Comparative Analysis of Life Expectancy Prediction for IGBT with Neural Network and Noise Data

This paper presents a study focused on utilizing neural networks (NN) to forecast the lifespan of electronic components. These components are susceptible to deterioration over time caused by aging, environmental factors, and other issues. Accurately predicting the remaining useful life of these components is crucial for enhancing reliability and minimizing downtime. Earlier, we introduce a neural network-based approach to estimate the life expectancy of electronic components with correct data. This study presents a logical extension, where the same NN is try against “noisy” data. Moreover, by simulating real-world variations and uncertainties, the model becomes more adaptable and capable of handling unforeseen circumstances in electronic component lifespan prediction. This noise-aided training approach ensures that the neural network is not overly reliant on ideal or noise-free data, making it more reliable and resilient in practical applications.