Reliability estimation by fusing multiple-source information based on evidential variable and Wiener process

Abstract

Wiener process has been combined with evidential variable in reliability estimation and shows superiorities, especially under small sample conditions. However, the previously published related reliability estimation methods only can utilize the degradation data under normal use stress, but cannot utilize the expert knowledge or accelerated testing data. While, the accelerated life tests and accelerated degradation tests are always adopted to collect enough data in a short testing period in engineering practice. Furthermore, the expert knowledge is also an important information source to estimate the reliability of the products. Hence, in order to utilize the above multiple-source information simultaneously, the previously published evidential variable based degradation model is developed into an accelerated testing model to describe the degradation process and the corresponding acceleration relationship. Furthermore, a related reliability estimation method is proposed. A simulation study and an actual engineering case study are performed to verify the effectiveness of the proposed method. The proposed method shows superiorities on parameter estimation, reliability and life distribution predictions. Furthermore, in order to consider the individual difference in population evaluation, the proposed method can also predict the plausibility and belief that the products can meet the requirements, including the mean time to failure and reliability.