A precision detection technique for power disturbance in electrical system

The sensitive electronic devices can be damaged or malfunction due to power disturbances in an electrical system. A fast and accurate detection of power disturbances in the system is the key to take appropriate corrective measures. This paper presents an innovative algorithm for fast and accurate detection of power disturbance. The algorithm is based on four major steps; segmentation of acquired signal resulting 10-samples frame, pre-processing, feature extraction and finally the decision between normal and disturbance signal using linear support vector machine. The performance of proposed detection technique was evaluated for two set of disturbance signals; first, group of synthetic disturbance signals generated using computer simulation and second, group of experimentally obtained real disturbance signals. The proposed method uses only two features; mean absolute deviation and energy. Hence, the algorithm reduces the computational complexity of the detection mechanism. Moreover, the system has shown tremendous performance with respect to detection accuracy. Proposed method may be implemented on microcontroller-based embedded system for a wide range of applications.