退火炉内罩中气体流动及钢卷传热的数值模拟

摘要：轧制力预报一直是热连轧过程控制模型的核心，浅层神经网络对复杂函数的表示能力有限，而深度学习模型通过学习一种深层非线性网络结构，实现复杂函数逼近。利用深度学习框架TensorFlow，构建了一种深度前馈神经网络轧制力模型，采用BP算法计算网络损失函数的梯度，运用融入Mini batch策略的Adam优化算法进行参数寻优，采用Early stopping、参数惩罚和Dropout正则化策略提高模型的泛化能力。基于上述建模策略，针对宝钢1880热连轧精轧机组的大量轧制历史数据进行了建模实验，对比分析了4种不同结构的前馈网络预测精度。结果表明，相比于传统SIMS轧制力模型，深度神经网络可实现轧制力的高精度预测，针对所有机架的预测精度平均提升21.11%。     

Prediction of the water content of biodiesel using ANN-MLP: An environmental application

Fuel quality, especially biodiesel, is highly dependent on its water content, and the major sources of water in the fuel relate to the transportation, production, and storage processes. In this present contribution, the multilayer perceptron artificial neural network (MLP-ANN) was applied to predict the water content of biodiesel and diesel blend in terms of temperature and composition. The proposed algorithm was trained and tested by utilizing 400 experimental data points which were extracted from the literature. Based on the results, the MLP-ANN model has great ability to estimate the water content of biodiesel and diesel blend. The R-squared (R²), root mean square error, average absolute relative deviation, and a?bsolute deviation parameters for the total data set are obtained, respectively, as 0.99784, 123919.1172, 3.3632, and 1.17%, which indicate the effective performance suggested by ANN. As the computational study is cheaper and easier than the experimental study, the developed software could be considered as an alternative for laboratory study, and the environmental effect of biodiesel and produced undesired product after biodiesel combustion which is directly related to the water content of biodiesel is estimable with the information released in this study.     

基于PSO-MLP神经网络的遥信插件质量识别方法研究

为了提高继电保护装置遥信插件的出厂质量,解决产品检验自动化的问题,提出了一种基于PSO优化多层感知器(Multi-LayerPerceptron,MLP)神经网络的遥信插件质量识别方法。首先,建立了继电保护装置遥信插件自动化硬件测试平台。然后,改进了PSO优化算法,调整了惯性权重?的滑动特性,使其根据粒子间距实时调整步进。最后,在SPSS中使用k-s检验对原始起动电压数据进行正态性检验,得到了具有正态性样本的频率分布及其拟合曲线,提取特征训练集,并对神经网络进行训练和测试。实验结果表明,该方法能够有效且准确地对遥信插件进行质量识别,实现了产品检验和质量识别的自动化及智能化。PSO-MLP神经网络训练时间短,收敛速度快,质量识别准确度高,约为97%,且泛化能力强。     

基于HALCON的发票号字符识别研究

为了解决人工方式识别发票号效率低下、费时费力的问题,以HALCON软件为平台,对发票号进行识别,首先,对读取的图像进行图像预处理,包括把原图像转化成单通道图像、图像增强、高斯滤波、阈值分割和连通区域分割,然后对预处理后的图像进行数学形态学处理,接着把粘连的数字拆分开,最后用基于多层感知分类器(MLP)对发票号数字进行识别.通过与人工方式的方法对比,结果表明,用基于HALCON的方法,发票号数字能被快速、精确的识别出来,有效的减少了人员的工作量,提高了工作效率.     

Parkinson’s Disease Detection Using Biogeography-Based Optimization

In recent years, Parkinson's Disease (PD) as a progressive syndrome of the nervous system has become highly prevalent worldwide. In this study, a novel hybrid technique established by integrating a Multi-layer Perceptron Neural Network (MLP) with the Biogeography-based Optimization (BBO) to classify PD based on a series of biomedical voice measurements. BBO is employed to determine the optimal MLP parameters and boost prediction accuracy. The inputs comprised of 22 biomedical voice measurements. The proposed approach detects two PD statuses: 0-disease status and 1- good control status. The performance of proposed methods compared with PSO, GA, ACO and ES method. The outcomes affirm that the MLP-BBO model exhibits higher precision and suitability for PD detection. The proposed diagnosis system as a type of speech algorithm detects early Parkinson’s symptoms, and consequently, it served as a promising new robust tool with excellent PD diagnosis performance.     

Modelling of Charpy V test rejection probability

Abstract

The purpose of this study was to develop a product design model for estimating the impact toughness of low alloy steel plates. The rejection probability in a Charpy V test is predicted with process variables and chemical composition. Joint modelling of the mean and deviation was used in order to improve the results. The proposed method is suitable for the whole production line, including all grades of steel in production and it is not restricted to a few test temperatures. Using the proposed model the product design group could have recognised most of the rejections before production. Next, the developed model will be implemented into a graphical simulation tool that is in daily use in the product planning department and already contains some other mechanical property models. The model will guide designers in predicting the related risk of rejection and in producing desired properties in the product at lower cost.     

Novel prescribed performance neural control of a flexible air-breathing hypersonic vehicle with unknown initial errors

A novel prescribed performance neural controller with unknown initial errors is addressed for the longitudinal dynamic model of a flexible air-breathing hypersonic vehicle (FAHV) subject to parametric uncertainties. Different from traditional prescribed performance control (PPC) requiring that the initial errors have to be known accurately, this paper investigates the tracking control without accurate initial errors via exploiting a new performance function. A combined neural back-stepping and minimal learning parameter (MLP) technology is employed for exploring a prescribed performance controller that provides robust tracking of velocity and altitude reference trajectories. The highlight is that the transient performance of velocity and altitude tracking errors is satisfactory and the computational load of neural approximation is low. Finally, numerical simulation results from a nonlinear FAHV model demonstrate the efficacy of the proposed strategy.     

Soft sensor for continuous product quality estimation (in crude distillation unit)

Due to the strict norm requirements of keeping products in crude refining units within specifications, laboratory testing and quality control of the products are necessary. Given this reason, virtual soft sensor for continuous quality estimation of light naphtha as the crude distillation unit (CDU) product was developed. Experimental data included available continuous measurements of CDU process streams (temperatures, pressures and flowrate) and laboratory analyses undertaken twice a day. The results are soft sensor models for light naphtha vapor pressure (RVP) estimation.Soft sensor models have been developed conducting multiple linear regression analysis and using neural network-based models such as LNN, MLP and RBF. Considering statistical and sensitivity analysis, the best results for both oils were obtained with MLP and RBF neural networks. The results show possible application of the soft sensor models for estimating light naphtha RVP as an alternative for laboratory testing.     

酪蛋白糖巨肽通过树突细胞发挥免疫调节作用的初步研究

研究用酪蛋白糖巨肽(CGMP)刺激体外培养的小鼠骨髓源树突细胞,观察CGMP对树突细胞成熟的影响,初步探讨CGMP作为功能性食品对免疫的调节作用。实验选用6～8周雄性C57BL/6小鼠,分离培养小鼠骨髓细胞,用重组小鼠白细胞介素-4 (rmIL-4)、重组小鼠粒细胞巨噬细胞集落刺激因子(rmGM-CSF)诱导其分化,在培养的第6天,分别加入RPMI1640完全培养基作为空白对照, 脂多糖(LPS)为阳性对照和0.1、1、10μg/mL的CGMP 3个实验剂量。培养至第8天收集细胞,进行三色荧光标记,用流式细胞仪检测细胞表面标志：CD11c、CD80、CD86、MHC-Ⅱ。另一方面将第8天的树突状细胞(DCs)和T淋巴细胞共孵育进行混合淋巴细胞反应,96h后用Cell Counting Kit-8 (CCK-8)检测T淋巴细胞的增殖。结果表明：3个剂量的CGMP均具有刺激DCs成熟的作用,其中CGMP为1μg/mL时,DCs的各种成熟标志均达到最高水平,其中,CGMP对MHC-Ⅱ的刺激作用最为强烈,而MHC-Ⅱ在DCs对外源抗原的提呈中发挥重要作用;混合淋巴细胞反应结果显示各个不同浓度CGMP作用的DCs均具有不同程度的刺激T淋巴细胞增殖的能力,并且DCs:T为1:10时T淋巴细胞的增殖作用最明显。     

基于变分贝叶斯算法和MLP网络的后非线性混合盲源分离方法研究

传统的后非线性模型往往要求其后非线性函数是可逆的,否则无法进行源信号的分离。然而在实际中,这一要求并不完全满足。针对此不足,结合变分贝叶斯推论和多层感知器网络,提出一种改进的多层感知器后非线性模型,它通过多层感知器来模拟后非线性函数,实现对不可逆后非线性函数混合的盲分离。仿真和实验结果表明该方法是有效的。