An Intelligent Classification Method for Vibration Signals Based on Random Forest

Due to the complexity of fiber optic vibration signals, using traditional classification methods to classify signals is inaccurate. Effective fiber optic sensor data classification has become an important research topic. Machine learning methods have been proposed for intelligent signal recognition and classification. This study aims to explore the use of machine learning methods to recognize and classify collected signals intelligently. The vibration signal data collected from the fiber optic sensors is first preprocessed, including removing trend terms and applying smoothing filters. Then, time-domain and frequency-domain features are extracted separately to construct a dataset for training and testing purposes. This paper focuses on the classification results of the Random Forest (RF) classifier and compares it with three other machine learning algorithms: Support Vector Machine (SVM), Back Propagation Neural Network (BP), and Radial Basis Function Neural Network (RBF). The results show that the Random Forest model optimized through feature selection outperforms other models in terms of classification accuracy, recall rate, and F1 score.