Rajesh G. Kavasseri † ~

Relay misoperations in power systems, especially during stressed conditions can initiate a cascade potentially leading towards a blackout. A typical event that aggravates a cascade is a trip decision by distance relays to faults in zone 3. We examine the use of energy functions as a discriminant in such scenarios to supervise the action of distance relays, in particular, to distinguish between load encroachment and zone 3 faults. The scenarios are tested on two individual test beds, the 39 bus New-England Test System (NETS) and a portion of the Western Electricity Coordinating Council (WECC) system. NETS is used to analyze and assess the choice of energy functions for protection, whereas the WECC system is used for an actual blackout simulation in the South Western United States corridor. Analysis on these test systems depict on how energy functions can help in detecting and potentially avert costly relay misoperations.

This paper envisages the mapping of particular faults associated with their real time occurring events in a large scale power system by exploring the underlying rich information embedded in the Structure Preserving Energy Function (SPEF) components. A sliding window based prediction is employed to continuously monitor the change in the SPEF components in pre-fault, during-fault and post-fault stages. A Stacked Potential Energy (SPE) matrix is introduced and subjected to Principal Component Sensitivity (PCS) to discern the effect of events by analyzing the upper and lower dimensional sub-spaces. The effectiveness of the proposed sequence of classification can be observed from the results on a 4-machine, 2-area test system, where a coherency can be noticed between types of events and the corresponding mappings in the respective SPEFs. Highly correlated energy functions can be sought-after by pointing at different events which may trigger some specific components of interconnected power systems.

This paper presents a benchmark model for a hybrid residential energy system comprised of AC as well as DC loads, DG and, a storage unit. The benchmark is connected to the grid through a simplified model of a Solid State Transformer (SST), developed in MAT-LAB/Simulink. Distributed generation (DG) is established by modeling the PV using the solar irradiance information obtained by the SAM software. A battery is modeled to be present in the grid with State-of-Charge (SOC) battery controls governing its switching on and off times. Solar irradiance for various times of the year were collected from the SAM software and were used as in-feed for various simulation schemes to test the fidelity of the model, subject to the hybrid nature of loads. The results demonstrate the bi-directional power flow capability of the the simplified SST in the context of hybrid generation and uneven load demand, over a range of scenarios incorporating the variability in PV penetration and its effects on the distribution system.

This paper presents a solution to address verification-related challenges during the design and validation for motor control applications that call for joint simulation of the control algorithm and the motor. One method used previously involved playback of a prerecorded motor response in order to get around a lack of an HDL model for the motor, resulting in erroneous results. To circumvent this, we present a workflow approach using FPGA-in-the-Loop simulation (FIL), which configures an FPGA with the HDL design and uses it as a computation block for closed-loop simulation. The approach is illustrated to implement, test, and validate two control schemes: a vector based Field Oriented Control (FOC) and the scalar voltage by frequency (V/f) algorithms on a permanent magnet synchronous machine. The proposed approach is useful to provide validation before hardware implementation.

This paper presents a Principal Component Analysis (PCA)-based method for online characterization of outliers in synchrophasor measurements. To that end, a linearized framework is established to analyze dynamical response from a system under nominal and off-nominal (e.g. faulted) conditions, which are contained in the same window of synchrophasor data. Inspired by the singular value perturbation theory, a bound on the change in the norm of the Principal Component (PC) scores as a function of system state matrices is presented. It is shown that in presence of bad data outliers these bounds for higher dimensional PC scores will be significantly larger compared to lower-dimensions. The effect of the number of samples in the data window on the results of the analysis is established. Case studies on a simulated test system and on field data collected from a US utility are presented to support the analytical results. Finally, an online classifier for characterization of outliers is developed to illustrate the usefulness of the proposed framework for machine learningbased methods.

This paper presents a method for online bad data outlier detection in PMU measurements. High resemblance of bad data in the form of outliers with that of system disturbances can cause control centers to take false decisions. The proposed method utilizes artificial neural network (ANN) in association with principal component analysis (PCA)-based feature extraction technique in order to classify bad data outliers and outliers caused by disturbances like faults. The extracted features using PCA from the reduced dimensional representation of the data is given as input to ANN. Bayesian regularization back-propagation-based learning algorithm is used for training ANN. The test results demonstrate online classification/detection of bad data of different amplitudes injected in measurements, both before and after disturbance, in two example power systems.

This study proposes a particle filter (PF)‐based dual estimation method for tracking the dynamic states of a synchronous generator. The authors consider the situation where the field voltage measurements are not readily available. The PF is modified to treat the field voltage as an unknown input which is sequentially estimated along with the other dynamic states. To exploit the tracking results from the estimator, the authors consider the application of use the estimated field voltage to identify internal failures in the excitation subsystem. The proposed method is tested on a 10‐machine, 39‐bus system to assess selectivity between estimation for regular external disturbances and both physical failure, and modelling abnormality within excitation system. The presented studies show that the proposed method (i) provides reasonable tracking results for the dynamic states and the field voltage simultaneously, (ii) rapidly tracks minor excitation loss due to exciter internal failure while maintaining selectivity and (iii) is robust to measurement noise.

This paper addresses two problems of interest in the present time, namely, the characterization of price variations and the corresponding load response to them. The paper begins by defining Price Elasticity Matrices (PEMs) and shows how they can be used to model demands deviation from their scheduled levels due to price differentials. It then explains the reason as to why extending PEMs characterization is necessary. Afterward, we show how Normal distribution lacks accuracy in modeling the variation of market clearing prices from one day to the next. Based on empirical data from the Midcontinent Independent System Operator (MISO) area, we propose the use of Stable distribution and demonstrate, how such statistical model is very accurate to characterize electricity market price variations. The models mentioned above are used in a Monte Carlo (MC) simulation to find the probability that different Direct Load Control (DLC) levels would not be enough to maintain the peak demand below its reference value. Since MC simulations are so computationally intensive, we also implement Importance Sampling Monte Carlo (ISMC) to substantially reduce the computational burden without sacrificing much of the accuracy.

There is a worldwide trend towards the integration of renewable energy in the form of distributed generation, leading to the formation of microgrids. Connection of these sources introduces new issues in the operation and management of distribution systems. An important issue is that of islanding, where the microgrid remains energized locally while isolated from the main grid. It is important to detect this islanding event quickly and accurately in order to prevent possible damage to the DG and load that remains connected to the DG side after islanding. This paper explores the problem of timely islanding detection by machine learning techniques. Several cases of islanding and non-islanding are simulated on the IEEE-13 bus distribution system. Different types of DGs are connected to the system and disturbances are introduced. We consider the use of the voltage, frequency and their rate of changes at the Point of Common Coupling (PCC) as features for event detection using classifiers. These features are extracted from the simulation results and used to train and test several types of classifiers. It is shown that the random forest classifier detects the islanding with a high level of accuracy and within a reasonable amount of time after the occurrence of the disturbance.