With the continuous increase in the proportion of new energy installed capacity, the operation mode of the power system is undergoing profound changes: the output of wind and photovoltaic power fluctuates with wind speed and light, and the traditional "experience based scheduling" is becoming increasingly difficult to cope with minute level supply-demand imbalances. Artificial intelligence has become one of the key technologies in the construction of new power systems, and its applications are mainly concentrated in the following four aspects.
1、 New energy output forecast
Wind and photovoltaic power generation are significantly affected by weather, and their output curves fluctuate greatly. A machine learning model based on numerical weather forecasting and historical power generation data can significantly improve the accuracy of short-term and ultra short term power forecasting, help scheduling agencies arrange backup capacity in advance, and reduce wind and solar power curtailment.
2、 Load forecasting and electricity management
The power load is influenced by multiple factors such as temperature, holidays, industrial structure, and user behavior. AI models can integrate meteorological, calendar, and historical electricity data to provide finer grained load forecasting, support spot trading quotes and demand side response, and guide users to use electricity off peak.
3、 Intelligent dispatch of power grid and equipment operation and maintenance
On the scheduling side, AI assisted decision-making systems can simultaneously consider safety constraints, economy, and new energy consumption goals, providing better unit combinations and maintenance plans; On the operation and maintenance side, unmanned aerial vehicle inspection combined with computer vision recognition identifies hidden dangers such as insulator damage and foreign object hanging wires, while fault diagnosis models help operation and inspection personnel quickly locate abnormalities.
4、 Virtual power plants and distributed resource aggregation
The integration of massive distributed photovoltaics, energy storage, and charging stations makes "virtual power plants" possible: AI is responsible for aggregating these fragmented resources and participating in peak shaving, frequency regulation, and electricity market transactions, which not only enhances the flexibility of the power grid but also creates revenue for resource owners.
Overall, the speed at which AI is implemented in the power industry depends on two factors: data quality and the trust and verification mechanism of humans in the scheduling scenario towards the model output. Technology itself is no longer the biggest bottleneck, engineering and mechanism design are the next focus.
[Reference source] Comprehensive compilation of industry information publicly released.