Risk early warning method based on coal mine accident disaster feature matching

The invention discloses a risk early warning method based on coal mine accident disaster feature matching, which relates to the technical field of coal mine accident disaster early warning, and comprises the following steps of: capturing amplitude feature information of a microseismic signal in real time by using a sensor at an initial sampling frequency, and transmitting the captured amplitude feature information to a ground monitoring center; amplitude characteristic information is preliminarily filtered, and the basic quality of data is ensured; and carrying out anomaly analysis on the filtered amplitude characteristic information, and preliminarily identifying potential abnormal micro-seismic signals. According to the invention, machine learning is utilized to evaluate a microseismic signal, the microseismic situation is divided according to a frequency spectrum peak value and a stress release rate, the early warning accuracy and real-time performance are improved, the false alarm rate is reduced, the man