Deep belief network-based voltage sag reason recognition method

Abstract

The invention discloses a deep belief network-based voltage sag reason recognition method, and belongs to the technical field of electric energy quality analysis methods. The method comprises the steps of: carrying out data preprocessing on voltage sag recording data and voltage sag reason tags; building a restricted Boltzmann machine network; carrying out non-supervision pre-training on the restricted Boltzmann machine network; building a deep belief network by using a restricted Boltzmann machine generated through the training; adding a softmax layer; carrying out supervised training on thewhole network; carrying out correctness verification on a generated model; and judging various probabilities output by the model, and automatically recognizing voltage sag reasons types correspondingto input data. According to the method, history voltage sag recoding and sag reason tags are utilized to carry out iterative training on a deep neural network, and possible voltage sag recording dataof supervision points is input into the model generated through the training to obtain corresponding voltage sag reason types. The method is supplementation to electric energy monitoring systems, andhas important practical significance.

Description

technical field [0001] The invention relates to a voltage sag cause identification method based on a deep belief network, and belongs to the technical field of power quality analysis methods. Background technique [0002] With the rapid development of industry and information society, the requirements of all walks of life for power quality are also increasing. Voltage sag is one of the unavoidable power quality disturbance events in the power system that is most likely to cause economic losses to sensitive industrial users. With the extensive use of sensitive equipment such as adjustable speed motors and precision control equipment, the losses caused by voltage sags can even be compared with power failures. In recent years, experts at home and abroad have carried out a lot of research on the feature extraction, cause classification, cause location, monitoring, and sensitive equipment tolerance of voltage sags, and have achieved many results. On the basis of the existing re...

AI Technical Summary

Problems solved by technology

The final generated model can automatically identify the cause of the corresponding voltage sag based on the wave recording data of the sag event, and at the same time, it can overcome the fact that artificial feature selection is easy to lose original information, the classification model is too complicated, and the convergence speed of traditional neural network training is slow and easy. Problems limited to local optima

Images