Power quality disturbance detection method based on yolo algorithm

Abstract

The present invention relates to a power quality disturbance signal detection technology, in particular to a power quality disturbance detection method based on the YOLO algorithm, including establishing a mathematical model of a power quality disturbance signal, using MATLAB to realize the construction of a power quality disturbance signal, traversing the disturbance signal parameter matrix, and generating power quality Disturb the sample data set; input the sample set Imagenet into the fully convolutional neural network Darknet-53, and obtain the initial weight parameters of the network after training; set the training parameters; adopt a multi-resolution training strategy to input the marked power quality disturbance sample data set into the global The convolutional neural network Darknet-53 is trained, and the network weight parameters are updated to obtain the power quality disturbance signal detection model; the power quality disturbance signal is randomly generated, and the disturbance signal is converted into a two-dimensional image and then input into the detection model to obtain the detection result. The method has high detection accuracy, wide application range, fast enough real-time monitoring, and can accurately identify various power quality disturbance signals.

Description

technical field [0001] The invention belongs to the technical field of power quality disturbance signal detection, in particular to a power quality disturbance detection method based on the YOLO algorithm. Background technique [0002] People's awareness of the importance of power quality has gradually deepened along with the development of power systems. The electrical loads in the early power system were mainly composed of linear loads, so the frequency offset and voltage offset were used to evaluate the power quality. With the large-scale use of new energy and power electronics technology, the nonlinear, asymmetrical, and impact loads within the power grid increase suddenly, which increases the frequency and types of power quality disturbances, prompting people to pay more and more attention to power quality issues. [0003] The basic idea of ​​power quality disturbance signal detection is that when a power quality disturbance occurs, by analyzing the voltage signal reco...

AI Technical Summary

Problems solved by technology

Fourier transform (FT) sampling needs to ensure that the sampling frequency is twice or more than the highest signal frequency, and may cause "spectrum aliasing" and "spectrum leakage" effects; short-time Fourier transform (FFT) is not applicable Due to the multi-scale analysis process, it is not suitable for the time window requirements of signal high and low frequency transformation; Prony analysis method is rarely used to analyze signal data in real time, because its calculation takes too long, and it is only suitable for offline power quality disturbance analysis method; although the artificial neural network has been widely used, there are also disadvantages such as the possibility of falling into local optimum and long training time; although the decision tree has the advantages of being reusable and easy to read, its algorithm itself leads to problems. When the training data is excessive or insufficient, it is easy to fall into local optimum or overfitting

[0004] The existing power quality disturbance detection methods have high requirements on the threshold value of the disturbance signal, and it is difficult to set the analysis parameters properly. These factors increase the difficulty of power quality disturbance detection, and the power quality disturbance signal itself has a large randomness and a wide variety , short duration, complex features and easy overlapping of edges

Images