Prediction of thermal profiles using artificial neural network

This paper present an intelligent computational system for the thermal management of microprocessor testing. The system is based on feed-forward neural network and used to predict the temperature profiles of the test die during testing process. By using this method, it can control the testing temperature proactively and prevent die overkill. We present initial phase of experiment results using our approach and it is shown that this method is suitable for thermal profiles prediction. We concentrate on the experimental verification to determine the practicability of neural network in this application and the obtained results strongly justify our approach.