A novel method for extracting mutation points of acoustic emission signals based on cosine similarity

Mutation point extraction in acoustic emission (AE) signals is always a complex and challenging task since the mutation is often hidden in AE signals and with a short duration. This paper proposes a novel method based on cosine similarity (CS) to detect change points in AE signals. Disregarding the specific value of AE signals, the proposed method extracts the similarity features from the adjacent waveforms. Compared with traditional AE analysis and state-of-art methods, the proposed method performs better for extracting mutation points in an AE monitoring laser scanning experiment. A combination of a de-negative step and a linear normalization step is applied in the preprocessing procedure to efficiently eliminate the oscillation and vibration in CS calculation. Key parameters (window length and sampling frequency) are demonstrated to affect the mutation points extracting accuracy. The proposed CS method provides an alternative for mutation extraction in AE signals, and can be used in other practical applications.