Prevention of the fracture problem occurring in automotive alternator heatsink blocks using artificial intelligence

Abstract

In this study, prevention of fracture in vibration fatigue testing of automotive alternator heatsink blocks was investigated using an artificial neural network. Automotive components such as alternator heatsink blocks are subjected to high cyclic vibration fatigue loads throughout their lifespan, which can lead to the formation and propagation of fatigue cracks and ultimately component failure. The basic parameters affecting the resonant frequency of the heatsink blocks, including geometry and loading conditions, are determined. Data-driven decision making provides advanced predictive insights to analyze data for prediction and decisions using artificial intelligence approaches. An efficient artificial neural network model was defined to predict the resonance frequency in the vibration fatigue test. While the artificial neural network was trained to establish a functional relationship between the parameters and the resonance frequency, regression analysis was used to develop a predictive model to detect the resonance frequency of the heatsinks. The proposed approach aims to provide a comprehensive framework for preventing fracture problems in vibration fatigue tests of automotive alternator heatsinks and ultimately contribute to the reliable design and performance of these critical components. While the artificial neural network approach achieved high classification accuracy in predicting the new natural frequency, the regression model was also able to make accurate predictions. The results of this study showed that the time spent on design and simulation can be significantly reduced in preventing breakage problems that may occur before dynamic tests such as vibration tests of alternator components.