Intelligent reverse analysis method for elastic-plastic properties of CuPb22Sn2 bearing shell coating via indentation experiment

Abstract

With increasing demand for reliability in internal combustion engines (ICE) and other power machinery, accurately analyzing key components like journal bearings is critical. Accurate analysis requires precise mechanical properties of bearing alloy layer material, such as CuPb22Sn2. However, its thin structure complicates the determination of mechanical properties through traditional methods. In this work, an intelligent reverse analysis method suitable for displacement control P-h curves is proposed to determine the mechanical properties of the CuPb22Sn2 alloy layer in bearing shells via in-situ indentation experiments. A trained artificial neural network (ANN) is used as a surrogate model, replacing finite element model (FEM) in optimization reverse analysis, reducing computation time from several hours to a few minutes, greatly improving efficiency. The ANN incorporates load samples from P-h curves and reference depth at unloading stage as output. This approach enables the method to handle displacement control P-h curves while maintaining complete loading and unloading information. The genetic algorithm (GA) is introduced to address sensitivity to initial values. Validation shows that the error of the proposed method is below 3.99%, meeting engineering requirements. In-situ experiments on CuPb22Sn2 alloy layers in bearing shells provide valuable mechanical properties data for journal bearing analysis in ICE.