Electromagnetic anomaly identification algorithm based on signal fingerprinting

This paper makes research on recognition algorithm of short-time （<10 min） geomagnetic anomaly data, based on signal fingerprint, to recognize the categories of geomagnetic disturbance signals. The signal fingerprinting used in this study can convert the waveform data for a given period of time into a 32-bit integer, basing on the combination of multiple data and text-processing methods, such as Fourier transform, wavelet transform, signal binarization and MinHash, which greatly compresses the feature information of signals, thus greatly reduces the amount of data to be located and classified in the following research. The experiment uses raw second-scale data of two sets of GM4 fluxgate magnetometers recorded at Hongshan geomagnetic station (LYH) in Longyao city of Hebei Province from 1 to 3 May 2016, and the results indicate that the algorithm in this paper can quickly and accurately recognize categories of interference signals, and provides technical support for automatic abnormal signals extraction of geomagnetic relative record data.