Reliability Assurance System Using Large Scale Multi-Media Databases and its Application

结果可归纳为以下四类：1。防止可靠性和安全性问题的信息共享系统为了防止可靠性和安全性问题的出现，数据的收集和分析、通过数据库对信息进行整合以及在实际现场的应用起着重要的作用。从管理方面、人员方面和硬件方面进行了研究，并提出了可靠性和安全性信息共享系统。提出了一种二元寿命试验的优化设计方法，使d -最优准则的估计精度达到最高。并对其序列设计进行了研究。很明显，大规模集成电路电迁移的分布形式服从对数正态分布，而不是威布尔分布。基于多媒体信息的状态监测维修利用传感器进行状态监测维修，对保证可靠性和安全性有着重要的作用。但是传感器有两种失效模式：误报和未报。为了减少这些故障，普遍采用k-out- n报警决策系统。本研究探讨了所有传感器报警阈值相等的最优性、自移动传感器的有效性以及带有灰色地带的阈值。基于大规模数据库的寿命数据分析目前，现场可靠性数据数据库已遍布全球，且规模越来越大。这些数据主要来自于各部门在产品保修期的维修记录。相反，销售金额信息来自销售部门的数据库。为了估计寿命参数，有必要将这两种不同的数据库结合起来。为了估计寿命参数，有必要将这两种不同的数据库结合起来。而且，有时每个月的销售金额都没有，只知道产品的总销售金额。本研究针对这些问题，开发了基于大规模数据库的寿命数据方法。少

Results are summarized into the following four categories :1.System for Sharing Information to Prevent Reliability and Safety ProblemsTo prevent problems in reliability and safety, collection and analysis of data, integration of information through databases, and their application on the actual spot play important roles. Researches on management aspect, human aspect and hardware aspect are implemented and systems for sharing information for reliability and safety are proposed.2.Design of Lifetime Evaluation and Its Distributional FormAn optimal design for binary lifetime experiment is developed which attains the highest precision of the estimation from D-optimal criterion. Also its sequential design is investigated. It becomes clear that the distributional form of an electro-migration of LSI follows a log-normal distribution, not Weibull distribution.3.Condition Monitoring Maintenance Aiming Utilization of Multi-media InformationCondition monitoring maintenance using sensors plays impo … More rtant roles for reliability and safety assurance. But a sensor has two failure modes : false alarm and failure to alarm. To decrease these failures, a k-out-of-n alarm decision system prevails. This research investigates the optimality of equal thresholds of all sensors for alarming, validity of self-moving sensors, and the thresholds with gray zone areas.4.Lifetime Data Analysis Based on Large Scale DatabasesNowadays databases of field reliability data have been spreading all over the world and getting larger and larger. Mainly these data come from repair records during product warranty period in each division. To the contrary, sales amount information comes from the databases of sales department. It is necessary to combine these two different databases for estimating lifetime parameters. It is necessary to combine these two different databases fro estimating lifetime parameters. Also, sometimes, sales amount of each month is missing and just total sales amount of the products are known. This research tackles with those problems and develops the methods of lifetime data based on large scale databases. Less