International Journal of Reliability, Quality and Safety Engineering

The testing of configurations with constraints still faces a great challenge. Although artificial intelligence (AI)-based algorithms perform better than greedy algorithms on $t$-way testing because of the good searching ability of optimal solutions, only a few AI-based algorithms can support constraints currently. Moreover, the AI-based algorithms can only ignore the conflicting candidate test cases subject to constraints, even though they are optimal. In this paper, we demonstrate two novel particle swarm optimization (PSO)-based constraint test cases generation (PCTG) methods. In the two methods, the strategies of avoiding the selection of conflicting test cases and replacing conflicting test cases are applied to handle constraints, respectively. They guide the process of searching for optimal solutions from different perspectives, according to different handling of constraints. We evaluate the availability of these two methods with some excellent existing strategies in terms of performance. The evaluation results indicate that our proposed methods, in most cases, outperform other strategies as far as the generated constraints covering array sizes.

This paper is concerned with a parallel system that have a random number of units. The distribution of the number of units is assumed to follow a power series class of distributions which contains well-known distributions such as modified or truncated Poisson, geometric, and logarithmic distributions. Optimal number of units and replacement time for the system which minimizes the mean cost rate are computed. The results extend the results in the literature from Poisson distribution to a wider class of distributions.

The main failure mode of cylindrical shock absorbers (CSAs) used by taxis in China’s Zhejiang Province is oil leakage because of a large horizontal force loaded on the cylinder wall by a questionnaire survey. The previous CSA durability bench test methods, which do not involve the interaction between horizontal load and vertical load and do not consider the user road spectrum, cannot simulate the real fatigue phenomena of it. In order to preferably reproduce the failure mode and accurately to predict the failure time of the CSA used in the area, a compilation method of multi-axial loading spectrum is proposed based on the road spectrum. The vertical loading spectrum is compiled by using the way of sine on sine, in which high frequency (HF) signal superimposed on low frequency (LF) signal. The LF is determined by the half-power bandwidth theory, which can reflect the power spectral density distribution of the random road spectrum. The HF is determined by the maximum line velocity principle to effectively reduce test time. The loading cycles and amplitudes of them are determined by using the rain-flow counting method. In addition, the horizontal loading force is taken into consideration in the multi-axial loading spectrum. To prove the validation and efficiency of the method described in this paper, the compiled multi-axial loading spectrum was used to conduct the durability bench test. The results show that the mean failure mileage of specimens is close to the mean mileage time of field samples. The failure modes of the specimens are the same as those of field samples. Thus the proposed method can be used as a reference for durability bench test of CSA. The test data can also be used to accurately study the reliability of CSA.

For systems with multiple redundancies, reliability evaluation in the redundancy allocation problem (RAP) constitutes a computational complexity. It has been demonstrated that neural network training provides an efficient approach to estimate the complex system reliability function. When executing the neural network algorithm, there are many parameters that need to be determined for improving the training performance. Therefore, robust experimental design method can be used to determine the neural network parameters. The traditional robust design methods are intended for a single response variable. However, the application of neural network method includes more than one measurement, such as estimation accuracy and time efficiency. In this paper, utility function is first estimated by neural network training, in which the algorithm parameters are determined by weighted principal component (WPCA)-based multi-response optimization which simultaneously optimizes more than one training performance measurements. Moreover, it is always desirable to simultaneously optimize several objectives in designing a system, such as reliability, cost, etc. Therefore, continuous WPCA-based multi-response design is then applied to obtain the best design of redundancies in RAP, which simultaneously optimize multiple objectives by taking into account the correlations between them.

The purpose of this paper is to propose an approach for testing multivariate uniformity. The proposed test procedure works for both complete sample case and incomplete sample case. What multivariate uniformity test does is to check whether the population distribution, from which a random sample is drawn is a multivariate uniform distribution. Particularly, it is desired to check whether the population distribution is a uniform distribution on a $p$-dimensional unit cube. There are some existing research papers in the literature. The proposed method works when only some arbitrarily selected order statistics are available for the test.