A connecting rod assembly deformation cognition method based on quality characteristics probability network

As the core component of marine diesel engines, the connecting rod quality directly affects the reliability and life of diesel engines. However, the deformation problem of the connecting rod assembly process, which caused the low pass rate of one-time assembly, has not been well solved. The deformation of connecting rod assembly is difficult to analyze by mechanism because it is influenced by many quality characteristics with complex and cross-process correlations. To address this issue, this paper proposes a connecting rod assembly deformation cognition method based on the quality characteristics probability network. Firstly, a quality characteristics probabilistic network model for the connecting rod assembly process is established. Secondly, based on the quality characteristics probability network model and the complexity of cross-process correlation, the K-shell entropy weight model is used to identify the key quality characteristics and construct their correlation model. Then, to deal with the uncertainty of the influence degree between quality characteristics, a Bayesian network is used to establish an influence analysis model of key quality characteristics and analyze their influence relationship from qualitative and quantitative perspectives. Finally, the simulation and actual assembly data of marine diesel engine connecting rod assembly process are used to train and verify the cognition method. The results show that the method can effectively analyze the relationship between the connecting rod assembly process quality characteristics and predict the connecting rod deformation degree. The proposed method can comprehensively consider the influence of multiple processes and quality characteristics, improve the accuracy of analysis and prediction of connecting rod deformation, and can be used to guide the dynamic adjustment of the connecting rod assembly process to increase the pass rate of one-time. Moreover, the proposed method can also provide a reference for other multi-process and multi-factor problem analyses.