Dong Shang Chang

Reliability bounds and critical time for the Birnbaum-Saunders distribution

The Birnbaum-Saunders distribution, under certain conditions, can be used to model the fatigue failure-time caused by the catastrophic crack size. Reliability bounds for the Birnbaum-Saunders distribution and point and interval estimates for the critical time of the failure (hazard)-rate are discussed. The confidence intervals are constructed under the assumption that both parameters are unknown. This method is not affected by censoring, as long as confidence intervals for the parameters can be established. Numerical examples illustrate the procedure. >

Reliability prediction using nondestructive accelerated-degradation data: case study on power supplies

This paper describes a conceptual framework for reliability evaluation from nondestructive accelerated degradation data (NADD). A numerical example of data sets from power supply units for electronic products is presented using this framework. The authors model NADD as a collection of stochastic processes for which the parameters depend on the stress levels. The relationship between these parameters and the associated stresses is explored using regression. The failure-time of power-supply units is modeled by the Birnbaum-Saunders distribution, for which the confidence bounds and tolerance limits can be easily obtained.

Percentile bounds and tolerance limits for the birnbaum-saunders distribution

In this paper, a confidence interval for the lOOpth percentile of the Birnbaum-Saunders distribution is constructed. Conservative two-sided tolerance limits are then obtained from the confidence limits. These results are useful for reliability evaluation when using the Birnbaum-Saunders model. A simple scheme for generating Birnbaum-Saunders random variates is derived. This is used for a simulation study on investigating the effectiveness of the proposed confidence interval in terms of its coverage probability.

Graphical analysis for Birnbaum-Saunders distribution

Abstract The Birnbaum-Saunders distribution has been shown to be the failure time distribution for fatigue failure in particular and for stochastic wear-out failure in general. In this paper, we present a simple graphical technique, analogous to probability plotting, to estimate the parameters and check for goodness-of-fit of failure times following the Birnbaum-Saunders distribution. Using known results from regression analysis, confidence intervals for the parameters can easily be established. A salient feature of our method is that it can be used for censored data where no analytical method is available for estimation. Finally a numerical example is given to illustrate the procedure and to compare our results with that of the maximum likelihood estimation.

Assessing quality performance based on the on-line sensor measurements using neural networks

Rapidly evolving sensor technologies, which employ advanced techniques, such as lasers, machine vision, and pattern recognition, have the potential to greatly improve quality control activities in the finished product inspection and process monitoring. In this paper, a neural network model was developed to probe the dependence between the quality of finished product and sensor measurements which were collected to monitor the failure (sudden fracture) of a tool in the manufacturing process. A real case in mass production is employed to illustrate the modeling procedure. Utilizing the trained neural network, the quality information of finished product can be further obtained from the online tooling sensor measurements. The result reveals that the tooling sensor measurements not only can be employed to detect the process condition (wear out or sudden fracture) but also can provide valuable information to monitor the quality performance of finished product simultaneously.

Bootstrap prediction intervals for the birnbaum-saunders distribution

The Birnbaum-Saunders distribution has been recognized as a versatile failure time model. However, it is not widely used in process control as some of its important characteristics have not been obtained. In this paper, we utilize the bootstrap method to construct a prediction interval for future observations from a Birnbaum-Saunders distribution. Monte Carlo simulations are carried out to evaluate the performance of the proposed procedure. The results reveal that the bootstrap intervals are satisfied with desired coverage probabilities and average lengths as sample size n is at least 30.

Random number generator for the Birnbaum-Saunders distribution

Abstract The Birnbaum-Saunders distribution has been shown to be the failure time distribution for fatigue failure in particular and for stochastic wear-out failure in general. In this paper, the random number generator for Birnbaum-Saunders distribution is addressed. In testing the proposed approaches several large samples were generated with various combinations of the paramenters. The probability plot is further applied to check the goodness-of-fit of each sample.

Reliability bounds for the stress-strength model

Abstract Stress-strength model has been applied in a wide variety of engineering areas for the purpose of reliability design. However, there is no suitable approach for the models reliability bound is found yet since the difficulty is arising from many parameters incorporating in the systems. In this paper, a conservative reliability bound is constructed by modeling the problem as a mathematical programming. A numerical example is presented for illustrative the methodology.

Reliability bounds and tolerance limits of two inverse Gaussian models

Abstract In this paper, we obtain confidence bounds for reliability function of the inverse Gaussian distribution where all the parameters are unknown. We consider two cases where the reliability function is parameterized with two different sets of parameters. In the first case, the parameters arise naturally as the drift and variance parameter of the underlying deterioration process. For the second case, the parameters are the mean and dispersion parameter. We describe the life testing methods where these two different forms of pdf arise. For both cases, we give estimators of these parameters and their respective confidence intervals. We demonstrate how to construct the proposed confidence bounds of the reliability function from the confidence intervals of the parameters. The corresponding β-content tolerance limits can then be derived from these reliability bounds. Finally, a numerical example for each case is provided to illustrate the procedure.

The formulation of performance indicators in evaluating the implementation of automation in the construction industry

The focus of the paper is to present the processes in formulating a system of performance indicators to evaluate the implementation of the National Construction Industry Automation Plan in the Republic of China. We approach the formulation as a ‘construction automation transformation system’ which consists of three major components: (I) the inputs, such as funding, manpower and supports, (2) the sub-systems, and (3)the outputs. There are three sub-systems as defined: the processing sub-systems, the receiving sub-system and the socio-economic sub-system. Corresponding to each sub-system, a set of output performance indicators is developed. Program output performance indicators are proposed to measure the effectiveness of the processing sub-system. The industry effectiveness performance indicators and the national welfare indicators are recommended to measure the effectiveness of the receiving sub-systems and the socio-economic sub-system respectively.