Processor Power Estimation with Dynamic Service-Class Models

The patent segments power consumption models into multiple service classes (first service class and second service class), each with its own dedicated power consumption model. The processor determines which service class an event belongs to and applies the corresponding model, thereby improving measurement precision without requiring a single overly complex universal model.

The patent implements dynamic selection of power consumption models based on service class determination. The processor dynamically switches between different power consumption models (first power consumption model for first service class, second power consumption model for second service class) according to the detected service type, allowing the system to adapt to changing service loads and maintain high estimation accuracy.

The patent performs preliminary classification of events into service classes before applying power consumption models. The processor determines the service class of each event in advance, which simplifies the subsequent model selection process and reduces the complexity of managing multiple models by establishing a clear classification hierarchy beforehand.

The patent applies the principle of local quality by assigning specific power consumption models to specific service classes rather than using a single uniform model. Each service class (first service class, second service class) has its own optimized model, allowing the system to maintain high precision for each local segment while managing complexity through localized model specialization.