Development of an algorithm for detection and classification of capacitor switching events

Shunt capacitor banks are widely used to improve power system operation by injecting reactive power into predominantly inductive systems. These banks may be continuously energized or switched according to the load levels. The primary goal of this work is to develop an algorithm for detection of capacitor switching events, both energizing and de-energizing. This identification allow us to assess the capacitor performance and detection of an unsuccessful energization, restrikes during the de-energizing operation, blown fuses, or even failed capacitors in one of the phases. The identification algorithms are developed based on the unique features associated to capacitor switching events, such as wavelet transform coefficients, voltage gradient at the switching instant, inrush current, steady-state voltage rise, and reactive power support. Moreover, a library with common range of values for these parameters is used to improve the algorithm accuracy. For each identified capacitor switching event, the following parameters are computed: peak voltage and current, voltage phase angle at the switching instant, duration and frequency of the transient, reactive power support in each phase, steady-state voltage rise, relative location of the switched bank, variation of power factor, voltage total harmonic distortion, and RMS line current. This capacitor switching identification module is tested with data from a power quality monitor, including different types of events. The results show that both energizing and de-energizing algorithms provide high accuracy levels. Moreover, using the scores assignment improves the performance of the capacitor energizing identification algorithm. As an application, these algorithms are used to evaluate the power factor over compensation in a distribution utility.Electrical and Computer Engineerin