基于自适应神经模糊推理系统的板形模式识别研究

 针对传统的最小二乘板形模式识别方法的抗干扰能力差、精度低和神经网络板形模式识别方法存在的网络学习时间长、易陷入局部最小值、应用效果不佳等问题,提出了一种基于ANFIS的板形模式识别方法。该方法融合了模糊理论和神经网络的优点,弥补了彼此的不足,有效的解决了上述问题。板形模式识别是一个多输出系统,而MATLAB中ANFIS指令仅有一个输出,针对这个问题,本文提出利用系统拟合的方法,有效的克服了这个缺陷。研究结果表明,该方法能有效识别出常见的板形缺陷,识别速度和精度有所提高,识别结果跟板形仪的实测板形非常接近。     

支持向量机改进及其在钢材质量管理建模中的应用研究

针对支持向量机参数难以选择的问题,提出了基于差分进化算法（DE）的参数选择方法,算例分析结果表明DE算法选择SVM参数有着迭代次数少、结果稳定的优点,能够很好的解决SVM的参数选择问题。随后将基于DE算法选择参数的支持向量机应用于一个钢材质量管理的建模实例中,并将其与神经网络建模方式所得结果相比较,结果表明经改进的支持向量机的预测性能更加优秀。     

基于LR-ANN-SVM的滑坡易发性评价

针对传统大数据机器学习等方法进行滑坡易发性评价时,存在过于追求模型评价精度,导致在中易发区与低易发区存在滑坡产生的风险,提出了风险预警来降低中与低易发区产生的滑坡灾害。选取神经网络模型（ANN）、逻辑回归模型（LR）、支持向量机模型（SVM）3种学习方法,对上犹县进行滑坡易发性评价,将上犹县分为高易发区、较高易发区、中易发区、较低易发区,低易发区。由受试者工作曲线（ROC）下的面积（AUC)显示:神经网络（ANN）的AUC=0.939, 逻辑回归模型（LR）的AUC=0.897, 支持向量机（SVM）的AUC=0.884,均具有较高的评价精度。根据以上的易发性评价结果,得到上犹县栅格的易发性指数（LSI）,然后基于MAX（LSI（LR）、LSI（ANN）、LSI（SVM））函数对上述模型的易发性指数取最大值,并对上犹县进行滑坡易发性评价。结果显示:LR-ANN-SVM的AUC=0.815,有较高的易发性评价精度。从高易发区与较高易发区所含滑坡占比来看,LR、ANN、SVM、LR-ANN-SVM的滑坡占比分别为80.6%、74.6%、91%、93.2%,表明根据ANN-LR-SVM易发性分区治理更安全。      

SVM在齿轮小样本故障诊断中的应用

将SVM的分类算法应用于齿轮小样本故障诊断中。选取识别能力好的时域无量纲指标和频域中的9段频谱融合作为支持向量机的特征矢量,对齿轮的三种典型故障进行分类,结果表明:SVM在解决小样本情况下的机械故障诊断的分类问题中具有良好的应用前景。     

支持向量机理论与算法研究综述

统计学习理论(statistical learning theory,SLT)是一种小样本统计理论,着重研究在小样本情况下的统计规律及学习方法性质.支持向量机(support vector maehinse,SVM)是一种基于SLT的新型的机器学习方法,由于其出色的学习性能,已经成为当前机器学习界的研究热点.该文系统介绍了支持向量机的理论基础,综述了传统支持向量机的主流训练算法以及一些新型的学习模型和算法,最后指出了支持向量机的研究方向与发展前景.     

板形模式识别的GA-BP模型和改进的最小二乘法

针对板宽变化时需要不同拓扑结构的神经网络才能完成板形模式识别任务，网络学习工作量大，网络存在收敛速度慢，易陷入局部极小等结构性能不佳的问题，首次建立了以勒让德正交多项式为基模式的只用6个输入信号、3个输出信号的板形模式识别GA-BP网络模型。该模型不仅结构简单，而且物理意义明确，识别精度较高，解决了板宽变化时神经网络结构形式不变的问题，从而实现了板形模式识别的智能化。又提出了基于勒让德正交多项式的板形模式识别最小二乘法，该方法简单、实用，识别精度较高，克服了传统的最小二乘模型板形模式识别的缺点和不足。为板形模式识别提供了两种简便实用的新方法，发展了板形模式识别理论和方法。     

支持向量机动态学习方法及其在票据识别中的应用

介绍了用支持向量机(SVM)进行动态学习训练的方法．解决了在机器学习过程中,训练样本获取比较困难,样本可随外界条件改变而变化的问题．实践证明,使用该方法可以动态跟踪样本的变化,保证SVM分类器的最优性能．利用该方法设计的银行票据OCR系统的实际应用说明了该方法的有效性．     

基于支持向量机的围岩定性智能分级研究

本文将数据挖掘的新方法支持向量机应用于隧道围岩分级.支持向量机是一种基于统计学习理论的新的学习算法,比神经网络算法能更好地解决小样本问题.选用岩层厚度、岩体结构、嵌合程度、风化程度、地下水特征、节理发育程度、榔头敲击声和地应力等 8 个定性指标作为评判因子,用泥巴山隧道采集的实际数据作为样本对不同核函数的支持向量机进行训练,并得到评判因子与围岩级别的映射关系,从而可以对未知的围岩样本进行级别判别.判别结果表明:采用多项式核的支持向量机对围岩级别进行判别有较高的准确率,是一种值得推广和应用的围岩智能分级方法.     

核函数方法(上)

支撑矢量机的成功引起了人们对核函数方法的兴趣。通过某种非线性变换将输入空间映射到一个高维特征空间，如果在其中应用标准的线性算法时，其分量间的相互作用仅限于内积，则可以利用函数的技术将这种算法转换为原输入空间里的非线性算法。Fisher判别法和主分析法是在模式分类与特征抽取中已经获得广泛应用的传统线性方法，近年出现的基于核函数的Fisher判别（KFD）与基于核函数的主分量分析（KPCA）是它们的非线性推广，其性能更好，适用范围更广，灵活性更高，是值得关注的应用前景看好的新技术。     

基于PCA－SVM的采空区危险性评价

介绍了主成分分析法和支持向量机算法。建立两种方法相结合的PCA－SVM模型。通过工程实例建立PCA－SVM评价模型,并利用样本数据对模型预测结果进行检测。能够较准确地预测采空区的危险性。该模型在矿山领域具有很好的应用前景。     

Neural networks predict response biases of female túngara frogs

Artificial neural networks have become useful tools for probing the origins of perceptual biases in the absence of explicit information on underlying neuronal substrates. Preceding studies have shown that neural networks selected to recognize or discriminate simple patterns may possess emergent biases toward pattern size of symmetry--preferences often exhibited by real females--and have investigated how these biases shape signal evolution. We asked whether simple neural networks could evolve to respond to an actual mate recognition signal, the call of the túngara frog, Physalaemus pustulosus. We found that not only were networks capable of recognizing the call of the túngara frog, but that they made remarkably accurate quantitative predictions about how well females generalized to many novel calls, and that these predictions were stable over several architectures. The data suggest that the degree to which P. pustulosus females respond to a call may often be an incidental by-product of a sensory system selected simply for species recognition.     

Overview of the Application of “Fuzzy Techniques” in Construction Management Research

During the last decade, “fuzzy techniques” have been increasingly applied to the research area of construction management discipline. To date, however, no paper has attempted to summarize and present a critique of the existing “fuzzy” literature. This paper, therefore, aims to comprehensively review the fuzzy literature that has been published in eight selected top quality journals from 1996 to 2005, these being Journal of Construction Engineering and Management, ASCE; Journal of Management in Engineering, ASCE; Construction Management and Economics; Engineering, Construction and Architectural Management; International Journal of Project Management; Building Research and Information; Building and Environment; and Benchmarking: An International Journal. It has been found that fuzzy research, as applied in construction management discipline in the past decade, can be divided into two broad fields, encompassing: (1) fuzzy set/fuzzy logic; and (2) hybrid fuzzy techniques, with the applications in four main categories, including: (1) decision making; (2) performance; (3) evaluation/assessment; and (4) modeling. The comprehensive review provided in this paper offers new directions for fuzzy research and its application in construction management. Based on a comprehensive literature review on the applications of fuzzy set/fuzzy logic, and hybrid fuzzy techniques in construction management research, an increasing trend of applying these techniques in construction management research is observed. Therefore, it is suggested that future research studies related to fuzzy techniques can be continuously applied to these four major categories. Fuzzy membership functions and linguistic variables in particular can be used to suit applications to solving problems encountered in the construction industry based on the nature of construction, which are widely regarded as complicated, full of uncertainties, and contingent on changing environments. Moreover, hybrid fuzzy techniques, such as neurofuzzy and fuzzy neural networks, can be more widely applied because they can better tackle some problems in construction that fuzzy set/fuzzy logic alone may not best suit. For example, neural networks are strong in pattern recognition and automatic learning while fuzzy set and fuzzy logic are strong in modeling certain uncertainties. Their combination can assist in developing models with uncertainty under some forms of pattern. Finally, an increasing trend of applying fuzzy techniques in the building science and environmental disciplines is also observed; it is believed that the application of fuzzy techniques will go beyond the construction management area into these disciplines as well.     

A Neural Network Recognition Method of Shape Pattern

The final objective of shape control is to makethe output shape data achieve the objective shapedata,then the adjusting magnitude of shape controlactuating mechanism can be defined in accordancewith the deflection value of the output shape dataand the objective shape data.In shape controlsystem,the first thing is to make shape faultpattern recognition based on the speciality andrequirement of shape control actuating mechanism ofrolling mill.The main assignment of shape patternrecognition is to…     

The application of neural networks in predicting the outcome of in-vitro fertilization

Infertility affects one in six couples at some time in their lives, with 48% of these couples requiring assisted conception techniques in order to achieve a pregnancy. Whilst the overall clinical pregnancy rate per embryo transfer is 23%, this varies widely between clinics. The Human Fertilisation and Embryology Authority has attempted to analyse the results of all units, with weighting of different factors affecting assisted conception, and the published data have invariably led to comparisons between units. However, statistical models need to be developed to eliminate bias for valid comparisons. Neural networks offer a novel approach to pattern recognition. In some instances neural networks can identify a wider range of associations than other statistical techniques due in part to their ability to recognize highly non-linear associations. It was hoped that a neural network approach may be able to predict success for individual couples about to undergo in-vitro fertilization (IVF) treatment. A neural network was constructed using the variables of age, number of eggs recovered, number of embryos transferred and whether there was embryo freezing. Overall the network managed to achieve an accuracy of 59%.     

A New Flatness Pattern Recognition Model Based on Cerebellar Model Articulation Controllers Network

In the traditional flatness pattern recognition neural network, the topologic configurations need to be rebuilt with a changing width of cold strip. Furthermore, the large learning assignment, slow convergence, and local minimum in the network are observed. Moreover, going by the structure of the traditional neural network, according to experience, the model is time-consuming and complex. Thus, a new approach of flatness pattern recognition is proposed based on the CMAC （cerebellar model articulation controllers） neural network. The difference in fuzzy distances between samples and the basic patterns is introduced as the input of the CMAC network. Simultaneously, the adequate learning rate is improved in the error correction algorithm of this neural network. The new approach with advantages, such as high learning speed, good generalization, and easy implementation, is efficient and intelligent. The simulation results show that the speed and accuracy of the flatness pattern recognition model are obviously im proved.     

硅钢退火网络型专家系统

介绍一种用于硅钢退火工艺的网络型专家系统。该系统将人工神经网络、分类模式识别技术和有关冶金物理化学知识相结合 ,建立了优化数学模型 ,以C语言和多媒体技术实现程序设计。现场生产运行表明该系统取得了预期效果。     

Model Induction with Support Vector Machines: Introduction and Applications

The rapid advance in information processing systems in recent decades had directed engineering research towards the development of intelligent systems that can evolve models of natural phenomena automatically—“by themselves,” so to speak. In this respect, a wide range of machine learning techniques like decision trees, artificial neural networks (ANNs), Bayesian methods, fuzzy-rule based systems, and evolutionary algorithms have been successfully applied to model different civil engineering systems. In this study, the possibility of using yet another machine learning paradigm that is firmly based on the theory of statistical learning, namely that of the support vector machine (SVM), is investigated. An interesting property of this approach is that it is an approximate implementation of a structural risk minimization (SRM) induction principle that aims at minimizing a bound on the generalization error of a model, rather than minimizing only the mean square error over the data set. In this paper, the basic ideas underlying statistical learning theory and SVM are reviewed, and the potential of the SVM for feature classification and multiple regression (modeling) problems is demonstrated by applying the method to two different cases of model induction from empirical data. The relative performance of the SVM is then analyzed by comparing its results with that of ANNs on the same data sets.     

Animats: computer-simulated animals in behavioral research

The term animat refers to a class of simulated animals. This article is intended as a nontechnical introduction to animat research. Animats can be robots interacting with the real world or computer simulations. In this article, the use of computer-generated animats is emphasized. The scientific use of animats has been pioneered by artificial intelligence and artificial life researchers. Behavior-based artificial intelligence uses animats capable of autonomous and adaptive activity as conceptual tools in the design of usefully intelligent systems. Artificial life proponents view some human artifacts, including informational structures that show adaptive behavior and self-replication, as animats may do, as analogous to biological organisms. Animat simulations may be used for rapid and inexpensive evaluation of new livestock environments or management techniques. The animat approach is a powerful heuristic for understanding the mechanisms that underlie behavior. The simple rules and capabilities of animat models generate emergent and sometimes unpredictable behavior. Adaptive variability in animat behavior may be exploited using artificial neural networks. These have computational properties similar to natural neurons and are capable of learning. Artificial neural networks can control behavior at all levels of an animat's functional organization. Improving the performance of animats often requires genetic programming. Genetic algorithms are computer programs that are capable of self-replication, simulating biological reproduction. Animats may thus evolve over generations. Selective forces may be provided by a human overseer or be part of the simulated environment. Animat techniques allow researchers to culture behavior outside the organism that usually produces it. This approach could contribute new insights in theoretical ethology on questions including the origins of social behavior and cooperation, adaptation, and the emergent nature of complex behavior. Animat studies applied to domestic animals have been few so far, and have involved simulations of space use by swine. I suggest other applications, including modeling animal movement during human handling and the effects of environmental enrichment on the satisfaction of behavioral needs. Appropriate use of animat models in a research program could result in savings of time and numbers of animals required. This approach may therefore come to be viewed as both ethically and economically advantageous.     

Some applications of neural networks for prediction of blast furnace irregularities

The on-line analysis of operational data and prediction of furnace irregularities, though difficult, are essential for the improvement of the control of blast furnace operation. Three models based on artificial neural networks for the recognition of top gas distribution, distributions of the heat fluxes through the furnace wall, and for the prediction of slips have been designed. The off-line test results showed that a trained perceptron network could recognise various types of top gas profiles. A classifier consisting of a self-organising feature map network and a learning vector quantizer could classify the characteristic patterns of heat flux distribution; and a model based on a back propagation network could properly predict the probability of upcoming slips in advance. The most important operational variables needed for predicting slips have also been extracted. It has been proved that the neural network used has a good capability of predicting furnace irregularities.