Solar cell panel defect detection method based on multi-scale joint convolutional neural network

Abstract

The invention discloses a solar cell panel defect detection method based on a multi-scale joint convolutional neural network. General image data processing of a neural network is involved. Firstly, asolar cell panel image sample is collected and data is processed; balancing the sample and then carrying out pretreatment operation; dividing the processed various types of images into a training setand a test set; building an experiment software environment and an experiment hardware environment after the data set is obtained; constructing a multi-scale joint convolutional neural network model;inputting the marked training set into the model for learning; and the test set is used for testing to complete the detection of the defects of the solar cell panel, so that the defects of single function, weaker generalization ability, incapability of ensuring the accuracy in an actual detection task, difficulty in universality of a set of parameters, fewer detectable categories, larger calculated amount and lower efficiency in the prior art of the defect recognition or detection method of the solar cell panel are overcome.

Description

technical field [0001] The technical proposal of the present invention relates to the general image data processing of neural network, in particular to the solar cell panel defect detection method of multi-scale joint convolutional neural network. Background technique [0002] With the development of society, the demand for green energy continues to increase. Due to the renewable, clean and pollution-free features of solar energy, the photovoltaic industry has developed rapidly, and solar panels are the core components of energy conversion, and their output is also rising. The battery board may be damaged during the process of production, transportation, installation and use. These defects caused by damage will seriously reduce the photoelectric conversion efficiency and service life of the solar panel. Therefore, it is necessary to detect the defect of the solar cell panel in time before using it, and replace the defective solar cell panel. In recent years, with the matu...

AI Technical Summary

Problems solved by technology

[0007] The technical problem to be solved by the present invention is to provide a multi-scale joint convolutional neural network solar panel defect detection method, first collect solar panel image samples and data processing, perform preprocessing after balancing the samples, and process each The various types of images are divided into training sets and test sets. After obtaining the above data sets, an experimental software environment and an experimental hardware environment are built, and then a multi-scale joint convolutional neural network model is constructed, and then the marked training set is input into the model for learning. And use the test set to test, complete the detection of solar panel defects, overcome the existing technology of solar panel defect identification or detection methods with single function, weak generalization ability, unguaranteed accuracy in actual detection tasks, and a set of parameters Defects that are difficult to generalize, have fewer detectable categories, and are computationally intensive and less efficient

Images