Assembly reliability modelling technology using function decomposing and lssvm

Assembly reliability is an important part of product quality, which greatly affects the reliability level of whole machine. Because of the limitations of FMEA and FTA, they are not suitable for assembly reliability of complex mechanical equipment. A new functional decomposition method for whole machine (FDWM) is proposed for assembly reliability modelling in this paper. Firstly, by decomposing the whole machine into function and connection function of main components, the success tree model for whole machine (STWM) can be established. Secondly, a hybrid method by combining modified grey relation and least squares support vector machine (MGR-LSSVM) is proposed to improve the accuracy of STWM model. Lastly, the proposed method is applied to the assembly reliability analysis for feed mixer. The results show that the FDWM method is feasible for assembly reliability modelling, and the MGR-LSSVM method has a higher accuracy, and is efficient for the reliability analysis.