采空区危险性的支持向量机识别

针对采空区危险性影响因素与其危险性等级之间存在着复杂非线性关系的特点,笔者提出采用支持向量机最优分类理论来识别采空区的危险性等级。研究选取岩体结构、地质构造、岩石抗压强度、弹性模量、采空区形状、矿体倾角、高跨比、空区体积等8个参数作为主要影响因素,根据支持向量机理论,提出了1-V-1的采空区分类算法,并在Matlab中编程,建立了分类预测的SVM模型。以某矿山的实测采空区为例,利用该模型进行了识别,并与BP神经网络预测结果作对比。实例研究表明,采用该方法的分类结果比神经网络更准确,与采空区调查结果一致性好,用支持向量机理论进行采空区危险性评价是可行的。     

基于遗传算法的BP神经网络在小麦蛋白质含量预测中的应用

为了快速、简便、准确地测定小麦蛋白质的含量,本文提出了应用近红外光谱分析技术结合遗传算法(GA)的BP神经网络的建模方法。采用SPXY算法对光谱数据进行了合理划分,并运用连续投影算法(SPA)将预处理过的数据压缩,对光谱数据提取最佳敏感波点作为GA-BP神经网络的输入,建立小麦蛋白质含量的校正模型。模型的预测均方根误差和预测相关系数为1.3379和0.979,并与BP神经网络所建立的校正模型进行了比较。结果表明：GA-BP神经网络所建模型收敛速度快、训练时间短、准确度也较高,能够实现对小麦蛋白质含量快速高效的检测。     

Using BP Neural Network to Determine the Membership Function of the Input of Fuzzy Control System

While design the fuzzy controller, it is very important to determine the membership function of fuzzy variables.The data can be broadly classified as fuzzy sets by using the classification property of the BP neural network. The author selects a BP neural network with one hide layer and uses S function to the input and hide layer, and linear function to the output layer.Advanced BP algorithm isused to train the BP neural network in the environment of MATLAB . The nearer to the target values is the better the last output is.With the trained BP network , the membership values of the inputs can be got ten. This method has high rate and low error.     

Design on Contingency Classifier of Functioned Link Neural Network

Contingency analysis is thekey computational issue in power system steady state security analysis and reliability calculations. This task requires a large amount of CPU time. In order to reduce effectively requirements of computationin outagesimulation ofbulk power system,a functioned link neural network (FLNN) classified model and algorithm employed to identify contingencies is presented. For the sake of gaining post-accident information of system states, a group of performance index (PI) is designed according to the performance characters relative to the changes of base case.Moreover, a neural network classifieris constructed. A varietyof the effects of PI and combinations of PI on the proposed classifieris discussed. That branch flow performance indices are better than the others is explanted. The resultsof classification by applied the FLNNclassifierto the IEEE-RTS24 show that it not only make network and algorithmsimpler, but also improvethe speed and accurate of contingency analysis.     

Artificial Neural Network Fault Diagnosis System Based on Rough Set Theory

On the basis of fault diagnosis neural network model, in this paper, knowledge representation system of rough set theory is taken as a major tool to delaminate the complex neural network and in which unnecessary properties are eliminated. This method overcomes some shortcomings, such as network scale is too large and the rate of classification is slow. The good effect that reduces the matching quantity of pattern search in classification course is gotten. The structure and algorithm of layered-mining neural network model based on rough set theory are also given. The example shows that this system has higher reference value in practical application.     

基于SSR标记的雪茄烟种质资源指纹图谱库的构建及遗传多样性分析

为从分子水平研究我国雪茄烟种质资源的遗传多样性差异并建立雪茄烟品种的DNA指纹图谱数据库,本研究利用43对多态性好的SSR引物对220份雪茄烟种质进行遗传多样性分析,筛选出14对核心引物对雪茄烟种质进行指纹图谱的构建。结果表明,43对SSR引物在220份雪茄烟种质材料中共扩增出243个等位基因,平均每个标记5.65个,变幅为2~13,每个位点的多态性信息量(polymorphism information content,PIC)变化为0.2078~0.9087,平均为0.6360。有效等位基因数(number of effective alleles,N_e)范围为1.3081~11.7876,平均有效等位基因数为3.9077;观测杂合度(observed heterozygosity,H_o)变化范围为0.0828~0.7639,平均为0.3191;预期杂合度(expected heterozygosity,H_e)的变化范围为0.2361~0.9172,平均为0.6809;种群平均Shannon遗传多样性指数(Shannon genetic diversity index,I)为1.3756,遗传距离在0.0233~0.9286之间,平均遗传距离0.6816。聚类分析表明,在遗传距离为0.74处,可将供试雪茄烟资源分为3个类群。Structure群体遗传结构分析和主成分分析将所有的供试材料划分为2个类群。根据引物的分析和表型鉴定结果,确定良种、辅善和满耳朵,山东大叶和牡丹江05-1,Florida513和CA0701为异名同种,一个品种保留一份种质,剩余216份不同种质。从43对SSR引物中筛选出14对可区分所有供试材料的SSR引物作为核心引物构建了216个雪茄烟品种的指纹图谱。我国雪茄烟种质资源具有较高水平的遗传多样性,本研究构建的雪茄烟种质资源SSR指纹图谱库及遗传分析的结果在分子水平上为筛选、鉴定优质雪茄烟种质资源、挖掘重要基因以及拓宽雪茄烟遗传育种基础等工作提供科学依据。     

Mouse with Capablility of Fingerprint Identifying

The mouse is developed with the purpose of providing reinforcement of database security for computer users, whose validity can be verified by taking advantage of the fingerprint identity, which is the uniqueness, immutability biological characteristic of human body. By one-by-one matching algorithm, rendering between pressured data and field acquisitions of finger print image, the vital heart of dactylogram verification with the mouse is realized. The mouse has been made possible by taking account of the roomy space within mouse,tiny chip patch for fingerprint identification and together with small size of available microprocessors. The mouse can be used as an normal pointing device while maintaining the capability of user identity distinguishing. The usage of this device is very easy and it is reliable. Development of this device mainly comprises of the hardware development, driver development, application software development, and the installation of network security verification center. The purpose, principle and system design of the mouse is introduced, while characteristics and applications of the device are also presented.     

Nonlinear Forecast System for River Water Pollution Based on GIS

Based on the analysis of the water pollution spatial distribution characters of Yangtze River in Chongqing,a new method based on the integration of BP neural network and genetic arithmetic(GA) is proposed.For some shortcomings existed in the standard BP neural network,this method has ultimately overcome these shortcomings by combining the GA with BP artificial neural network through altering stimulating function,adding momentum factor to power value for BP algorithm and introducing genetic arithmetic to searching for the knots of the hidden layer,momentum factor and learning level.Using this method can easily overcome the difficulty of measuring the water prediction model's parameters.GIS is used as a tool for data management and spatial analysis,and the prediction result of the model for the water pollution spatial distribution characters of Yangtze River in Chongqing is visualized and explored with the precision of more than 78%.     

Polynomial Feedforward Artificial Neural Network

A new artificial neural network,i.e.,polynomial feedforward artificial neural network(PFANN), which has three layers(input layer,hidden layer and output layer) is presented. The neural activation functions of hidden layer and output layer are f(x)=x p and linear function, respectively. The learning method of hidden-output layer weights is the steepest descent method and the one of input-hidden layer weights is genetic algorithm(GA) . During the learning process, the error function is decreased monotonely. So the learning algorithm is convergent and the network ,which can approximate to arbitrary continuous function , is stable. Some applying samples of PFANN, which reveales the remarkable quality, are proposed,too.     

BP神经网络在储粮害虫分类中的应用研究

综合利用计算机视觉技术和BP神经网络技术,实现对粮仓害虫的无损检测.通过对粮仓害虫图像的CCD图像预处理,获取了几何特征和不变矩等15个特征参数,并通过优化选取其中七个参数输入神经网络进行训练.仿真结果表明训练网络对粮仓四类常见害虫的识别率达到了85%,得到了较好的识别结果.     

大型商场建筑夏季冷负荷动态预测模型

夏季建筑冷负荷的正确预测是实现大型复杂中央空调优化运行、节能降耗的关键。笔者探讨了商场建筑冷负荷的主要影响因素,确定了建筑动态冷负荷预测模型的输入,提出了夏季基于新风机组供电频率的商场顾客率间接测量方法,解决了商场内顾客量难以检测的难题。还提出了AFC-HCMAC神经网络预测模型算法,实现了大型商场建筑冷负荷的动态预测。仿真结果表明:顾客率在商场冷负荷预测中占有重要地位,在冷负荷预测模型中增加商场顾客率可显著提高预测精度;AFC-HCMAC神经网络预测算法与传统的HCMAC神经网络算法比较,可有效降低神经网络节点数,提高预测精度。     

Application of Incremental Genetic RBF NN in Converter Vanadium Recover

Converter vanadium recover is a very sophisticate reaction which is diverse and non- line. From the point of view of statistics and reaction mechanism, it is difficult to build up end- point control static model. Aim at this problem, the paper puts forward a model identify method based on incremental genetic RBF neural network to build up such a model, which can perfectly resolve the problem of random selection of RBF cluster center number and sample data clustering. Furthermore, in order to ensure structure of neural network to fit with continuous incremental data set, the paper presents a method of incremental data dealing, which is applied to amend the parameters of neural network. Then the request of continuous production is satisfied. Finally the result of test shows that after adopting the algorithm, the error of result is less than before and end- point hitting ratio satisfies to ninety percent. These indicate the algorithm has the engineering practicability.     

基于深度卷积神经网络的玉米病害识别

为了提高玉米病害的识别率,本文提出了一种在自然环境条件下基于深度卷积神经网络的玉米病害识别方法。该方法以玉米常见的10类病害为研究对象。算法模型是先将图像预处理,应用Triplet loss双卷积神经网络结构学习玉米图像特征,再使用SIFT算法提取图像纹理细节,最后通过Softmax对图像进行标签分类。训练集采用正常玉米图像与玉米病害图像相结合的方式,使用深度相似性网络学习正常玉米图像特征表示,再使用迁移学习方法学习玉米病害图像的特征,最后对特征进行分类识别。研究结果表明,该方法可准确识别10种常见玉米病害,正确率可达90%以上,为玉米病害的防治提供了有效的技术支持。     

甜菜品种SSR指纹图谱的构建及遗传多样性分析

筛选出20对多态性高、带型清晰、重复性好的SSR标记引物对供试甜菜品种进行PCR扩增,根据扩增产物构建SSR指纹图谱,并分析其遗传多样性。结果显示,共检测出112个等位基因,平均每对引物5.6个,利用获得的等位基因计算遗传距离,107个品种的遗传距离变化范围在0.065~0.467之间,平均遗传距离0.298。Shannon’s多样性指数变异范围0.78~2.90;PIC值介于0.08~0.83;Nei’s指数介于0.39~1.87。利用类平均法（UPGMA）进行聚类分析,可将107个甜菜品种分为2个类群。类群Ⅰ29个品种,类群Ⅱ78个品种。类群Ⅰ和Ⅱ又可分为多个亚群。结果表明,每个甜菜品种有区别于其他品种唯一的数字指纹,说明用于试验的20对SSR标记适用于甜菜品种真实性的鉴定,同时甜菜品种指纹图谱库的构建也为甜菜品种鉴定提供技术基础。     

棉花DUS测试标准品种的SSR指纹数据库构建

基于SSR核心引物,采用荧光毛细管电泳检测系统与多重PCR技术相结合,构建我国棉花DUS(Distinctness,Uniformity and Stability)测试标准品种的DNA指纹数据库并进行遗传多样性分析。依据多重PCR组合的基本原则与五色荧光检测系统的特点,采用40对核心引物构建了10个4重PCR组合,利用DNA遗传分析仪进行指纹检测与数据采集。40对荧光引物在30份DUS测试标准品种中共扩增产生146个等位变异,每对引物的等位变异数为2~7个不等,平均每对引物产生3.65个等位变异。海7124与陆地棉品种明显划分为2类,来源于新疆的新陆早1号区别于其他陆地棉品种,单独聚为1类。多色荧光检测系统相比常规聚丙烯酰胺凝胶电泳具有高精度、高通量、自动化程度高的优点,尤其适用于大规模指纹数据库的构建,提出了通过构建已知品种DNA指纹数据库,将分子标记技术应用于棉花DUS测试的初步设想。     

The Study on the Branched Feedforward Neural Network

Based on the BP algorithm,the Branched Feedforward Neural Network,one ofthe artificial neural network medel used for pattern classification is proposed in this paper, and therelated algorithm is also offered here, Several typical examples of pattern classification are studiedwith computer simulation, The simulation results show that the training time of Branched Feedforward Neural Network is obviously reduced and the classifying effect is much better as compared withgeneral BP Network.     

Human Face Location Based on Motion Information in a Complex Background

This paper proposes an improvement for the traditional Optical Algorithm, and presents a new way to image segmentation in a complex background. In addition, combined with the neural network, the system can locate the possible human faces successfully by means of two-step location model. In our system, the searching and locating of the human face is the most important stage. According to this, the authors adopt the two-step way to run, firstly they take up the segmentation of the candidate human face areas and then the accurate face locating based on the neural network is used. This algorithm is fast and robust. Experimental results with real scene images are given out there, and all these prove that two-step method gains many advantages in the course of human face location with motion information, such as real-time, robustness and practicality. In addition, the proposed system is also the fundamental and important part of the perfect human face recognition system.     

A radial basis function neural network based self adapting predictive decoupling control system for gas collector pressure in coke ovens

Pressure is measured at different levels in the loop layer because self adapting predictive decoupling control systems are strongly coupled, disturbed, and non linear and there is a long time delay for gas collector pressure systems in coke ovens. By combing the traditional neural network control and proportional integral differential(PID) controllers based on radial basis function(RBF) neural network identification, the gas collector pressure is ensured to reach the desired technology range. The prediction model of an RBF neural network is used for advanced prediction of the actual output pressure to overcome delays in general gas collection. The simulation results and application indicate that the method can obtain ideal control results.     

Improved Ellipsoidal unit Neural Networks and Its Applications in CSTR

To overcome the limitations of the standard ellipsoidal unit neural networks, some new approaches used in ellipsoidal unit neural networks have been proposed. These new approaches address three main issues: firstly, to understand better and represent the nature of fault classification boundaries; secondly, to determine the network structure without the usual trial and error schemes; lastly, to avoid erroneous generalizations. The application in CSTR shows that the ellipsoidal unit networks can possess arbitrary nonlinear classifying ability, nonlinear interfacial describing ability, and obtain accurate and efficient diagnosis results.