Automating Power Quality Analysis

Abstract: The increased requirements on supervision, control, and performance in modern power systems make power quality monitoring a common practise for utilities. Large databases are created and automatic processing of the data is required for fast and effective use of the available information.

Aim of the work presented in this thesis is the development of tools for automatic analysis of monitoring data and in particular measurements of voltage events. The main objective of the analysis is the identification of the event origin (event classification). It is shown that event classification can be achieved by considering the voltage magnitude of the three phases. In the group of events that cause a temporary decrease in voltage magnitude (voltage dips) three classes are found: fault-induced events, transformer saturation events and induction motor starting events. Measurements and simulations are used for the analysis of these events. Emphasis is given on fault-induced events that present different stages of magnitude (multistage dips) and transformer saturation dips.

Different aspects regarding voltage magnitude estimation are studied using Kalman filtering. Two segmentation algorithms are proposed to divide voltage waveforms into several possible events. Kalman filtering is also used for voltage dip detection.

An expert system is developed for automatic event classification and analysis. The expert system uses the voltage waveforms and distinguishes the different types of voltage dips as well as interruptions. A method for classification is used based on the proposed segmentation algorithms. The expert system is tested using over 900 measured voltage recordings. The results show that the expert system enables fast and accurate analysis of power quality measurements. One more method for automatic event classification is proposed. The method uses discrete rms voltage measurements. Discrete rms voltage measurements form a memory saving option that power quality monitors offer instead of saving the actual voltage waveforms. It is shown that classification is possible even with rms measurements using the segmentation-based approach.

Power system transients are also studied. Measurements and simulations are used for analysis of these events. Aspects related to the frequency contents of these events are discussed.

Overall, this thesis shows that automatic processing of power quality monitoring can be achieved by following a number of well-defined steps. Automatic classification can be applied to large databases and simplify the processing and analysis of monitoring data.