Electricity price prediction method, device, equipment, storage medium and program product
By integrating power grid data to form a time-series dataset and utilizing feature extraction and prediction models, the problems of data feature loss and period limitations in existing electricity price prediction methods are solved, achieving high-precision electricity price prediction and long-term adaptability, and optimizing the scheduling of energy storage systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING HYPERSTRONG TECH CO LTD
- Filing Date
- 2026-04-14
- Publication Date
- 2026-07-21
AI Technical Summary
Existing electricity price forecasting methods rely on multi-stage information distillation, which leads to the loss of original data features. This makes it difficult to meet the high-precision requirements of energy storage power stations for accurate peak and valley identification, and it cannot be dynamically extended to longer periods for forecasting, resulting in energy storage system scheduling decisions being limited to local optima.
By acquiring power grid data and integrating it into a time-series dataset, and utilizing time-series features and electricity price prediction models, predictions are made based on the dependency relationship between historical and current electricity prices. Convolutional networks and long short-term memory neural network models are used to extract feature correlation vectors, and sliding window technology is combined to dynamically adjust the data length, thereby improving the accuracy of electricity price prediction and the ability to predict long-term periods.
It improves the accuracy and adaptability of electricity price forecasting, can dynamically extend the forecast period, meet the high-precision peak-valley identification requirements of energy storage power stations, and optimize the scheduling decisions of energy storage systems.
Smart Images

Figure CN122434586A_ABST