Academic literature on the topic '可靠度'

偶氮四唑锌（ZnATZ）不易吸湿，假密度较小，具有良好的火焰感度，撞击感度较高，能用作点火药和击发药。作为击发药组分，ZnATZ具有合适的撞击感度，而且能避免含铅起爆毒性大致使分解气体有黑烟或者残渣量较大问题。此外，ZnATZ还可以用作飞片式雷管的施主药。偶氮四唑锌（ZnATZ）为不含铅材料，同时具有所需的热稳定性，并且可由极小的热能瞬间可靠点火。ZnATZ药剂作为点火药，可广泛应用于电火工品制造中，具有良好的应用前景。

随着人民生活质量的提高,对配网的电力供应提出了新要求。 配网供电系统是面对客户用电的最后一道关卡,电力供应,物资先行,决定客户平均停电时间、客户平均停电次数、电压合格率等安全生产指标的关键在于配网运维,配网运维中的核心是人力资源和物资资源,如何管控物资资源的可靠质量、供应速度,可以直接反应到安全生产指标的建设;物资资源的管理直接也关系到配网运维高质量、高效率发展,如何在保障配网运行安全、可靠、稳定的同时,确保物资可靠高效管理,是一个迫切问题。 利用大数据互联网平台,搭建配网物资就地供应体系,推进配网物资管理穿透力,从配网自动化、电网发展新形势对配网物资供应的可靠高效提出新要求,从物资大数据、物资管理新体系方向中寻求突破。

碩士
中正理工學院
兵器工程研究所
86
To estimate the reliability of a repairable system, the present statistics methods such like Probability distributions, Non-Parametric methods, Graphic solutions and Mixture of populations to figure out the reliability value are not suitable. In order to determine whether the prototype products or weapon system used in a demonstration test meet the specified reliability requirement, we here establish a suitable life data analysis procedure which based on Non-Homogeneous Poisson Process with Weibull Intensity Function and can estimate the reliability reasonably. During the research and development stages, we can also get some valuable information about parts life then to suggest a preventive maintenance. Furthermore, it will conduct us to make an optimum acceptance specification under a given producer''s and customer''s risks.

博士
國立中央大學
統計研究所
88
We develop models of non-Gaussion Kalman filter to apply in reliability. There are four themes in this paper: software reliability, accelerated life testing, fatigue-crack-growth model, and burn-in procedure. The reliability of computer software is a crucial measure about the quality of computer system. We have presented a Kalman filter to predict software failure time, which is a generalized JM model. Two important indexes for quality of software have been reached: the reliability function and the probability that the program is perfect. In practice, the system equation can be modified to satisfy the different situation. It can be applied to the imperfect debugging situation as well as in evolving programs. By examining a set of data on evolving program failures, the effect of evolving program model is amply proved. For estimating the reliability of long-life products, many techniques are used, such as the accelerated life testing (ALT), the degradation performance measuring, and the accelerated degradation testing (ADT). We propose a Kalman filter which is applied to the ALT and the degradation performance measuring. It assumes a Gamma type distribution for the observations, conditionally on unobserved state variables. A simple simulation study and an example for ALT are shown. The simulation result shows that when the observations are i.i.d., the inference about unobserved state parameters based on this model will be near to the true model; and the example shows that the predicted value of our model may better than m.l.e. in some cases. We can use this model to predict the future degradation performance. If failure is defined in terms of a specified level of degradation, a degradation model defines a particular time-to-failure distribution. The distribution of our model will be estimated by a fatigue-crack-growth data. In the fatigue-crack-growth example, there are some units censored. The estimate of time-to-failure distribution of our model based on the censored data is similar to the nonparametric estimate in full data, and the computing time is sooner than bootstrap method. Burn-in is a manufacturing operation that is intended to eliminate early failure. It becomes particularly meaningful when the distributions of the components'' lifetime have a decreasing failure rate in a neighborhood of the origin. In our problem, there are several components being tested at once, and a loss function, as the cost of burn-in which given by Clarotti and Spizzichino (1990) and Spizzichino (1991), will be introduced. We use a Bayesian steady forecasting model to describe the variation and uncertainty of the failure rate in testing period, and assume the burn-in is based on discrete time. We obtain an optimal strategy and simplify computing by an approximate method.

碩士
國立清華大學
應用數學研究所
76
S 為－r-out-of-k系統，考慮應力－強度模式（Stress-strength model ），則S 作 用若且唯若在S 的k 個分量中，至少有r 個作用，即至少有r 個分量，它們所受的應 力比本身的強度小。本文以無母數理論研究應力與強度皆為多變量時，系統S 之可靠 度對應力與強度之分配無關的信賴區間（disstribution-free confidence interval ）。當 r＝ k＝１，過去曾有多位作者討論，１９８７年，Halperin，Gilbert 與La chin也做了這方面的研究，但其方法僅適用於r ＝k ＝１的情形。本文最主要的目的 在修改並推廣Halperin等的方法，使之適用於一般具有k －維之應力－強度模式的問 題。 本文分成四部分說明研究結果： （１）過去一些相關的結果：敘述過去可應用於本文之問題的主要結論。 （２）推廣H.G.L.方法至k ≧２：說明如何修改與推廣Halperin等的方法與結果。 （３）模擬分析：以蒙地卡羅法的模擬分析方式，對過去部份之結論與本文之結果作 一比較。 （４）結論：把以上所討論的做一總括性的結論。