A Pattern Recognition Method for Vibration Events

Abstract

The invention discloses a method for pattern recognition of vibration events, comprising the following steps: collecting and acquiring an original vibration signal through a vibration sensor, the original vibration signal including a vibration signal and a non-vibration signal; and dividing the vibration signal in the original vibration signal; Perform denoising processing on the vibration signal; perform feature extraction on the denoised vibration signal to obtain a feature vector, including three features: feature A. Perform wavelet packet decomposition in the time-frequency domain to obtain an energy feature vector; feature B, Perform cepstral analysis to extract cepstral parameter features; feature C, extract signal features in the time domain; establish a recognition model, which is composed of two-level classifiers; the first-level classifier is based on the support vector machine SVM classifier, using from vibration The feature vector extracted from the signal is used as input, and the vibration events are divided into non-intrusion events and intrusion events; the secondary classifier is to identify the intrusion events based on artificial neural network and obtain the classification results.

Description

technical field [0001] The invention relates to the field of pattern recognition, in particular to a method for pattern recognition of vibration events. Background technique [0002] Vibration sensors are everywhere along national borders, railways, industrial pipelines, important places, and residential quarters, and have been widely used in security monitoring and other fields. For example, in the defense along the railway line, potential safety hazards can be found, so that the staff can prevent them in advance; another example is the perimeter defense, which can be used in places such as homes, schools, residences, important institutions, financial centers, etc., to detect external intrusions immediately; in addition, Vibration sensors have also found important applications in earthquake and tsunami detection. However, at present, commercial perimeter intrusion warning systems have appeared at home and abroad, and most of them can only be used to locate the location of ...

AI Technical Summary

Problems solved by technology

However, at present, commercial perimeter intrusion warning systems have appeared at home and abroad, and most of them can only be used to locate the location of possible abnormal events. The method used to identify intrusion behavior is usually to analyze the time domain and frequency domain of the vibration signal, but this method It is somewhat single and limited for identifying complex vibration events

The recognition rate of intrusion behavior is low, that is, it is impossible to effectively distinguish interference events from real intrusion events, resulting in the inability to accurately know the way of intrusion behavior, so that intrusion events cannot be effectively controlled and disposed of

Images