A method for power system frequency security control based on convolutional neural network

Abstract

The invention discloses a power system frequency security control method based on a convolutional neural network, which includes: autonomously mining the time-series feature quantity of the power system frequency in wide-area measurement data as input data, and performing preprocessing operations on the data, and Define three kinds of frequency indicators as output data; use deep learning architecture to establish nonlinear mapping relationship between time series feature quantity and frequency security to realize end-to-end frequency security assessment; optimize key parameters of power system frequency security assessment model to improve assessment Accuracy, with the help of the measurement error of wide-area measurement data and the analysis of wind power penetration rate, the anti-jamming performance of the proposed frequency security assessment method is discussed. The invention realizes the rapid evaluation of the frequency security of the power system under the power disturbance, and effectively makes up for the shortcomings of the traditional method limited by the data processing ability and the generalization ability.

Description

technical field [0001] The invention relates to the field of power systems, in particular to a method for frequency security control of power systems based on convolutional neural networks. Background technique [0002] With the access of power electronic equipment such as large-scale renewable energy and direct current transmission systems in my country [1-2] , making the dynamic characteristics of power system frequency, voltage and power angle more complex and changeable [3] , seriously threatens the safe and stable operation of my country's power system, and frequency security is an important basis for evaluating the anti-interference ability of power systems, which can be comprehensively judged according to the maximum frequency, quasi-steady-state frequency and maximum frequency change rate indicators when active power is disturbed System frequency security [4] , since the access of power electronic equipment will further aggravate the high-dimensional nonlinear chara...

AI Technical Summary

Problems solved by technology

In terms of power system feature construction, the "three-stage" feature construction method is generally adopted, which includes part of the time dimension information (steady state time, fault initial time and fault end time), but it is difficult to capture the overall time dimension information of power system operation. Limits the performance improvement of power system frequency safety assessment

In terms of evaluation model construction, when the system scale is large, it is prone to serious problems such as dimensionality disaster and low accuracy

The convolutional neural network (CNN), an important branch of machine learning, uses massive wide-area measurement data as the driving force, autonomously learns the characteristic connotation of the wide-area measurement data in the whole process of disturbance, and builds a model containing A deep learning framework with multiple hidden layers, CNN has been used in wind farm power prediction [12] , small disturbance stability evaluation [13] , Power Transformer Fault Diagnosis [14] etc. have been widely used, but the research on power system frequency security assessment is still lacking.

Images