Multidisciplinary Design Optimization And Uncertainty Of Research

With the development of society and technology, the project systems are becoming more and more complex in the domain of industrial design. Therefore, the limitation of the traditional methods of design and optimization appeared. In order to solve this problem, multidisciplinary design optimization (MDO) was proposed. MDO is a meth-odology of designing complex system through deeply exploring and utilizing the inter-active cooperation mechanism in the project system, and its main idea is organizing and managing the procedure of the fairly complex system by distributed computer network in the design procedure. The aim of MDO is to obtain the optimal solution of the system by using the cooperation effect produced by the interactions between different disci-plines. In this paper several common MDO methods were studied and verified by the same example. After the comparison, the advantages and disadvantages of these meth-ods were given.For any design process, the reliability of the product should be considered. Re-search on reliability has obtained great achievement in the traditional design and opti-mization methods and plenty of experience was accumulated simultaneously. However, in the field of MDO there is great deal of difficulty with research on reliability of mul-tidisciplinary system due to MDO has the characteristics of multi-object, multidiscipline, multi-coupling and large calculation. In this paper, reliability index was introduced into the multidisciplinary system modeling and then the model of reliability based MDO (RBMDO) was established based on the traditional methods of design optimization. In-dividual discipline feasible (IDF) for RBMDO method was introduced and then the ef-ficiency of this method was verified by two examples.In project analysis, modeling and simulation based design, there may be insuffi-cient or imperfect knowledge on design which leads to difference between the model-predicted performance and the actual system performance and there also may be fuzziness in problem description and decision-making, all of which consist of uncertain-ties in project design. In MDO a critical issue is that the uncertainties of one discipline may be propagated to another discipline through the linking variables and the final out- put from the integrated multidisciplinary system has an accumulation of the uncertain-ties from the individual disciplines which may influence the reliability of the design. In order to solve this problem a new method for uncertainty analysis in multidisciplinary system: collaborative uncertainty analysis (CUA) was proposed. This method estimates the mean and variance of the uncertainty in design parameter and model and the uncer-tainty analysis was performed concurrently between disciplines. Compared with the Monte Carlo Simulation method, CUA can reduce the amount of the calculation sig-nificantly in the system level, so the efficiency of MDO was improved. In the end a mathematical example and an electronic packaging problem are used to verify the effec-tiveness of CUA. The result shows that CUA is not only feasible, but also has high ac-curacy. So CUA is an effective method for uncertainty analysis in multidisciplinary sys-tem.