Research on Computational Method of Fault Probability for New Product Development Based on Intelligence and Integration

Abstract

Computing fault probability is an effective method of predicting the possible weakness of new product. Aiming at computing fault probability of each component of new product, and ensuring simplification of computational process and reasonableness of result, in the paper, according to design proposal and layering of new product, fault tree meta-model (FTMM) is established based on node-knowledge-representation method (NKRM) to express logical relation between each failure in different level of product and collaboration between each part. Secondly, RST and Bayes theory are applied to mining the decision rule of FTMM which is that fault of part in lower level causes fault of part in upper level, and joint probability of part is computed directly which avoids difficulty of computing because of unknown prior probability, then fault probability of part is computed. On this basis, weighted mean algorithm based on Bayes-and-RST (BRMA) is proposed to revise fault probability further, thus the result is more exact and reasonable, which can help product development personnel predict possible weakness of product in time and carrying out reliability design further in the design stage of new product development. Finally, an instance is analyzed