Data Center Cooling Control Using Causal Signal Injection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face inefficiencies in energy management due to the short lifespan of machine learning models used for optimizing cooling systems, which require extensive data and time for retraining, leading to 'catastrophic forgetting' and suboptimal performance.
Innovation Solution
Implementing active data center management through randomized signal injections and causal analytics to monitor and optimize cooling system parameters, allowing for faster and more robust learning by inferring causal effects on utility metrics like power utilization efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If advanced machine learning techniques are used to optimize cooling systems, then energy reduction is achieved, but model accuracy degrades over time due to catastrophic forgetting
Solution Approach 1:
The patent implements dynamic model updating by continuously adapting the machine learning model to new data as equipment is refreshed. Instead of relying on a static model that degrades over time, the system dynamically adjusts to changing conditions and equipment configurations, preventing catastrophic forgetting while maintaining energy optimization performance.
Solution Approach 2:
The system incorporates feedback mechanisms that monitor model performance and trigger retraining when accuracy degradation is detected. This feedback loop ensures the model maintains reliability by identifying when catastrophic forgetting occurs and initiating corrective retraining actions to restore optimal performance.
2Reliability
If machine learning models are retrained frequently to maintain accuracy, then model reliability is improved, but time and computational resources are consumed
Solution Approach 1:
The patent prepares models in advance for potential equipment refreshes by implementing proactive retraining schedules and maintaining multiple model versions. This preliminary action ensures that when equipment changes occur, the system can quickly switch to pre-prepared models or rapidly retrain without significant downtime, reducing the time loss associated with model updates.
3Measurement precision
If extensive operational data is collected for model training, then learning accuracy is improved, but data storage and processing requirements increase
Solution Approach 1:
The patent extracts only the most relevant features and data points necessary for training accurate models, rather than storing and processing all available operational data. By identifying and extracting critical variables that drive cooling system performance, the system achieves high learning accuracy with a reduced data footprint, minimizing storage and processing requirements.
Data Source
AI summary
Methods for active data center management by injecting randomized controlled signals in the operational controls of the cooling infrastructure of the data center and receiving response signals corresponding with the injected signals. The injected signals are used to adjust the operational controls of the cooling infrastructure, and the response signals relate to the operational conditions in the data center. Based upon the response signals along with independent and external variables, the randomized signals are continually injected into the cooling infrastructure and fine-tuned based upon the response signals. Optimum or improved parameters for controlling the cooling infrastructure of the data center are determined based upon the signal injections and corresponding responses.