Improving accuracy of automatic optical inspection with machine learning

Electronic devices require the printed circuit board(PCB)to support the whole structure,but the assembly of PCBs suffers from welding problem of the electronic components such as surface mounted devices(SMDs)resistors.The automated optical inspection(AOI)machine,widely used in industrial production,can take the image of PCBs and examine the welding issue.However,the AOI machine could commit false negative errors and dedicated technicians have to be employed to pick out those misjudged PCBs.This paper proposes a machine learning based method to improve the accuracy of AOI.In particular,we propose an adjacent pixel RGB value based method to pre-process the image from the AOI machine and build a customized deep learning model to classify the image.We present a practical scheme including two machine learning procedures to mitigate AOI errors.We conduct experiments with the real dataset from a production line for three months,the experimental results show that our method can reduce the rate of misjudgment from 0.3%–0.5%to 0.02%–0.03%,which is meaningful for thousands of PCBs each containing thousands of electronic components in practice.     

南水北调中线总干渠藻类的生态调度

南水北调中线总干渠无在线调蓄水库,对藻类生态调度过程中出现的问题开展生态调度实现策略和实施方式研究。主要实现策略包括:划定自身的调蓄区,隔离生态调度对下游的影响;采用高效的渠池运行方式,减少生态调度时蓄量的反复调整;综合考虑安全、快速、平稳等需求,设定生态调度实施进程和方式。具体实施方式包括:将总干渠划分为流速调控区、调蓄区和正常运行区,分别实施等体积、控制蓄量和闸前常水位方式运行;将生态调度过程划分为充水阶段和泄水阶段,基于流速调控目标值、持续时长和水位降幅约束条件,确定各阶段时长和各分区的闸门群调控方案等。基于2018年3月输水工况,采用明渠一维非恒定流模型,仿真总干渠上游15个渠池的藻类生态调度过程。结果表明,生态调度可在3.5 d内完成,各渠池的平均流速由0.48 m/s增至0.93 m/s,持续时间超过2 h。在整个生态调度过程中,水位变化平稳,水位变幅符合安全阈值要求,下游渠道的正常运行未受生态调度明显影响。     

基于VMD和BSA-KELM的高陡边坡位移预测模型研究

边坡位移的时间序列曲线存在复杂的非线性特性,传统的预测模型精度不足以满足预测要求。为此提出了基于变分模态分解的鸟群优化-核极限学习机的预测模型,并用于河北省某水泥厂的边坡位移预测。该方法首先采用VMD把边坡位移序列分解为一系列的有限带宽的子序列,再对各子序列分别采用相空间重构并用核极限学习机预测,采用鸟群算法优化相空间重构的嵌入维度和KELM中惩罚系数和核参数三个数值,以取得最优预测模型。最后将各个子序列预测值叠加,得到边坡位移的最终预测值。结果表明：和KELM、BSA-KELM、EEMD-BSA-KELM模型相比,基于VMD的BSA-KELM预测精度更高,为边坡位移的预测提供一种有效的方法。     

Application of artificial neural network method to predict the breakage properties of PGE bearing chromite ore

In the present investigation, systematic grinding experiments were conducted in a laboratory ball mill to determine the breakage properties of low-grade PGE bearing chromite ore. The population balance modeling technique was used to study the breakage parameters such as primary breakage distribution (B_{i, j}) and the specific rates of breakage (S_i). The breakage and selection function values were determined for six feed sizes. The results stated that the breakage follows the first-order grinding kinetics for all the feed sizes. It was observed that the coarser feed sizes exhibit higher selection function values than the finer feed size. Further, an artificial neural network was used to predict breakage characteristics of low-grade PGE bearing chromite ore. The predicted results obtained from the neural network modeling were close to the experimental results with a correlation of determination R² = 0.99 for both product size and selection function.     

Evaluation of a sulfidogenic system fed with microalgal biomass of Chlorella pyrenoidosa as an electron donor: Sulfate reduction kinetics

In this study, a sulfidogenic reactor fed with microalgal biomass of Chlorella pyrenoidosa as an electron donor was operated in a continuous mode. This study evaluated the influence of various initial sulfate concentration from 1.0 to 2.5 g/L on anaerobic sulfate reduction kinetics by a sulfidogenic enrichment culture predominantly Desulfovibrio sp. VSV2. It was observed that volumetric sulfate reduction rate (VSRR) was consistently increasing with an increase in volumetric sulfate loading rate (VSLR) across the retention time of 7–10 days. For a retention time of 7 days, the maximum VSRR was noted as 0.0050 g/(L.h) with a corresponding VSLR of 0.0089 g/(L.h). When retention time was maintained for 10 days, a maximum sulfate reduction of 65% and a maximum bacterial concentration of 1.632 g/L were achieved for an initial sulfate concentration of 1.5 g/L. It was concluded that VSLR facilitated through both dilution rate and initial sulfate concentration had a significant influence over sulfate reduction kinetics. The results of the study suggested that the microalgal-fed sulfidogenic system could be effectively employed for reduction of sulfate from sulfate-rich wastewater.     

An analytical model for gas leakage through contact interface in proton exchange membrane fuel cells

Sealing performance between two contacting surfaces is of significant importance to stable operation of proton exchange membrane (PEM) fuel cells. In this work, an analytical micro-scale approach is first established to predict the gas leakage in fuel cells. Gas pressure and uneven pressure distribution at the interface are also included in the model. At first, the micro tortuous leakage path at the interface is constructed by introducing contact modelling and fractal porous structure theory. In order to obtain the leakage at the entire surface, contact pressure distribution is predicted based on bonded elastic layer model. The gas leakage through the discontinuous interface can be obtained with consideration of convection and diffusion. Then, experiments are conducted to validate the numerical model, and good agreement is obtained between them. Finally, influences of surface topology, gasket compression and gasket width on leakage are studied based on the model. The results show that gas leakage would be greatly amplified when the asperity standard deviation of surface roughness exceeds 1.0 μm. Gaskets with larger width and smaller thickness are beneficial to sealing performance. The model is helpful to understand the gas leakage behavior at the interface and guide the gasket design of fuel cells.     

PID Control of Planar Nonlinear Uncertain Systems in the Presence of Actuator Saturation

This paper investigates PID control design for a class of planar nonlinear uncertain systems in the presence of actuator saturation. Based on the bounds on the growth rates of the nonlinear uncertain function in the system model, the system is placed in a linear differential inclusion. Each vertex system of the linear differential inclusion is a linear system subject to actuator saturation. By placing the saturated PID control into a convex hull formed by the PID controller and an auxiliary linear feedback law, we establish conditions under which an ellipsoid is contractively invariant and hence is an estimate of the domain of attraction of the equilibrium point of the closed-loop system. The equilibrium point corresponds to the desired set point for the system output. Thus, the location of the equilibrium point and the size of the domain of attraction determine, respectively, the set point that the output can achieve and the range of initial conditions from which this set point can be reached. Based on these conditions, the feasible set points can be determined and the design of the PID control law that stabilizes the nonlinear uncertain system at a feasible set point with a large domain of attraction can then be formulated and solved as a constrained optimization problem with constraints in the form of linear matrix inequalities (LMIs). Application of the proposed design to a magnetic suspension system illustrates the design process and the performance of the resulting PID control law.      

基于卷积神经网络的5G蜂窝网络无线定位方法

5G蜂窝网络发展迅猛，其覆盖面积将逐渐增大，因此使用5G蜂窝网络进行定位是有研究潜力的研究方向。本文提出一种新的深度学习技术来实现高效、高精度和低占用的定位，以代替传统指纹定位过程中繁重的指纹库生成以及距离计算。该方法建立了一个特殊的卷积神经网络，并根据5G天线信号的接收信号强度指示、相位和到达角等特征量，选择合适的输入数据格式构造样本组建训练集，对该卷积神经网络进行训练。训练得到的卷积神经网络可以替代指纹定位中的庞大指纹库，非常有利于直接在5G移动设备端实现定位。虽然卷积神经网络在训练过程中需要大量时间，但在训练完毕后直接进行分类定位的速度非常快，可以保障定位实现的实时性。本文所实现的卷积神经网络权重与偏置所占内存不到0.5 MB，且能够在实际应用环境中以95%的定位准确率以及0.1 m的平均定位精度实现高精度定位。