电磁兼容的人工神经网络预测技术分析

电磁兼容预测是实现电子设备或系统电磁兼容性(electromagnetic compatibility,EMC) 的必要步骤.提出了应用人工神经网络对电磁兼容问题进行快速预测的方法.通过选择有效的电磁干扰参数作为输入预测因子,用误差反向传播的神经网络 (back propagation,BP)构造输入预测因子与敏感设备骚扰响应之间的映射关系,并用电磁场数值计算方法获得的训练样本集和测试样本集对构造好的BP网络进行训练,建立了基于BP网络的电磁兼容快速预测模型.最后以导线间的串扰问题为预测算例,表明了该方法的有效性.     

一种基于衰减系数的混沌时间序列预测法

为提高加权一阶局域模型的预测精度,提出一种改进型混沌时间序列预测方法.该方法用衰减系数和时间延迟修正向量距离公式,调节邻近点与中心点的相关性,同时,只用邻近点中与预测值相关性最大的分量进行线性拟合.利用该方法对Henon混沌时间序列进行预测的结果表明,衰减系数取最佳值时,相对于现有算法,该方法可以更精确地预测混沌时间序...     

Finds in Testing Experiments for Model Evaluation

To evaluate the fault location and the failure prediction models, simulation-based and codebased experiments were conducted to collect the required failure data. The PIE model was applied to simulate failures in the simulation-based experiment. Based on syntax and semantic level fault injections, a hybrid fault injection model is presented. To analyze the injected faults, the difficulty to inject (DTI) and difficulty to detect (DTD) are introduced and are measured from the programs used in the code-based experiment. Three interesting results were obtained from the experiments: 1) Failures simulated by the PIE model without consideration of the program and testing features are unreliably predicted; 2) There is no obvious correlation between the DTI and DTD parameters; 3) The DTD for syntax level faults changes in a different pattern to that for semantic level faults when the DTI increases. The results show that the parameters have a strong effect on the failures simulated, and the measurement of DTD is not strict.     

基于混合模型的多重故障诊断

介绍了基于模型的多重故障诊断，它可通过ＡＴＭＳ来实现，能克服基于规则的诊断系统的一些局限性，但当运行模型和故障模型同时被运用或者等级模型被运用时，普通ＡＴＭＳ不能处理这种情况．本文应用分层ＡＴＭＳ来实现基于混合模型的故障诊断方法，同时给出广义最小诊断的定义，从而克服关于最小诊断定义中的不足．由于分层ＡＴＭＳ能表示和处理论据之间的优先级，因而能运用混合模型，能结合基于模型和基于规则的诊断，及提高诊断效率．     

川东北宣汉-达县地区储层预测技术

川东北宣汉-达县地区以构造圈闭为主要勘探对象,钻探成功率低,为此开展了以储层预测技术为主导的构造-岩性复合圈闭勘探.通过山区地震数据高分辨率采集、处理、储层沉积相研究、储层测井响应特征、储层物性特征及储层层位标定、地震相研究、储层地震预测方法研究,确立了研究区内的三大储层类型及各类储层的研究方法,提出了川东北复杂储层预测以野外地震资料高分辨率采集为基础、多种特殊处理手段相结合,并以沉积相研究为指导的综合性研究思路,为川东北宣汉-达县地区储层预测方法研究提供了指导性意见,为该地区复杂储层勘探的深入与突破打了坚实的基础.     

浙江省地壳形变监测网络的构建及其应用前景展望

近年来,GNSS技术取得了长足进展,特别是GNSS的相对定位精度有了显著提高.GNSS技术已经成为监测地壳形变和板块运动的主要手段,它可以提供各活动构造带地壳水平运动的可靠结果,建设浙江省的地壳形变监测观测网,可获取大量监测区域地壳形变震兆信息,对于研究浙江省的地壳形变运动,监测板块边缘及内部的构造变形,提高浙江省地震预报水平有很重要的意义.     

Predicting the spread of nuclear radiation from the damaged Fukushima Nuclear Power Plant

Japan suffered a M9.0 earthquake and massive tsunami on March 11, 2011, which seriously damaged the Fukushima Nuclear Power Plant and caused a nuclear crisis. The spread of nuclear radiation from the power plant through the atmosphere and ocean was predicted with a short-term climate forecasting model and an ocean circulation model under some idealized assumptions. If nuclear matter were leaked in the near-ground layer of 992 hPa, the climate model results show that the nuclear radiation would cover North America 10 days after the initial leakage, with the concentration at the forefront dramatically reduced to 10 millionths of the initial model concentration at the source. The radiation would span Europe in 15 days and cover much of the Northern Hemisphere in 30 days. If the initial leakage was assumed to occur in the layer 5000-m above the ground, the radiation would cover Europe in 10 days and cover much of the Northern Hemisphere in 15 days. Moreover, under the assumption that the nuclear matter leaked in the 10000-m layer, the radiation would affect much of China after 10 days. The ocean circulation model indicates that the nuclear material would be slowly transported northeast of Fukushima and reach 150°E in 50 days, and the nuclear debris in the ocean would be confined to a narrow band. Compared with the spread in the ocean, the area affected by leaked nuclear radiation in the atmosphere would be very large. Atmospheric monitors in North America and Europe will be helpful for estimating the effect in China of any leaked nuclear material.     

基于Attention机制的LSTM股价预测模型

针对时间序列分析方法和神经网络对于股价预测具有一定局限性的问题,将基于Attention机制的LSTM模型应用于股价预测;以2014-01-02—2020-09-22日的上证工业指数、上证环保指数等相关数据为样本,在LSTM模型中引入Attention机制,使模型聚焦于重要的股价特征信息,预测股票第二日的最高价;实证研...     

一类线性离散切换系统的容错控制

文章研究了一类线性离散切换系统的容错控制问题;当执行器失效或部分失效时,利用Lyapunov函数法,建立了切换闭环系统混杂状态反馈容错控制器存在的充分条件;运用线性矩阵不等式将容错控制器设计问题转化为一组线性矩阵不等式的可行解问题,可借助Matlab中线性矩阵不等式工具箱求解;通过数值算例,验证所提出设计方法的有效性。     

弓型折流板换热器动态响应神经网络预测

采用ARX动态模型设计了一种用于换热器动态特性辨识及预测的人工神经网络结构模型,基于有限的实验数据将人工神经网络技术应用于以水、油为换热工质的弓型折流板换热器动态特性的预测当中.采用Levenberg-Marquardt算法对网络进行训练和测试,辨识和预测了换热器单侧流量扰动情况下油侧出口温度的响应情况及换热器油侧进口发生温度扰动情况下油侧出口温度响应情况,并与数值计算进行了对比,神经网络的预测结果好于数值计算.分析了神经网络模型的泛化能力.计算结果表明,人工神经网络对于复杂系统的动态辨识及预测是相当成功的,与基于数理模型的数值预测相比,有更好的预测精度,且模型的泛化能力也很强.