Signal Change Detection Using a Novel Algorithm with Matlab

Signal change / event / anomaly detection using a novel short-time conditional local peaks rate feature.

The present code is a Matlab function that implements a new time-domain based method for signal change detection using a frame-by-frame approach and a novel signal feature named “conditional local peaks rate” (CLPR) – the rate of the local signal peaks which are above their neighbors by some predefined threshold level. The basic unit of the proposed feature is “local peaks per sample” (lpps).

Several real-world examples are given in order to clarify the usage of the function. They indicate that CLPR could serve as a good detection procedure for a variety of anomalies or events in data. The examples show that in most of the cases the CLPR outperforms the classic change detection methods – short-time energy, short-time zero-crossing rate and short-time kurtosis.

For convenience, the input and output arguments are given in the beginning of the function.

The code is based on the theory described in:

[1] H. Zhivomirov, N. Kostov. A Method for Signal Change Detection via Short-Time Conditional Local Peaks Rate Feature. Journal of Electrical and Electronics Engineering, ISSN: 1844-6035, Vol. 15, No. 2, Oct. 2022, pp. 106-109, 2022. (https://electroinf.uoradea.ro/images/articles/CERCETARE/Reviste/JEEE/JEEE_V15_N2_OCT_2022/ZHIVOMIROV_JEEE.pdf)