Low-Carbon Energy Planning With LLM-Guided Multi-Time Scale Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tools for low-carbon energy system design and planning lack comprehensive functionality, fail to depict chemical fields like methanol and ammonia, require manual data input, lack built-in databases, have high usage thresholds, and lack real-time optimal scheduling solutions.
Innovation Solution
A method and apparatus combining a large language model with a design and planning tool to facilitate low-carbon energy system design and planning through natural language interaction, utilizing a built-in database and optimization modules for optimal planning across multiple time scales.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing design and planning tools are used, then conventional planning scenarios are covered, but they cannot meet requirements of low-carbon energy system scenarios and lack connection with large models
Solution Approach 1:
The patent merges the large language model with the design and planning tool to create an integrated system. The large model serves as the front-end interface for natural language interaction, while the design and planning tool handles the actual optimization calculations, resolving the contradiction between adaptability and complexity by combining these two components into a unified platform
Solution Approach 2:
The integrated system provides universal functionality by supporting both natural language interaction (through the large model) and professional design and planning operations (through the optimization tool). This multi-functional approach allows the system to handle diverse low-carbon energy system scenarios while maintaining a unified interface
2Ease of operation
If the large model is used for planning, then natural language interaction is enabled, but complex mathematical calculations and optimization algorithms cannot be performed
Solution Approach 1:
The patent introduces an intermediary mechanism where the large model translates natural language queries into formal optimization problems that the design and planning tool can solve. This intermediary layer preserves the ease of natural language interaction while ensuring reliable mathematical calculations through the specialized optimization tool
Solution Approach 2:
The system replaces the need for users to directly perform complex mathematical operations with an automated computational engine. The large model handles the interface layer while the design and planning tool executes the rigorous mathematical optimization, substituting manual or simple automated processes with a more sophisticated computational system
3Adaptability or versatility
If existing tools are used, then some planning functions are available, but comprehensive and refined functionality including chemical fields and variable operating conditions is lacking
Solution Approach 1:
The patent implements dynamic capabilities by enabling variable operating conditions for power generation technologies and incorporating chemical fields such as methanol and ammonia. The system can adapt to different operational scenarios and adjust parameters dynamically, transforming static planning tools into dynamic, versatile platforms
4Productivity
If existing tools are used, then simulation-based planning is provided, but real-time optimal scheduling solutions and multi-time scale planning are lacking
Solution Approach 1:
The patent implements multi-time scale planning by performing day-ahead, intra-day, and real-time scheduling in a coordinated manner. The system prepares optimization schedules in advance while maintaining the capability for real-time adjustments, applying preliminary action principles to improve planning efficiency without sacrificing real-time responsiveness
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for designing and planning a low-carbon energy system based on source-grid-load-storage-hydrogen-chemical includes obtaining (S1) user natural language information; constructing (S2) a large model for planning the low-carbon energy system; the large model includes a large language model and a planning module; identifying (S3) a user intention by inputting the user natural language information into the large language model, and invoking the planning module based on the user intention identified; converting (S4) the user natural language information into a preset format of input data and inputting the preset format of input data into the planning module, and outputting a planning report by optimization solving of the planning module; displaying a planning result corresponding to the planning report by the large language model.