A photovoltaic system arc fault detection method based on machine learning and multi-time-frequency features

Abstract

The invention discloses a photovoltaic system fault arc detection method of machine learning with fusion of multiple time-frequency characteristics. A signal xn is acquired by a time window of the length of Ts, and the corresponding square matrix distribution form of the xn in a time-frequency domain is obtained through Gabor transformation; and the corresponding matrix distribution form of the xn in the time-frequency domain is obtained through Radon-Wigner transformation, integration of the time dimension is performed on the elements of the two matrixes, the specific component of the frequency dimension is selected to be processed in different ways so as to obtain the corresponding multiple time-frequency characteristics, the multiple characteristic quantities are fused based on the trained hidden Markov model and thus whether the photovoltaic system has fault arc in the current time window can be judged. Multiple effective time-frequency characteristics are fused to accurately identify multiple fault arc forms in the grid-connected photovoltaic system so that fault arc operation can be accelerated and misoperation of multiple types of arc conditions is ensured not to occur, and thus the capacity of safe and stable operation of the grid-connected photovoltaic system can be enhanced and the problem of potential misoperation of the grid-connected photovoltaic system in case of facing the external interference can be solved.

Description

technical field [0001] The invention belongs to the technical field of photovoltaic electrical fault detection, and specifically relates to a fault arc detection of a photovoltaic system that uses Gabor transform and Raydon-Wegener transform to obtain multi-time-frequency feature quantities and uses an implicit Markov model to fuse multi-time-frequency feature quantities method, thereby speeding up the operation of the corresponding fault branch under fault arc conditions, ensuring that various types of arc conditions do not occur in error, and improving the safe and stable operation of the grid-connected photovoltaic system. Background technique [0002] Before the large-scale application of photovoltaic products, the most widely used is alternating current. For the prevention and control of AC fault arcs, there are corresponding regulations, standard test methods and corresponding industrial products, such as AC fault arc circuit interrupters (ACAFCI). As early as 1999, t...

AI Technical Summary

Problems solved by technology

However, when the grid-connected photovoltaic system is running, the power output of the photovoltaic system is greatly affected by the operating environment. It is possible to disturb the power output signal of the grid-connected photovoltaic system and form an arc-like working condition similar to the fault arc to interfere with the correct judgment result of the fault arc

If the detection method malfunctions when an arc-like working condition occurs, an undesired shutdown will occur during the normal operation of the grid-connected photovoltaic system, which will greatly reduce the running time of the grid-connected photovoltaic system and reduce the operating efficiency of the grid-connected photovoltaic system

Images