Power quality disturbance identification method and system based on variable recurrent neural network, and medium

Abstract

The invention discloses a power quality disturbance identification method and system based on a variable recurrent neural network, and a medium. The method comprises the following steps: extracting power quality disturbance waveform data characteristics, inputting the characteristics into a pre-trained variable recurrent neural network, and generating an output probability curve corresponding to Ndisturbances through an activation function; wherein the variable recurrent neural network comprises a random layer used for introducing a potential random variable to the output layer, and finally determining the disturbance type at a certain moment and / or the start and stop moments of a certain disturbance according to the output probability curve. According to the invention, the invention canachieve the recognition and classification of power quality disturbance types at any time, can accurately recognize the start and stop time of disturbance, and can achieve the multi-node, quick and accurate power quality disturbance recognition after the algorithm is introduced into a common household electric meter.

Description

technical field [0001] The invention relates to the technical field of power quality detection, in particular to a method, system and medium for recognizing power quality disturbances based on a variable cyclic neural network. Background technique [0002] The current existing power quality monitoring methods generally use signal processing (short-time windowed Fourier, wavelet, variational mode decomposition and other power quality disturbance detection algorithms) to perform disturbance detection and extraction features, and then use Bayesian criterion, A series of algorithms such as k nearest neighbor, fisher criterion and artificial neural network are used for identification. However, because these methods are not targeted at the processing of sequence data, they can only identify disturbances by time period. The neural network structure can make full use of a large amount of existing power quality data. After adding new data during model training, the recognition accur...

AI Technical Summary

Problems solved by technology

The entire network structure of the ordinary cyclic neural network is completely determined, the front and back are logically related, the probability affects each other, the randomness is lost, and problems such as gradient bursts are prone to occur.

Images