Power system natural frequency characteristic coefficient interval prediction method based on deep learning

The invention discloses a power system natural frequency characteristic coefficient interval prediction method based on deep learning. The method comprises the steps of 1) determining main influence factors of a natural frequency characteristic coefficient beta; 2) establishing a mapping relation among system power disturbance, reserve capacity, a unit start-stop mode and a natural frequency characteristic coefficient beta by using a DNN model; 3) establishing an interval prediction model of a natural frequency characteristic coefficient beta; (4) obtaining a Bootstrap training set, and training the DNN model; and 5) estimating the variance of the prediction error of the natural frequency characteristic coefficient beta, and calculating the confidence interval of the prediction result of the natural frequency characteristic coefficient beta under a given confidence level by using the interval prediction model. The method can be used as a reference basis for setting the frequency deviation coefficient B in an