Radial basis function neural network application to power system restoration studies

Area of Science:

• Electrical Engineering
• Power Systems
• Artificial Intelligence

Background:

• Overvoltages during transformer switching pose a significant risk to power system equipment and restoration efforts.
• Accurate prediction of these transient overvoltages is crucial for ensuring grid stability and reliability.

Purpose of the Study:

• To develop and validate a novel method for accurately estimating transformer switching overvoltages.
• To enhance the generalization capability of the proposed model for diverse operational scenarios.

Main Methods:

• A radial basis function neural network (RBFNN) was developed to model transformer switching overvoltages.
• Equivalent network parameters were incorporated as inputs to the RBFNN to improve generalization.
• The RBFNN was trained using worst-case switching angles and remanent flux, then tested on typical system cases.

Main Results:

• The proposed RBFNN demonstrated high accuracy in estimating the peak values of switching overvoltages.
• The model accurately predicted the duration of the switching overvoltages.
• Simulations on a partial New England 39-bus test system validated the technique's effectiveness.

Conclusions:

• The developed RBFNN provides an accurate and reliable method for estimating transformer switching overvoltages.
• This technique can significantly aid in power system restoration by mitigating potential equipment damage.
• The incorporation of equivalent parameters enhances the model's robustness for practical applications.