A Combined Algorithm For Solving Reliability-based Robust Design Optimization Problems

Ameneh Forouzandeh Shahraki, Rassoul Noorossana
2013 Journal of Mathematics and Computer Science  
In the most of the design optimization problems, we encounter uncertainties in design variables and problem parameters. In these problems, robustness and reliability of design are so important. Both robust design and reliability-based design approaches take into consideration these aspects. However, the individual application of them doesn't ensure the stability of product during its life cycle. In this paper, we combine both robust design and reliability-based design approaches into one model
more » ... nd propose a genetic and reliability analysis combined algorithm to solve this kind of problem. Moreover, to increase the efficiency of the genetic algorithm, we use the design of experiment (DOE) to find the optimal levels of the parameters of this algorithm. The application of the proposed methodology is demonstrated using a numerical example. Introduction. Deterministic optimization techniques have been successfully applied to a large number of product design problems. However, deterministic optimal designs are pushed to the limits of design constraint boundaries and leaving no room for uncertainties. The probability of failure of such designs is high because of impact of uncertainties existed in modeling, manufacturing phase and operation condition. Therefore, such designs obtained without considering uncertainties can be unreliable and might lead to failure of the product being designed [1]. Different approaches have been developed to consider uncertainties and optimize the designs under the uncertainties. Among them, robust design optimization (RDO) and reliability based design optimization (RBDO) are two major approaches. RDO is a method for improving the quality of product through minimizing the effect of variation without eliminating the sources of variation [2] . The robustness of performance is important in this approach. On the other hand, RBDO approach emphasizes on maintaining design feasibility at different reliability levels and in order to consider the uncertainties, the constraints are modeled as probabilistic [3, 4] . Robust design, originally proposed by Taguchi (1993) to improve the quality of products by minimizing the effect of variation without eliminating the source of variation. The different approaches of robust design are divided into three main categories: Taguchi's experimental design technique [2], RSM-based RD [5] general optimization techniques and [6] . The main goal of all of them is optimizing the mean and minimizing the variance of performance function. In these problems, the accuracy and efficiency of
doi:10.22436/jmcs.07.01.06 fatcat:pobtu5vht5hxxadanv4yyn2la4