Gas biogas identification method for extracting typical event characteristics based on machine learning

The invention discloses a gas marsh gas identification method for extracting typical event characteristics based on machine learning. The method comprises the following steps: step 1, data acquisition: acquiring methane alarm data acquired by monitoring equipment installed in an underground space adjacent to a gas pipeline in real time; step 2, data preprocessing: preprocessing the collected methane alarm data, including missing value filling, abnormal value processing and data resampling; step 3, constructing a sample data based on an early warning event field verification result, dividing historical methane early warning events into gas leakage and biogas accumulation, and incorporating methane alarm concentration data into a sample database corresponding to each event to construct a gas leakage and biogas accumulation sample database; according to the method, a large number of gas leakage and biogas gathering event sample features are extracted, a gas leakage and biogas gathering event sample library is co