一个基于故障树模型的诊断系统

提出一种基于故障树模型的诊断方法，并基于该方法实现了一个验证性的诊断系统。基于故障树的层次诊断模型，提出了确定性推理方法和可能性推理方法及面向故障树的基于框架和广义规则的混合知识表示方法。最后在ＣＯＭ－ＰＡＱ４８６微机上，Ｗｉｎｄｏｗｓ环境下，用ｂｏｒｌａｎｄＣ＋＋实现了该诊断系统，并通过对一个电路系统实验台的诊断验证了方法的有效性。     

应用加权中值滤波法检测薄壁梁式挠性器件下摆量

应用加权中值滤波法对薄壁梁式挠性器件下摆量进行高准确度CCD检测。薄壁梁式挠性器件在测量过程中测量图像存在高频抖动现象,因此测量准确度不高,运用加权中值滤波后使用最小二乘法细分技术,可以在实现提高测量准确度的同时,消除高频抖动带来的误差,从而可以有效提高CCD测量薄壁梁式挠性器件下摆量的准确度。实验表明,该方法比传统测量方法准确度提高20倍以上。     

无刷直流电动机的新型自适应模糊神经控制

 为无刷直流电动机提出了一种自适应模糊神经控制方法.这是一种建立在开关控制、模糊控制和自适应控制相结合基础上的控制方法,并用神经网络实现了模糊控制器和自适应机构.在无刷直流电动机的双闭环调速系统中,电流控制器是PI控制器;转速控制器是由1个开关控制器和1个包括自适应机构在内的模糊控制器相结合组成的,且用1个3层前向神经网络离线学习实现了模糊控制器,学习算法采用的是改进的BP算法.用1个单神经元通过在系统运行过程中的动态学习实现了自适应机构,学习算法选用了有监督的Hebb学习算法.由电机所处的运行阶段决定哪一个控制器工作.此控制算法的仿真结果说明,它使系统具有良好的动、静态特性和自适应性.     

同无黑子耀斑相协的微波爆发机制的探讨

1980—1984年期间,我们在云南天文台9375MHz,3653MHz和2902 HHz三个波段的射电观测资料中,查到11组微波爆发与无黑子耀斑相伴随。本文研究了这些爆发与耀斑的大气层高度,射电爆发类型以及通过落进峰值流量-时间图的位置,确认出爆发机制是活动区中作麦克斯韦分布电子的热轫致辐射。     

Image restoration of the asymmetric point spread function of a high-resolution remote sensing satellite with time-delayed integration

The calibration and validation team of the Korea Aerospace Research Institute has calibrated and validated the image data of the KOMPSAT-2 since the launch of KOMPSAT-2 on July 28, 2006. The asymmetric phenomenon of the point spread function of the KOMPSAT-2 image data is evident in both the along and the across direction, most likely because KOMPSAT-2 has a 1 m ground sample distance high-resolution camera with time-delayed integration. Furthermore, because KOMPSAT-2 is in space, the KOMPSAT-2 image data has been corrected with good results by means of modulation transfer function compensation.     

Exploration and implementation of commonality valuation method in commercial aircraft family design

Commercial aircraft family design can reduce development costs, shorten development cycles, and expand the market coverage of aircraft. Commercial aircraft family development has become one of the most important features of modern aircraft design. This paper explores the effects of commonality on different aircraft models in a commercial aircraft family. The existing product commonality indexes are summarized and their limitations in the application to aircraft design are discussed. Then a new component commonality index is proposed based on the component decomposition structure. A model for calculating the aircraft program value is established,which considers development costs, manufacturing costs, sale price, operation costs and residual costs. The effects of aircraft commonality on time and economic costs of both development and manufacturing, and on sale price, are analyzed and quantified. The commonality evaluation strategy is obtained, which features comprehensive consideration of the aircraft program value and time costs. The break-even analysis of aircraft is proceeded on the basis of costs and price data. By using a real option method, the strategy considers the uncertainty of the aircraft program and the flexibility of the manufacturer. This strategy proves to be rational and applicable to aircraft design based on the calculation of three examples and the analysis of parameter sensitivity.     

磁选态单束铯束管束光学参数的模拟计算和设计

为提高铯束管跃迁信号强度,在束光学参数设计中,用蒙特卡罗方法模拟计算铯原子通过率和未跃迁比例。通过改变模拟计算中的束光学参数,分析铯原子通过率、未跃迁比例和倍增器首极电流的变化趋势,得到了主要束光学参数的合理取值范围。在铯束管实际装配和试验中使用束光学参数设计结果,有效地增加了倍增器首极电流。     

电子接插件在线质量检测研究

近年来,电子信息产业的迅猛发展,作为配套支持的接插件的生产规模也不断扩大。由于电子设备厂商对接插件的质量要求越来越高,接插件的检测也得到了越来越多的重视。由于检测速度快、运行稳定可靠的特点,机器视觉检测技术被广泛用于对接插件外观质量的在线检测。本文应用机器视觉检测技术对电子接插件质量进行了在线检测研究,并以其典型的冲压质量缺陷的检测为例对机器视觉在线检测系统的构成、核心处理模块、应用中存在的问题以及解决问题的新方法进行阐述,最后针对系统的可靠性进行了分析。     

黄渤海晴空像元下雾霾识别的机器学习算法

为实时有效地识别和监测雾霾，选取2013年10月4日与10月5日的两景MODIS遥感图像，利用不同类型像元的光谱差异，基于神经网络模型，建立海上雾霾的遥感识别机器学习算法。经检验表明，在晴空条件下，机器学习算法反演精度相对较高，已超过98%，利用算法可以实现对中国近海上空雾霾的监测。