源探测领域具有一定应用价值。