Research on Power Image Big Data Classification Based on Deep Learning
Compared with ordinary image data classification,the similarity between power image data is higher,which brings greater difficulty to the data classification work.In order to improve the accuracy of power image big data classification,a deep learn-ing based power image big data classification method is proposed.By using data mining to collect big data from power images,prepro-cessing of the initial power image big data is completed through steps such as grayscale and filtering.By utilizing convolutional neural networks in deep learning algorithms and training through forward and backward propagation,the extraction results of big data features in power images are obtained.Finally,through feature matching,the type of power image big data is determined,and the data classi-fication work is completed.Through performance testing experiments,it was concluded that compared with traditional methods,the optimized design method has improved classification accuracy and recall by 4.8％and 2.35％,respectively,indicating that the opti-mized design method has significant advantages in classification performance.
deep learningpower imagepower big datadata classification
万方数据
维普
