Arc detection method and system

Abstract

The embodiment of the invention discloses an arc detection method and system. The arc detection method comprises the following steps: station end equipment detects an arc through a first arc detection model, and uploads an arc detection sample to model training equipment to update an original arc detection sample in the model training equipment; the model training device retrains a second arc detection model according to the updated arc detection sample to obtain an updated second arc detection model; and the model training equipment updates the first arc detection model according to the updated second arc detection model. Compared with the prior art, according to the embodiment of the invention, the models are respectively trained for different application scenes, the fault tolerance rate and universality requirements of arc detection can be met, and the sensitivity requirement of arc detection can also be met.

Description

technical field [0001] Embodiments of the present invention relate to the field of photovoltaic technologies, and in particular to an arc detection method and system. Background technique [0002] Arcing is a gas discharge phenomenon that produces high temperatures, brightness and radiation. When an arc occurs, if the arc is not extinguished in time, it may cause cable damage, loss of power generation, or even fire. In photovoltaic systems, one of the common arc types on the DC side is the series arc. In order to cut off the branch in time when an arc occurs, it is necessary to add an arc detection function in the photovoltaic system. [0003] In the prior art, the arc detection method mostly collects the AC component of the DC current (referred to as the current signal), and analyzes the signal in the time domain and frequency domain, extracts the characteristic parameters and compares them with the preset thresholds to determine whether an arc occurs. The use of machine...

AI Technical Summary

Problems solved by technology

[0004] However, taking the photovoltaic system as an example, affected by the characteristics of photovoltaic power generation, the inverter models and grid interference signals are different in different application scenarios. It is difficult to cover all working conditions when establishing the initial training set

In addition, using sample data containing various working conditions to train the model improves the fault tolerance and universality of the model, but reduces the sensitivity of the model—the model cannot be optimized for a specific working condition

Images