A hybrid approach of process reasoning and artificial intelligence-based intelligent decision system framework for fatigue life of belt grinding

Belt grinding is widely used as the final step in the fabrication of fatigue-resistant surfaces of nickel-based superalloy components, and fatigue life after grinding is one of the most concerning issues. However, the response mechanism of fatigue life under different grinding parameter excitation conditions is not well understood for a long time. In this study, a system framework of fatigue life prediction for nickel-based superalloy abrasive belt based on process reasoning and artificial intelligence algorithm is proposed. Based on the process reasoning method, the mathematical relationship between grinding parameters and fatigue life is established. The equation is solved by RNN and LSMT algorithms embedded in the system, and the excitation response model of process parameters to fatigue life is obtained. The results show that the prediction accuracy of the system is high. The mean squared error (MSE) of the LSTM algorithm is below 0.0441, and the R -squared can be above 0.9956. In addition, experimental verification has been carried out, the observation of the specimen section shows that the process parameters have an effect on the initiation position, distribution, and crack length of the fatigue crack source, which are related to the stress concentration and residual stress distribution at the depth of the grinding scratches. Furthermore, using Spring Boot framework, an intelligent decision-making system based on this system framework is developed by using java and python.