Vuk Malbasa

Regression tree for stability margin prediction using synchrophasor measurements

A regression tree-based approach to predicting the power system stability margin and detecting impending system event is proposed. The input features of the regression tree (RT) include the synchronized voltage and current phasors. Modal analysis and continuation power flow are the tools used to build the knowledge base for offline RT training. Corresponding metrics include the damping ratio of the critical oscillation mode and MW-distance to the voltage collapse point. The robustness of the proposed predictor to measurement errors and system topology variation is analyzed. The optimal placement for the phasor measurement units (PMUs) based on the importance of RT variables is suggested.

Sensitivity Analysis of Voltage Sag Based Fault Location With Distributed Generation

The presence of distributed generation (DG) in distribution networks may seriously affect accuracy of the voltage sag based fault location method. An approach toward quantifying the adverse effect of DG on the fault location calculation is described. A series of realistic scenarios is used to illustrate how DG impacts synchrophasor measurements during disturbances. Alternative Transients Program-Electromagnetic Transients Program models are used to obtain steady-state solutions in the time domain, while Sobols approach to sensitivity analysis is used to quantify the effect of DG and imperfections of measuring instruments on fault location. Various test cases reveal that DG may adversely affect the voltage characteristic and therefore the accuracy of voltage sag based fault location.

Locating sub-cycle faults in distribution network applying half-cycle DFT method

It is necessary to accurately detect and locate sub-cycle faults in order to prevent unexpected outages. However, conventional fault location methods cannot locate these faults as typically data windows longer than the faults signature are used for phasor extraction. This paper presents an overall analysis of how the single-phase-ground sub-cycle fault in the distribution network can be located using voltage sag. The half-cycle Discrete Fourier transform is used for phasor extraction in the timedomain simulations. Our results reveal that the proposed approach is capable of accurately locating sub-cycle faults whose duration is between 0.5 and 1.0 cycles. The results also suggest that the placement of meters may significantly affect the capability of the proposed approach to locate sub-cycle faults.

Sensitivity analysis of voltage sag based fault location algorithm

In this paper, the global, variance-based, sensitivity analysis is used to quantify the impact of measurement imperfections on voltage sag based fault location. This kind of fault location requires voltage phasor information from meters to be compared to simulated cases in order to locate faults. However, meters are prone to measurement imperfections. It is therefore critical that the impact of measurement imperfections, such as measurement and loading errors, are fully assessed to account for uncertainty in algorithm inputs. Sensitivity analysis was used to attribute responsibility for uncertainty in fault location to uncertainty in the inputs of the fault location algorithm. The results demonstrate that the single most detrimental factor to precise fault location is large fault resistance, both alone and in combination with other factors. Although accurately deduced by the algorithm, other impacts of this fault feature adversely impact accuracy.

Uncertainty of measurement error in intelligent electronic devices

This paper focuses on methodology to quantify uncertainty in measurements obtained from Intelligent Electronic Devices (IED). IEDs have emerged in distribution systems as a prevalent source of measurements in monitoring and protection, as well as for different kinds of applications beyond IEDs primary purposes. These measurement devices are installed across a system, from substations down to the customer locations, and provide measurements of a wide array of quantities. We report how IED measurements respond to external disturbances, which may lead to possible accuracy impacts in various applications. The example used to illustrate the approach is highly accurate fault location in distribution systems based on voltage sag measurements.

Analysis of voltage stability issues with distributed generation penetration in distribution networks

This paper presents an overall analysis of how the penetration of distributed generation in low-voltage secondary distribution networks affects voltage stability. It is critical that the voltage collapse point be carefully studied under different system operating points to prevent degradation of service. System components have been sophisticatedly modeled in ATP/EMTP. DGs are allocated in a probabilistic fashion to account for uncertainties in future allocation. A large number of experiments under both light and peak load conditions have been carried out to provide realistic results. Results indicate that voltage stability is positively correlated with penetration of DG, but large induction type DG may lower the voltage stability margin.

Voltage Stability Prediction Using Active Machine Learning

An active machine learning technique for monitoring the voltage stability in transmission systems is presented. It has been shown that machine learning algorithms may be used to supplement the traditional simulation approach, but they suffer from the difficulties of online machine learning model update and offline training data preparation. We propose an active learning solution to enhance existing machine learning applications by actively interacting with the online prediction and offline training process. The technique identifies operating points where machine learning predictions based on power system measurements contradict with actual system conditions. By creating the training set around the identified operating points, it is possible to improve the capability of machine learning tools to predict future power system states. The technique also accelerates the offline training process by reducing the amount of simulations on a detailed power system model around operating points where correct predictions are made. Experiments show a significant advantage in relation to the training time, prediction time, and number of measurements that need to be queried to achieve high prediction accuracy.

Sensitivity of voltage sag based fault location in distribution network to sub-cycle faults

Single-phase-to-ground sub-cycle faults in the distribution network can be located using voltage sag fault location. This paper illustrates how a sensitivity study of measurement imperfections can be used to quantify the impact of sub-cycle faults on voltage sag based fault location. Our results suggest that there is a complex relationship between factors influencing error in fault location because the design of the study covered a wide range of conditions. The more complicated, higher order interactions have a stronger influence on error than any particular input factor alone.

Improved Transmission Line Fault Location Using Automated Correlation of Big Data from Lightning Strikes and Fault-Induced Traveling Waves

This paper investigates how correlating cross-domain big data from the lightning surge and the traveling wave measurements in time and space can be used to improve fault location accuracy. The integration and correlation of big data in time and space using Global Positioning System and Geographic Information System respectively improves knowledge about faults on transmission lines caused by lightning. The benefits of proposed method are: (a) the decision process can be accelerated through automation, and (b) better accuracy of fault location result can be provided due to the data correlation. The benefit is a more efficient outage management procedure.

Power system online stability assessment using active learning and synchrophasor data

Analysis of synchrophasor measurements using data mining tools, in pursuit of precise stability assessment, requires a sufficiently large training data set. Traditionally the process of learning the underlying power system behavioral patterns introduces a significant computational burden such that exhaustive simulations of all possible system operating conditions are necessary. Advancements in machine learning make it possible, in some cases, to reduce the amount of operating conditions that need to be analyzed without impacting the accuracy of stability assessment. By using a probabilistic learning tool in the described active learning scheme to interactively query operating conditions based on their importance, we show that significantly fewer data needs to be processed for accurate voltage stability and oscillatory stability estimation. Results show that the advantage of active learning is greater on more complicated power networks, where larger training data sets are involved.