Feature extraction technology of unstructured data for intelligent inspection terminal
Most of the power equipment data collected by intelligent inspection terminals are unstructured data such as images,videos and sounds,which are characterized by complexity and diversity.The accuracy of unstructured data extraction determines the monitoring capability of power equipment.Therefore,a new feature extraction technology is proposed for unstructured data in intelligent patrol terminal.Identify the image feature values,video feature values,and sound feature values of the data in the intelligent inspection terminal separately.Based on the recognition results,after normalization,K-L transformation is used to complete the dimensionality reduction of data samples,and the feature extraction of unstructured data of intelligent patrol terminal is realized.The experimental results show that the difference between the sample length of structured data extracted by the proposed method and the sample length of transmission and distribution data required by intelligent patrol terminal host is always less than 0.05×109 MB,improving the accuracy of unstructured data feature extraction.
万方数据
维普
