Pulse profile and metal transfer in pulsed gas metal arc welding: droplet formation,detachment and velocity

The influence of current profile and pulse parameters on droplet formation and transfer was investigated. One profile has an exponential ramp up and down in the current pulse shape, while the second is nearly square shaped. High-speed photography, synchronised with a high-speed data acquisition system, was used to monitor the droplet formation and transfer. It was found that for long-tail current profile, most of droplet formation and detachment occurs before background current is reached. While, for the nearly square pulse, most of droplet formation and transfer occurs during background current, giving a stable and smooth metal transfer. The arc attachment position was found to vary for the different profiles. Droplet speed was measured, and it was found that it is proportional to the peak current and inversely proportional to background current. Dimensionless process parameters were defined and used to predict droplet speed using a neural networks algorithm.     

基于NN的无信号混合交通交叉口自行车冲突避让行为建模仿真

在我国城市无信号交叉口,自行车与混合交通流的其他车辆(如机动车、自行车、行人)的冲突避让行为,难以通过传统数学方法建立准确的仿真模型.本文在以往交叉口自行车微观模型基础上,采用冲突车辆与自行车的相对位置矢量、相对速度矢量以及期望速度与实际速度相对矢量作为输入,自行车二维空间加速度矢量作为双输出的神经网络建立数学模型.利用实地采集自行车-机动车冲突、自行车-自行车冲突和自行车-行人冲突样本数据,进行神经网络训练和仿真验证,并对骑车者性别、冲突车辆类型和冲突速度进行了模型灵敏度分析.结果表明,模型能较好地反映无信号交叉口自行车与其他车辆及障碍物的冲突避让行为.     

基于遗传神经网络的军用油库选址优化模型

影响军用油库选址的因素众多且具有不确定性，传统的凭人主观选址和线性规划等方法选址不能充分体现各影响因素的主次成分及相互关系，为了解决该问题，将遗传算法和神经网络相结合，利用德尔斐法建立选址指标体系并进行指标的量化及归一化，将各底层指标的归一化值作为神经网络的输入，将代表选址等级的布尔变量作为神经网络的输出，利用遗传算法来优化神经网络的连接权值和阈值，然后用足够的样本借助 Matlab 工具训练此模型，通过模型的自适应学习，直到得到能正确表示网络内部特征的那组阈值。实际应用表明，所建模型的操作性和实用性强，为军用油库实际选址提供直接的决策依据。     

基于PSO-RBF的航空相机电源板故障诊断技术研究

针对机载航空相机电源板故障率高，传统故障诊断方法技术不足而造成的相机维护难度大的实际问题，提出一种基于PSO-RBF神经网络的电源板故障诊断方法。考虑选取RBF网络训练算法中隐含层节点数和中心参数的难题，采用实用粒子算法约简了RBF神经网络，设计了航空相机电源板故障诊断系统方案，并给出了诊断系统的软件模块和实现方法，实现了从计算机仿真到工程应用的电源电路故障诊断。仿真与实际检测实验结果均表明，系统在不依赖任何标准设备和附加测点时，可对航空相机电源板进行实时、全自动化故障检测，其故障现象的检测覆盖率为100%，故障诊断平均可靠性可达到97.73%，故障器件定位率可达到96.89%。     

A hybrid lithium‐ion battery model for system‐level analyses

We describe an advanced lithium‐ion battery model for system‐level analyses such as electric vehicle fleet simulation or distributed energy storage applications. The model combines an empirical multi‐parameter model and an artificial neural network with particular emphasis on thermal effects such as battery internal heating. The model is fast and can accurately describe constant current charging and discharging of a battery cell at a variety of ambient temperatures. Comparison to a commonly used linear kilowatt‐hour counter battery model indicates that a linear model overestimates the usable capacity of a battery at low temperatures. We highlight the importance of including internal heating in a battery model at low temperatures, as more capacity is available when internal heating is taken into account. Copyright © 2016 John Wiley & Sons, Ltd.     

分层流速控制水槽的模拟及智能化控制

设计了一种多路循环分层流速控制水槽,并建立了用于模拟分层控制的数学模型及实验平台. 通过建立的实验平台,采用实验和模拟结合的方法研究喷口流速组合对水槽中水体流速分布形态的影响. 针对同一种水流形态,通过改变其速度大小来观察水流形态的变化以及速度大小的分布和有效区域的范围变化. 通过改变进水口初始速度,得到不同的水流形态. 研究中模拟了常见的几种水流形态(直线型、上快下慢型和抛物线型),利用模拟所得各形态特征的数据,结合神经网络实现对实验平台智能化控制,从而实现对水槽的智能化控制.     

Structural Synthesis of Modular Weakly Connected Neural Networks. I. Methodology of Structural Synthesis of Modular Neural Networks

Structural models of modular neural networks without loops are considered. The notion of a weakly connected network is introduced. A genetic principle of generation of weakly connected neural networks is proposed, and its reproductive power is proved on the basis of the theory of categories. Algorithmic realizations of structural synthesis are considered. Some examples are given.     

夹支边矩形薄板裂纹诊断研究

 采用计算机仿真技术建立了四边夹支矩形板裂纹损伤的标准谱, 并研制出相应的基于人工神经网络的裂纹诊断系统。     

四边简支矩形板损伤诊断

本文建立了四边简支矩形板的固有频率与裂纹长度及裂纹位置关系的人工神经网络模型，该模型可用于结构的损伤监测，预报故障的发生。