Micro-Grid Energy Storage Control for Price-Adaptive Cost Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current micro-grid systems face challenges in intelligently regulating energy storage to minimize electricity costs due to market price fluctuations and data accumulation requirements for machine learning algorithms, leading to increased user costs.
Innovation Solution
The proposed solution involves an energy storage control method that uses a dynamic programming model for short-term operations and a deep reinforcement learning algorithm for long-term operations, allowing for intelligent regulation throughout the micro-grid system's life cycle by dividing time scales and utilizing historical data effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a deep reinforcement learning algorithm is used for energy storage control, then the adaptability to market price fluctuations is improved, but the requirement for data accumulation increases, leading to higher initial costs and longer implementation time
Solution Approach 1:
The patent pre-trains the deep reinforcement learning model using historical data before actual deployment. This preliminary action allows the model to have initial predictive capabilities without requiring real-time data accumulation, thus reducing the implementation time and allowing the system to start generating value immediately while maintaining adaptability.
Solution Approach 2:
The patent segments the energy storage control into two phases: a pre-training phase using historical data to build initial model capabilities, and an online optimization phase where the model continuously learns from real-time data. This segmentation allows the system to achieve adaptability without requiring extensive real-time data accumulation.
2Device complexity
If traditional control methods are used for energy storage, then the system complexity is reduced, but the ability to minimize electricity costs is insufficient
Solution Approach 1:
The patent introduces a pre-trained deep reinforcement learning model as an intermediary between the simple control system and the complex market environment. This intermediary provides intelligent decision-making capabilities without requiring the entire control system to be complex, thus maintaining ease of implementation while achieving superior cost reduction performance.
Solution Approach 2:
The patent replaces traditional rule-based mechanical control logic with a data-driven deep reinforcement learning model. This substitution enables the system to automatically adapt to market fluctuations and optimize energy storage operations, significantly improving electricity cost reduction capability while the pre-training ensures the model is ready for immediate deployment.
3Duration of action of stationary object
If the entire life cycle of the micro-grid system is considered for energy storage regulation, then the long-term electricity cost reduction is improved, but the computational complexity increases
Solution Approach 1:
The patent implements a dynamic control approach where the deep reinforcement learning model continuously adapts its parameters based on real-time market prices and system state. This dynamic adjustment allows the system to optimize energy storage regulation throughout the entire life cycle without requiring a static complex model, as the model evolves and learns from accumulated data over time.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
An energy storage control method and apparatus for a micro-grid system and a micro-grid system are provided. The method includes: for a target period within a first operation period, based on first predicted operation parameters of the micro-grid system in the target period, obtaining, via a first control model, a first control parameter of the energy storage unit in the target period; and for a target period within a second operation period, based on first historical state parameters of the micro-grid system before the target period, obtaining, via a second control model, a second control parameter of the energy storage unit in the target period. In this way, the energy storage regulation of the entire life cycle of the micro-grid system can be realized, effectively reducing the user electricity cost.