The Inherent Fault Tolerance of Multi-Layer Neural Networks

The inherent fault tolerance of multilayer neural networks (MLNs) is due to thedistributed properties of information processing and information storage in an MLN , and the inher-ent fault-masking capability of every neuron. Also,the concept of inherent fault tolerance of MLNsis formalized, Finally, perceptrons and three-layer nets are shown inherently fault tolerant underreasonable statistical conditions,respectively.     

The Convergence Rates of Improveed Kernel Estimators for Regression Function

The kernel function of kernel estimator for regression function is often defined with compact support, the moment of Y is bigger than one. The convergence rate of the improved kernel estimator is discussed under the improved kernel function. The result is also discussed under random censorship. The class of applicable kernels include those having unbounded support and even not integral .The condition of moment of Y is discarded under the improved kernel estimator. The results are wholly the same as usual and generalize the relative results of the paper .     

Synthetic Assessment of Power Transformer Condition Based on Matter-element Theory

Condition assessment on power transformer is an important approach to improve reliability of transformer and provide proper decision-making for condition-based maintenance. For long, static assessment method used in the past has already not adapted to the development of modem electricity transmission and distribution technology. So based on the deficiency of fuzzy synthetic assessment method, matter-element synthetic assessment method is introduced. The method can evaluate transformer condition not only in quality but also in quantity. Because each index weight can be given fairly and fast through improved analytic hierarchy process, which need not carry on consistency chec accuracy of assessment can be improved. Examples analysis has proved that assessment method of power transformer condition based on matter-element theory is rational and effective.     

Automatic Fingerprint Classification by GA-Based Neural Network

Fingerprint classification can provide an important indexing mechanism in a fingerprint database. An accurate and consistent classification can greatly reduce fingerprint matching time for large database. In the paper, by combining genetic algorithm and neural network is presented a fingerprint classification algorithm which is able to achieve an accurate classification. By inputting the global feature represented by directional image to three layer neural network trained by genetic algorithm, the fingerprints were classified into six categories: whorl, right loop, left loop, arch, double loop and undiscerning type successfully.     

A Muti-Objective Network Planning Model for Project

Based on the network planning model developed by the author for an overhaul project of a rolling mill production line in a heavy steel plant, an new multiobjective network planning model is proposed in this paper. The formulations of network time parameters are derived. The optimization problem of multiobjective network planning is also discussed. And it has been found out that the fractional single-objctive network is only one of the cases in this model.     

A MULTI -OBJECTIVE FUZZY DECISION -MAKING APPROACH FOR POWER NETWORK PLANNING UNDER UNCERTAINTY

This paper proposes a multi - objective fuzzy decision - making approach for power network planning under uncertainty ,which simutaneously considers not only the least investment cost,the minimum power loss,the maximum reliability,and the least environmental impacts, but also the uncertainty about the future load growth and the capital investment availability. The validity and effectiveness of the proposed approach is verified with numerical examples.     

The Method of Diagnosing Brain-dysfunction Based on Membership Function and BP Neural Network

Membership functions.formed with the characteristic values obtained by applying the continuous reaction time (CRT) theory from controls and heptic encephalopthy and proposedin this paper.whith these membership functions,the CRT data to be diagnosed are calculated in advance,and then discriminated by BP meural network to differentiate patients with or without braindysfunction.The troubles of low accuracy and efficiency,encountered in training BP network withfuzzy samples.scattered data and extended samples,are solved.     

基于SOFM神经网络的核心种质取样算法

介绍了核心种质和自组织特征映射网络的基本概念，使用自组织特征映射网络的自组织特性对种质性状数据进行了聚类，利用聚类的分组结果，提取核心种质，通过对核心种质的各种参数的检验并与完全随机取样和一般的系统聚类的对比。结果表明，SOFM的取样算法，是一种比较有效快捷的算法，可以在实际中加以应用。     

基于BP神经网络的信息融合发动机故障诊断的研究

为了解决发动机喷油器故障诊断中基于单传感器信息的方法诊断精度低的缺点，在神经网络分析的基础上，提出了一种基于气缸压和、缸盖振动信号和燃油压力等多传感器信息融合的喷油器故障诊断新方法。该方法能有效地提高其故障诊断精度，明显增加了诊断过程的准确性和智能化。