Abstract:Nowadays, surveys show that the cooling system accounts for 30% of the energy consumption of data centers. The challenge of designing an effective cooling control system to minimize data center energy consumption cost as much as feasible while maintaining normal operation is a pressing one to address. The traditional cooling control method produces excessive cooling energy consumption, which causes a large amount of energy waste and economic expenses; Moreover, the existing Model Predictive Control (MPC) method based on the physical model has high modeling difficulty and model migration cost, and can only provide short-term control strategy. In this paper, we propose an MPC method based on a dual-stage and multi-step forecasting model (DMF-MPC), which can forecast the cooling system energy consumption in data centers accurately and provide a cost-saving control strategy of the cooling system for the next 24 hours on a combination of the peak-valley time-of-use (TOU) electricity price. We firstly propose a dual-stage and multi-step forecasting model (DMFM), a pure data-driven deep learning model, which adopts the historical server features of data centers, future weather factors, and cooling control to forecast the cooling system energy consumption for the multiple future periods, also a multi-step iterative training method is proposed in the training process to reduce the multi-step forecasting errors. In the control part of DMF-MPC, we introduce an optimal control model for the data center cooling system based on DMFM. The cost-saving control strategy for the next 24 hours under the peak-valley TOU electricity price standard can be obtained by solving the model through optimization algorithms. Finally, the effectiveness of the method is verified by validating DMF-MPC in a real campus high-performance computing (HPC) data center located in China. Our method achieves a 15%-25% cost reduction in the experiment.

Dynamic Energy Storage Control for Reducing Electricity Cost in Data Centers

Optimal Power Cost Management Using Stored Energy in Data Centers

Online Learning-Based Server Provisioning for Electricity Cost Reduction in Data Center

Battery-Assisted Online Operation of Distributed Data Centers With Uncertain Workload and Electricity Prices

Distributed Energy Management for Multiple Data Centers with Renewable Resources and Energy Storages

Online Electricity Purchase for Data Center with Dynamic Virtual Battery from Flexibility Aggregation

Long Term Operational Optimization of Data Center Network under Uncertainty

Proactive Demand Response for Data Centers: A Win-Win Solution

Dynamic Allocation of Power Delivery Paths in Consolidated Data Centers Based on Adaptive UPS Switching.

Demand Charge Control for Energy-intensive Enterprises Based on Deep Reinforcement Learning

RE-UPS: an Adaptive Distributed Energy Storage System for Dynamically Managing Solar Energy in Green Datacenters

An optimal solutions-guided deep reinforcement learning approach for online energy storage control

DMF-MPC: A Dual-Stage and Multi-Step Forecasting Model Based Cooling System Control Method for Data Center Energy Cost Minimization.

Energy Cost Minimization with Risk Rate Constraint for Internet Data Center in Deregulated Electricity Markets

Minimizing Electricity Cost: Optimization of Distributed Internet Data Centers in a Multi-Electricity-Market Environment

An Energy Saving Control Strategy Based on Multi-Agent Q-Learning Algorithm for Data Center

Energy Cost Minimization with Job Security Guarantee in Internet Data Center

Data-based Energy Storage System Configuration and Operation Considering Degradation Cost

GreenFlowing: A Green Way of Reducing Electricity Cost for Cloud Data Center Using Heterogeneous ESDs

Lowering Down the Cost for Green Cloud Data Centers by Using ESDs and Energy Trading

Energy Cost Minimization for Distributed Internet Data Centers in Smart Microgrids Considering Power Outages