Research on feature Optimization algorithm of Optical Fiber Sensor recognition system

For fiber-optic sensing signal recognition systems, signal feature extraction is the key to determining the accuracy of signal recognition and classification. Mel frequency cepstral coefficients (MFCC) can be used for feature extraction of optical fiber signals. In order to further improve the accuracy of fiber vibration signal pattern recognition, this paper proposes a feature extraction algorithm based on Compensation Distance Estimation Technology (CDET). The algorithm first extracts the MFCC feature vector from the optical fiber vibration signal, uses the compensation distance estimation technology to evaluate the feature through the feature screening strategy, and optimizes the feature by deleting the redundant vector. In order to verify the effectiveness of the algorithm, the traditional MFCC algorithm and the PCA algorithm dimensionality reduction and the proposed algorithm are used to compare the classification experiments. The experimental results show that the feature extraction method based on CDET can further improve the accuracy of the classification results by 5%. After adding different signal-to-noise ratio noises, classification experiments are performed to verify that the proposed algorithm has certain anti-noise performance.