During product design phase, aiming at the problem of lacking reliability data, lower of product reliability, feature similarity-based new product design reliability analysis and prediction model were proposed. Putting the new product features as an evaluation objectives, an approach named Technique for Order Preference by Similarity to Ideal Solution(TOPSIS) was established firstly for selecting similar features products; Then, in order to realize the reliability analysis relational mapping

more »

... h the new product design, the failure structure of the similar features products was quantified and the product failure structure matrix (FSM) was established, respectively; Afterwards, the Group Decision Making Method (GDMM) was presented for determining the improvement factor of the similar features products failure causes, on that basis, the new product features failure structure was generated to predict the reliability of new designing products. Finally, feasibility and effectiveness of the model were verified through an example of new Smart Mobile Phone product design. 242 Copyright ⓒ 2014 SERSC reliability, but the drawback is the testing cost is too high and the analysis process is too complicated, and it would not obtain useful reliability information if the initial tests were wrong. Case-based reasoning [9] (CBR) is a solving strategy to new problems by analogizing and associating the empirical knowledge, but this method focuses more on instance representation, case retrieval and storage, etc., less on reasoning research, also the retrieval process of CBR is sometimes not as accurate as the human observation, like expert decision making. Product reliability simulation technology (PRST) [10] [11] to some extent reduces test cycle time and the resource consumption by using parameter characterization method, but PRST is insufficient to some degree mainly in (1) complex simulation modeling and needs experts with specialized knowledge in simulation process, which restricts the application of PRST. (2)The existing data -driven simulation modeling techniques cannot well express the large number of semantics and relationships between decision making work and simulation parameters. (3) Simulation parameters are commonly selected from engineering experience, which is not very accurate. Built-in reliability[12](BIR) and design for reliability [13] (DFR) both emphasize the need to consider and control the factors and variables affecting product reliability in early stages of product design, and take measures to optimize the reliability design. This paper based on the research results analyzing of design reliability home and abroad, combined with the theories and methods of BIR/DFR and product feature similarity, to analyze and predict the design reliability in product design phase.