Simple and Practical MATLAB Code Implementation with Feature Extraction

In the field of signal processing, feature extraction is a crucial step for distilling key information from raw data. MATLAB serves as a powerful scientific computing tool particularly well-suited for implementing such tasks. This article introduces a simple yet practical MATLAB program designed for data feature extraction. The program selects multiple key features from both time-domain and frequency-domain perspectives: Time-Domain Features Waveform Factor: An indicator reflecting signal waveform shape that can differentiate various signal types Kurtosis: A statistical measure describing the sharpness of signal distribution, highly sensitive to impulse signals Frequency-Domain Features Frequency Band Energy Percentage: Reveals frequency characteristics by analyzing energy distribution across different frequency bands The implementation follows a modular design approach: first preprocessing input signals, then computing individual features separately, and finally integrating them into a feature vector output. This methodology maintains code simplicity while facilitating future addition of new feature indicators. The program employs MATLAB's built-in functions like kurtosis() for statistical calculations and fft() for frequency analysis. For frequency band energy calculation, the code typically uses bandpass filtering or spectral segmentation followed by energy summation within target bands. This program is applicable to various scenarios including mechanical equipment monitoring and biological signal analysis. Users can adjust feature combinations according to specific requirements to balance computational efficiency with feature representation capability. The modular structure allows easy customization through function modifications or additional feature modules.