Research of Machine Learning Algorithm for Broadcasting Spectrum Signal Processing

FM broadcasting is mainly used to transmit sound and other signals in the form of wireless transmission. This thesis is based on a project for a radio department, and mainly focuses on the broadcasting frequency band, uses radio monitoring equipment to scan the spectrum signal of the broadcasting frequency band, and performs the data preprocessing, signal feature extraction and classification processing of the radio frequency spectrum signal on the extracted spectrum signal, so as to extract abnormal signals such as pseudo base stations, black radio, cheats in exams, etc. Firstly, relevant pre-processing is performed on the spectrum information of the broadcast frequency band. Through the improved K-Means algorithm, the original sample data including the glitch signal is eliminated, and the original signal is processed through the wavelet analysis of the spectrum signal. The original signal was decomposed by wavelet and wavelet reconstruction, so as to achieve the purpose of denoising the original signal. Secondly, based on the analysis of signal characteristics and the comparison of a large number of spectrum signals, a method for extracting individual features of spectrum signals is summarized. Finally, the grey relational degree cluster Analysis is used to extract the features of the spectrum signal, which provides a certain basis for the subsequent classification algorithm.