Chip Temperature Prediction With Pre-Operation Frequency Control

The system performs preliminary temperature prediction using a trained artificial neural network before executing applications that may cause temperature rise. Based on the predicted temperature increase, the system proactively adjusts clock frequencies of the processor and memory in advance, rather than reactively reducing frequency only when temperature thresholds are exceeded. This preliminary action prevents sudden power off situations by maintaining temperatures within safe operating ranges throughout application execution.

The system collects temperature data, application execution data, and other operational parameters during normal device operation to automatically train the artificial neural network. The training process utilizes existing operational data without requiring external training datasets or additional hardware resources. The system continuously improves its temperature prediction accuracy by learning from its own operational history, thereby achieving high prediction precision while minimizing the increase in device complexity.