Model Card for DMP-PCFC
Advanced neural architecture , DMP-PCFC is an interpretable and accurate model for multi-step energy loads prediction in integrated energy systems, or the broader task of time series forecasting.
Model Details
Model Description
- Developed by: Xingyu Liang and Min Xia.
- Model type: DMP-PCFC (Dual-Resolution Channel Multi-Period Cross Reconstruction Parallel Closed-Form Continuous-Time Network)
- Language(s) (NLP): Not applicable
- License: MIT License
- Finetuned from model [optional]: Original implementation
Model Sources [optional]
- Repository: https://github.com/nuist-xf/DMP-PCFC
Uses
Direct Use
Energy engineers and researchers can do so directly using the DMP-PCFC framework:
- Multi-energy load forecasting for integrated energy systems (electricity, cooling, heat)
- Multi-resolution dynamic capture and long-term trend analysis
- Interpretable feature interaction analysis based on biological neuron dynamics
- Predicting sudden changes in energy demand patterns under extreme climate events such as hurricanes and heat waves
- Provides highly accurate input signals for demand response systems
Downstream Use [optional]
- Smart Grid Real-Time Dispatch System
- Energy consumption optimisation for industrial IoT devices
- Assessing the carbon reduction potential of renewable energy alternatives
Out-of-Scope Use
- Non-periodic time series forecasting (e.g., sudden event detection)
- Non-energy sector forecasts (e.g., financial time series)
- Unstructured data processing such as image/text
Bias, Risks, and Limitations
- Training data limitations: current validation of an IES system based on climatic conditions in Arizona, USA
Recommendations
- Recommended for migrated learning in conjunction with local data
How to Get Started with the Model
All data acquisition, preprocessing, loading, hyperparameters of the model, inference speed, number of parameters and the rest of the relevant content about the experiment have been presented in the repository： https://github.com/nuist-xf/DMP-PCFC
Training Details
Training Data
More Information Needed
Training Procedure
More Information Needed
Training Hyperparameters
- Training regime: More Information Needed
Speeds, Sizes, Times [optional]
More Information Needed
Evaluation
Testing Data, Factors & Metrics
Testing Data
More Information Needed
Factors
More Information Needed
Metrics
More Information Needed
Results
More Information Needed
Summary
More Information Needed
Environmental Impact
- Hardware Type: NVIDIA GeForce RTX 3090
- Hours used: More Information Needed
- Cloud Provider: More Information Needed
- Compute Region: More Information Needed
- Carbon Emitted: More Information Needed
Model Card Authors
Xingyu Liang
Model Card Contact
For technical support or data access requests:
- Liguo Weng
📧 [email protected] 🏛 Nanjing University of Information Science & Technology