Smart vibratory peening: An approach towards digitalisation and integration of vibratory special process into smart factories

Abstract

Shot peening is a widely used technique in aerospace to enhance the fatigue life of a part. However, it leads to an increased surface roughness, which necessitates an additional Vibropolishing technique to improve the fatigue life and aerodynamic efficiency. Alternatively vibratory peening, a relatively new technology, can replace both Shot peening and Vibropolishing and achieve similar fatigue life and aerodynamic efficiency. This paper reports the development of a novel smart vibratory peening system, using which several experimental investigations are conducted to characterize the part and validate them using the conventional Almen strips. Data from the experimental investigations are used to train a machine learning model and predict the characteristics of the given part and hence eliminating the use of Almen strips. With the help of the feature redundancy removal technique, 79% of the redundant features were removed to retain 94% of the process information. The results show that out of various established machine learning models, the Kernel Ridge model demonstrated the best performance with an RMSE of 0.0443. With the introduction of the smart vibratory peening system, the overall process time has been significantly reduced by 76%.