International Journal of Reliability, Quality and Safety Engineering

By probabilistic optimization of partial safety factors for a class of structures a more uniform reliability of practical design of individual cases within the class is aimed at. The paper deals with an ultimate limit state design of a class of low rise industrial buildings subject to climatic loads and permanent loads. The optimization is carried out on selected representative structures, the reliability of which is calculated by the level II reliability method FORM. Probabilistic models of cross section resistance and of climatic loads pertinent to a continental climate are based on measurements. Peculiarities of the optimization procedure, particularly optimization stages in conjunction with the choice and possible differentiation of partial factors are discussed.

In this paper, we consider the optimal design of the binary weighted k-out-of-n system. The binary weighted k-out-of-n: G system works if and only if the total utility of all working components is at least k. In the design process, we need to evaluate system reliability repetitively. The universal generating function (UGF) approach is used for this purpose when the system size is small or moderate. When the size of the system is large, the recursive approach is used, which is more efficient. Two optimal models are formulated. One is to minimize the expected total cost while guaranteeing the system reliability higher than a pre-specified value; the other is to maximize the system reliability with the constraints on total system cost. Genetic algorithms (GA) and Tabu Search (TS) methods are both used to solve the proposed optimization models. Since the key to a good TS algorithm is usually quite problem-specific policies and memory structures, there is no existing general TS tool available. Therefore more details of the TS approach used in this paper are discussed than the GA approach. The results obtained with these two methods are compared. The results illustrate that both methods are powerful tools for solving these kinds of problems. However TS is more efficient than GA in computation. The materials in this paper have been published in 19.

Although process design optimization issues have received considerable attention from researchers for more than several decades, and a number of methodologies for modeling and optimizing the process have been developed, there is still ample room for improvement. Most research work has rarely considered the use of raw data from a manufacturing process database into the process design. However, the use of cumulative raw data can be a vital component in optimizing processes. To address this, we propose a new process design procedure called robust-Bayesian data mining (RBDM). First, we show how data mining techniques and a correlation-based feature selection (CBFS) method can be applied effectively to the selection of significant factors. Second, we then show how RBDM can be incorporated into robust design. Third, we present how the proposed RBDM estimates process parameters by considering the concept of robustness of the estimated parameters while incorporating the concept of noise factors. Finally, we present numerical examples to illustrate the efficiency of the proposed RBDM as a design tool for optimizing manufacturing processes.

A framework for human error modes and effects analysis is presented. The methods of hierarchical task analysis and event tree analysis are introduced as the basic analytical tools, and a novel generic framework for human errors classification is presented as the guidance to identify human errors. Firstly, the process of human error modes and effects analysis is discussed, and the hierarchical task analysis method is introduced briefly. Secondly, the framework for human errors classification is discussed in detail. Thirdly, the multi-states event tree is designed to model the accident scenario. Finally, two scenarios are selected as the examples to illustrate the proposed process of human error modes and effects analysis.

A model of software complexity and reliability is developed that uses an evolutionary process to transition from one software system to the next while complexity metrics are used to predict the reliability for each system. Systems are tested until the software passes defect presence criteria and is released. Testing criteria are based on defect count, defect density, and testing efficiency predictions exceeding specified thresholds. In addition, another type of testing efficiency — a directed graph representing the complexity of the software and defects embedded in the code — is used to evaluate the efficiency of defect detection in NASA satellite system software. Complexity metrics were found to be good predictors of defects and testing efficiency in this evolutionary process.