Intelligent Protection Schemes for Smart Grids
Loading...
Authors
Asgari Gashteroodkhani, Oveis
Issue Date
2020
Type
Dissertation
Language
Keywords
Data Mining , Microgrids , Power System Protection , Protective Relays , Signal Processing , Smart Grids
Alternative Title
Abstract
A smart grid integrates advanced sensing technologies, control methods, and integrated communications into the electricity grid. In smart grids, fast and accurate fault detection/location along power transmission and distribution networks can be achieved by deployment of modern technologies used for data recording and analysis combined with intelligent algorithms. Smart protection schemes will result in power system reliability improvement, quick restoration of the power service and reduction in outage time. In this dissertation, intelligent protection schemes for microgrid, distribution and transmission systems are developed by taking advantage of available modern technologies used for data recording, analytics and information extraction.A microgrid is considered as an important part of future smart grids. A microgrid is a low voltage distribution network with a set of loads, distributed generations (DGs), and local energy storage systems (ESS) working as an entity. A microgrid is generally capable of operating in islanded and grid-connected modes and can provide higher reliability, better power quality, and optimum operation. On the other hand, due to the bidirectional power flow in feeders and different fault current levels under different modes of operation, the protection of the microgrids is challenging. In this dissertation, two different schemes are proposed for microgrid protection. In the first scheme, an appropriate index based on the TT-matrix z-score vector is introduced for fault detection/classification. A threshold is used for fault detection and classification within a few cycles from the fault inception for both grid-connected and islanded modes. A threshold selection approach using the unscented transformation (UT) is used to accommodate low signal-to-noise ratios (SNRs). In the second scheme, a combined deep belief network (DBN) and TT-transform are used for fault detection and classification for microgrids. Several statistical metrics are derived from the processed data and used for training a DBN. In both schemes, test scenarios consisting of all shunt and high impedance faults are investigated in microgrids with radial and loop topologies for grid-connected and autonomous modes of operation. The results verify that the proposed schemes can reliably detect and classify faults. Also, the proposed methods are robust to changes in fault parameters, and performs well under highly noisy conditions. Hybrid transmission lines in medium and high voltage networks have become more widespread due to underground cables that are used to connect the off-shore wind farms to overhead lines. It is expected that the proliferation of these lines significantly increases in the future grid. However, fault detection/location will be more complicated in such hybrid lines because of different traveling wave velocities in overhead lines and underground cables as well as reflection and propagation of traveling waves in junction points. An effective hybrid fault location technique in hybrid lines is proposed. Time-time (TT)- and S-transforms are used to extract information from the transient voltage signals measured at a single end. This information is used to train a support vector machine (SVM) which determines whether the fault occurs in the first or second half of the underground cable or overhead line. After the identification, the fault is located using Bewley's lattice diagram and the transformed voltage signal. The effectiveness of the proposed method is confirmed under various parameters such as fault type, resistance, inception angle, and location.In distribution systems, High Impedance Faults (HIFs) have been challenging researchers and utilities for years since their current magnitudes are below traditional overcurrent relay pickups. Most of HIFs include igniting arc between a conductor and a ground surface that may cause fire and human safety issues. Therefore, in this dissertation, HIF detection is extensively evaluated in the real distribution feeders of Northern Nevada. Two different HIF detection methods based on non-harmonic and odd-harmonic content of fault currents that can be used in a substation-based protective relay are described and tested in a hardware-in-the-loop (HIL) platform using a real-time digital simulator (RTDS). HIFs in different ground surfaces such as wet sand, dry sod, dry grass, wet sod, wet grass, and reinforced concrete are modeled in RSCAD software. The practical usefulness of a utility-based relay capable of detecting HIF is evaluated. The limitations of the relay and the solutions for mitigating the fire and human safety hazard are discussed. Also, operating behaviors of different grounding systems are described and compared and the best possible approach is suggested for increasing the HIF detection rate and speed.Line impedances estimation is of great importance for protection engineers due to various applications such as protective relays setting and short-circuit analysis. Different software are used in utilities to calculate line impedances using the line conductor and structure data. The calculated values cannot be reliable due to inaccurate information or change in atmospheric condition. Also, since there is a lack of communication between different departments such as system protection, line construction, etc., a line conductor/structure may be changed while the affected departments are not notified. Most of the impedance estimation methods use phasor measurement units (PMUs) in the protection relays which are connected to protection class transducers that do not provide accurate currents/voltages under normal operating conditions. These methods not only require PMUs in the substations, but also fail to measure zero-sequence voltages and currents due to inadequate zero-sequence components under normal operation. In parallel transmission lines, zero-sequence impedance estimation is even more challenging. In this dissertation, a method is introduced for estimating the self and mutual zero-sequence impedances for mutually coupled transmission lines using recorded fault data. Extensive simulation and protective relay test results in RSCAD/RTDS indicate that the proposed approach has high accuracy in estimating the self and mutual zero-sequence impedances as well as the fault location in parallel transmission lines.