Static Stability Prediction Of Power System Based On Deep Learning

With the expansion of power system interconnection and the promotion of power market system reform,the load on the power grid is constantly increasing,and the voltage stability situation is facing challenges.Solving loadability limit quickly is the basic requirement of online assessment for static voltage stability of power system.Traditional analysis method of static voltage stability based on CPF(continuation power flow)can’t meet the real-time requirements of online applications because of its high computational complexity.Meanwhile,the construction of WAMS(Wide Area Measurement System)and the development of smart grid bring massive amounts of real data,which provide an opportunity for deep learning methods to be applied in the field of electricity.In this paper,we propose a method for assessing the static stability of power systems based on deep learning.Different from the method based on BP neural network,the method we proposed is more concise,which can automatically learn features without the need for pre-characterization of data.Specifically,we propose a deep neural network model that contains three kinds of layers,including convolutional layers,pooled layers,and fully connected layers.The model takes the operation data of power grid as input and outputs the predicted value of loadability limit.Key features extracted by the convolutional neural network in the model are input to the MLP(Multi-Layer Perceptron)for calculating the predicted value as the output of the model.We simulate the static stability limit data for training and testing model on IEEE 39 system by using CPF method.The deep neural network model we proposed has achieved very good results in both prediction accuracy and efficiency,and shows more excellent performance in comparison with other models.