关键词: 光伏功率等级分类, 深度学习, 残差网络, Mobilenet, 轻量化, 平行网络

Abstract: To improve the accuracy of photovoltaic power generation level classification to meet industry requirements, this paper proposes a classification model based on Res-MobileCom parallel network. In data processing, denormalized linear interpolation is used to retain data features to the greatest extent. Then, by simplifying the structures of Resnet and Mobilenet, their outputs are jointly input into the combination network after parallel training to further extract data features, and finally the classification results are obtained. Experiments show that compared with common deep learning models, the model proposed in this paper retains the feature extraction capability and lightweight nature of Resnet and Mobilenet. At the same time, the accuracy has been significantly improved, and the model also has good balance performance and generalization ability. In addition, experiments also find that the values of random numbers in the combination network can improve the accuracy of the model, but it lacks Interpretability. This paper proposes the denormalized linear interpolation in data processing and the combination network for further extracting data features, which improves the accuracy of the model by over 10%. It provides new methods and ideas for improving the accuracy of the photovoltaic power generation level classification model.

Key words: photovoltaic power rating classification, deep learning, residual network, mobilenet, lightweight design, parallel network