Digital protective relays are widely used in power systems, including industrial and commercial power systems. These modern protective devices have demonstrated several performance advantages over electromagnetic ones in terms of accuracy, reliability, and response speed. Performance advantages of digital protective relays are always dependent on the resolution of data used as their inputs, along with algorithms for fault detection and identification. This paper tests the performance of time-based, frequency-based, and time-frequency-based digital protective relays for various sampling rates. Tested sampling rates include low, medium, and high sampling rates (ranging from 16 to 256 samples/cycle). In this paper, the three types of digital protective relays are tested when deployed as the digital 87T transformer digital protection. Performance testing of these digital protective relays is carried out in terms of the accuracy and response speed. Test results show that low sampling rates adversely impact the accuracy and response speed of time-based, frequency-based, and time-frequency-based digital protective relays. Obtained results also reveal that medium and high sampling rates help improve the accuracy and response speed of digital protective relays.
A fast SOC estimation method based on electrochemical impedance spectroscopy (EIS) dataset combined with a simplified electrochemical model is proposed in this paper. Firstly, the solution-phase lithium-ion diffusion model is stripped from the simplified electrochemical model to characterize the lithium-ion concentration change. Then, the key frequency feature for SOC estimation is extracted based on the correlation analysis between the solution-phase lithium-ion concentration and the EIS data. Finally, the proposed SOC estimation method is expanded to consider the influences of ambient temperature. Experimental study indicates that the integration of lithium-ion diffusion can significantly improve the accuracy and reduce the computational time compared with the traditional broadband domain method and the fractional order equivalent circuit model (FOECM).
The growing interests in utilizing Photovoltaic (PV) systems are usually faced with challenges of accurate and reliable protection of these distributed generation units. A desired protection of PV systems has to effectively and accurately detect and respond to internal faults (within the PV system) and external faults (in the host grid and/or fed loads). This paper presents and tests the performance of a digital ground fault protection (DGFP) for grid-connected PV systems. The presented DGFP detects faults based on the energy contents of high frequency sub-bands of the ground current. These energy contents are extracted using the phaselet transform that can process signals with non-stationary variations in their phases. Energy contents of the high frequency sub-bands can provide accurate, fast, and reliable detection and identification of faults experienced by a PV system operated. The phaselet-transform (PHT)-based DGFP is implemented for performance testing using PV systems that are operated in the grid-connected mode, and subjected to various fault and non-fault events. Performance results demonstrate accurate, fast, and reliable detection and response to different types of fault and non-fault events. Response features of the PHT-based DGFP are complimented with minor sensitivity to the level of power production and/or type and location of faults.
This paper describes what caused phase-phase arc fault in the cable compartment of a medium-voltage (MV) arc-resistant metal-clad switchgear without any evidence of external switching surge on the power system. Switchgear fault occurred on a sunny day to activate overcurrent protection relay to clear the fault instantaneously. After the arc fault incident, maintenance crew along with technical staff noted insulation damage was caused by signs of moist air and dust particles which led to insulation tracking to cause phase-phase arc fault. Air entry with dust particles and moisture was observed from the operation of arc-blast louver above the arc-exhaust chamber which has been opening and closing. Every time louver opens and closes some air will trap and go inside the cable compartment, which is against the basic design requirement of arc-resistant enclosure. Basic characteristics of arc-pressure increase and decrease (when arc-blast move through the arc exhaust chamber and louver) is included in the paper to understand important design requirements: ‘‘no outside air is allowed to go inside the arc-exhaust chamber.’’ Lessons learnt from this electrical incident, maloperation of arc-exhaust louver will require design modifications and proper coordination with the manufacturer of the arc-resistant chamber at arc-resistance switchgear. Papers ends with design recommendations for an arc-resistant switchgear application in inclement weathers and no outside air should get into the arc-exhaust chamber coming inwards from hinged flaps at the arc-resistance switchgear.
Accurate distributed photovoltaic power prediction plays a crucial role in supporting power retailers in making optimal trading strategies and distribution network operators in making reasonable scheduling plans. The mining and utilization of distributed photovoltaic power correlation is an effective way to improve the forecasting accuracy of the target station. However, the current photovoltaic power forecasting methods must rely on data sharing in correlation mining, which may lead to serious data privacy problems. In order to protect the privacy of data during power modeling while ensuring the effective utilization of spatiotemporal correlations, the spatiotemporal federated learning based distributed photovoltaic ultra-short-term power forecasting method is proposed in this paper. In this method, the local server takes the Gated Recurrent Unit-based autoencoder as local model to extract local temporal features and forecast the local power. The central server uses Federated Averaging to aggregate and update the parameters of each local forecasting model and uses t-distributed Stochastic Neighbor Embedding to process the temporal features into the global feature which implies the spatiotemporal correlations between stations. Subsequently, the local model combines local temporal features and global spatial features to achieve power prediction. Through the doublelayer sharing of local model parameters and local temporal features, spatiotemporal correlations are effectively utilized while ensuring the privacy of data and the forecasting accuracy of each distributed photovoltaic is improved. Finally, the case study via real-world data is conducted to verify the effectiveness of the proposed method.
The Soft Open Point (SOP) is an emerging power electronics device in distribution networks to replace normally open points (NOPs). During faulty conditions, service restoration can be effectively conducted by coordinating SOPs and distributed generation (DG) units. In this paper, a novel two-stage service restoration method is proposed for a distribution network by coordinating multiple SOPs and DGs. In Stage 1, a dynamic load-shedding scheme is applied during a fault at the upstream grid or distribution substation of a distribution network, and power supply to priority loads is kept uninterrupted as much as possible with DGs. Considering the ramp rate constraints of controllable/dispatchable DGs (CDGs) in Stage 1, their active power generation is kept at the same set points as they were before the fault. In Stage 2, CDGs are dispatched to maximize restoration in the outage area. In both stages, active and reactive power of SOPs are regulated to maximize restoration. A mixedinteger nonlinear programming (MINLP) model is developed using AC power flow to formulate the proposed restoration method. The centralized (coordinated) and decentralized (uncoordinated) optimization of SOPs and DGs are conducted and compared to validate the proposed restoration method using the modified IEEE 33-node test system.
Microgrids are building blocks of smart grids. With increasing penetration of renewable energy-based distributed generation (DG) in distribution networks, microgrid formation is an effective way to improve resiliency by transforming a conventional distribution network into its active form. In this paper, a new method for the microgrid planning through optimal microgrid formation in distribution networks is proposed through performance optimization, where the total power losses of the whole system are minimized, and their adequacy and reliability are improved. Brute Force search algorithm and Backward Forward Sweep method are used to solve the optimization and load flow problems, respectively. The IEEE 33-node test system is used to validate the proposed method in MATLAB.
Over the past years, more and more extreme weather events have been recorded across the globe. These events put thousands of people without electricity and cause utility companies lose millions of dollars. Therefore, power system resilience which focuses on high-impact, low-probability events become a crucial consideration for the system operator. This paper proposes a dayahead resilience oriented unit commitment that aimed to minimize the cost needed to mitigate the impact of typhoons. By using a weather-dependent fragility curve for the transmission components, the proposed framework can minimize system load shedding under typhoons. In order to address the high-impact nature of typhoon, robust optimization is used in our framework, which ensure the worst-case performance of unit commitment. The trilevel problem is reformulated to a bi-level robust problem using Column&Constraint Generation algorithm. To solve the intrinsic drawback of robust optimization in producing an over-conservative solution, the proposed model introduces Transportable Battery Energy System (TBES) to further improve the power system resilience under the proposed framework. RTS-24 system is used to demonstrate the effectiveness and efficiency of the proposed framework.
Solar power forecasting technology includes the procedure of data processing and the establishment of forecasting models. There is a high correlation exists between power generation and solar irradiance. Thus, if solar irradiance can be predicted accurately, the forecasting accuracy of solar power generation can be greatly improved. However, there is still room for improving the solar irradiance forecasting due to the limits of numerical weather prediction (NWP). Therefore, it is significant to improve other forecasting related technologies, such as data processing, to improve the power forecasting accuracy. According to literature reports, the classification of weather patterns is one of the important methods for data preprocessing. If different weather patterns can be classified, and the corresponding prediction model can be established in each weather classification, the forecasting accuracy can be improved effectively. This paper attempts to establish a variety of different methods to classify weather patterns; these methods include typical clustering classification, seasonbased classification, time-based classification (morning or afternoon), and the classification according to data’s amplitude and variance. The research results demonstrate that the forecasting methods that consider weather classification can improve the accuracy of day-ahead forecasts.
A full electrical system study for a large refinery with major expansion is reported. The expansion added two new gas turbine generators (GTG) as well as one new steam turbine generator (STG) to the system. A gas-insulated switchgear (GIS) lineup was installed at the 22 kV main substation with a new current limiting reactor to tie the main buses. Four of the substations are new as part of the expansion to increase the total number of substations to fourteen. Due to the new generators, a complete system study to size equipment, adjust protection settings, verify operating conditions and ensure system stability was carried out. The system study scope includes load flow, short circuit, protection and coordination, motor starting, and transient stability. The project lasted and covered the entire construction stages from For Review (FR) and For Design (FD) to For Construction (FC). The detailed system study helped the construction contractor and refinery owners to validate system design, confirm operation procedures and finalize protection schemes under both normal and abnormal conditions. Further, this paper shares experiences gained from performing and managing a complete system study project for an industrial facility in after front-end engineering design (FEED) phase in terms of validating system models, selecting study scenarios, leveraging computer software features, analyzing study results and proposing recommendations.
High Impedance differential protections (87Z) schemes for bus and transformer Restricted Earth Fault (REF) applications are common in the industry because of their security, dependability, and simplicity. 87Z schemes consist of a dedicated set of CTs, one per terminal, with all the secondary windings connected in parallel and terminated at a resistor, referred to as a stabilizing resistor, and MOVs. The interaction of these components results in voltage and current signals that deviate from typical power systems signals. In this paper, we present data from testing at a 2 MVA test facility and draw conclusions from the effect of these distortions on the signal processing of protective relays. We show that these effects are limited to a narrow range of fault currents. We also present signal processing and logic for the 87Z scheme that is secure and dependable.
External short circuit (ESC) testing is one of the many steps towards making lithium-ion batteries safe for different applications. The nature of testing is sensitive due to the concern of safety and data integrity. Great care and planning must be done before conducting these tests. Bouncing in recorded data is observed in literatures with high sampling rates due to switching and moving mechanical parts. This impacts the estimation accuracy of battery internal parameter values especially the effective ohmic resistance (Ro). This paper proposes a zero bouncing circuit design to eliminate this problem. Multi-rate Pulse Discharge Testing (MPDT) is conducted at safe current levels up to 7C at 25°C. To extract the values of internal battery parameters, Python is used, and the values are compared with those from an ESC test conducted using the same setup. In a bid to prevent more batteries from entering the waste stream, internal resistance of a post ESC battery (PESCB) is measured and compared to the internal resistance change during cyclic ageing of a healthy battery (HB).
Fault detection and failure mode diagnosis are of crucial importance in operation and maintenance (O&M) of photovoltaic (PV) power stations. In this work, advanced artificial intelligence techniques are exploited to optimize these O&M tasks for 150 PV power stations in Taiwan with total power rating around 54 MW. First, the response of each inverter under the maximal power tracking is monitored and analyzed by machine learning algorithms in every five minutes. The alert of fault detection will be activated if the power output of each inverter is significantly different from its nominal output. Prompt notification will be sent to user by mobile devices or emails immediately. To further enhance the performance of power prediction for multiple oriented roof-top PV systems, the power prediction model will be upgraded by simulated plane of array irradiance instead of direct measurements from only one pyranometer. Two-year field test results from 74 PV power stations with 4,792 inverters indeed demonstrate the effectiveness of the proposed AI-based O&M scheme for PV power stations.
With global warming, the need for a low-carbon transition of the power system is becoming more urgent. Previous DR potential forecasting models for load aggregators (LAs) only consider the peak-shaving potential of customers, ignore their potential to reduce carbon emissions, resulting in higher carbon emissions. To this end, A logistic sparrow search algorithm-back propagation neural network (LSSA-BPNN) based DR potential forecasting model for LAs in a low-carbon operation mode is proposed in this paper, which considers the dual incentive of electricity and carbon. First, customers are divided into different types according to their willingness of reducing economic cost and carbon emissions, and then the HEMS model considering the dual incentive of electricity and carbon is built. Second, the multiple influencing features are sorted according to the degree of importance by the RF model, which could reduce the redundancy of input features. Finally, based on the selected features, the LSSA-BPNN model is introduced to forecast the DR potential for LAs.
Accurate solar irradiance forecasting is the basis for accurate photovoltaic (PV) power forecasting. However, previous minutely irradiance forecasting methods based on allsky images have difficulty in adequately extracting the cloud features that are essential for future irradiance fluctuations. Therefore, in this paper, a minutely irradiance prediction method based on multidimensional feature extraction of all-sky images is proposed. The raw images are first pre-processed and classified into four weather types by cloud-sky identification. Then the cloud displacement vector is calculated using the optical flow (OF) method, and capture the subimages of the cloud domain that will cover the sun in the future dynamically according to the calculation results. Subsequently, we use the convolutional neural network (CNN) to extract multidimensional features including the local features and overall features. The multidimensional features are combined with relevant meteorological factors and historical irradiance to construct irradiance mapping models for each of the four weather types to achieve irradiance prediction on a ten-minute scale. Simulation results show that the proposed method can better introduce the key cloud information and improve the prediction accuracy.
This paper presents the wind farm generation effect on transient recovery voltage (TRV) values based on the location of the wind farm additionally this paper will examine the values of transient recovery voltage (TRV) above the capability of the breaker by applying the breaker transient recovery voltage capability curves within IEEE/IEC standards by determining the peak value (PV) and rate-of-rise of the transient recovery voltage waveforms (RRRV). The simulation of a grid-connected wind farm and a substation under study is performed using PSCAD. The studies show the effect of Wind farm location on TRV values, A substation breaker near the Wind farm can operate safely and the TRV values are within the breaker rating capability while as the substation breaker located far from the Wind farm has TRV values above the breakers rating capability, the breaker will also operate unsafely.
The primary energy sources in the world are fossil fuels like coal and oil. Overuse will result in issues like the greenhouse impact and energy depletion. Finding high-efficiency, clean, and renewable energy is a must in order to address the conflict between the ongoing rise in energy consumption and the deterioration of the natural environment. In order to optimize the Photovoltaic (PV) power in distribution system, this paper propose an efficient algorithm based on neural network. The main idea is to consider the main factors for evaluating the system performance through principle factor analysis technique and deploys them as input for the prediction. Then, the spatial search algorithm is used to select the optimal weight threshold of the neural network. Simulation results show that the proposed algorithm has better performance as compared with existing methods.
This paper addresses the provision of flexible ramping products by electric vehicle (EV) aggregators under renewable energy production and EV availability uncertainty. The problem is formulated using an adaptive robust optimization model to minimize system operational costs considering the unit commitment decisions, day-ahead, and real-time markets. The solution to this problem provides protection against EV availability, modeled using uncertainty sets while characterizing renewable energy production via scenarios. This problem is solved by Benders decomposition technique using primal cuts. The efficacy of the methodology identifying the robust scheduling of generators and EV aggregators is demonstrated through a realistic case study.
This paper discusses challenges in detecting, identifying, and responding to low voltage (<1000V) arcing current faults (ACFs), which can occur on the secondary side of 3$\phi$ medium voltage-to-low voltage power transformers. Secondary side ACFs trigger currents with magnitudes lower than those triggered by conventional faults, thus reducing the ability of medium voltage (MV) side protective devices to detect and respond to such faults. In many cases, the reduced ability to detect and respond to LV side ACFs prolongs the duration of these ACFs, and leads to a significant increase in the incident energy (may exceed acceptable limits). This paper presents an analysis of MV side currents to extract signature information to detect and identify LV side ACFs. The desired LV side ACF signature is extracted as high frequency components that have non-stationary phases. Such frequency components can be extracted using a multi-channel filter bank composed of digital high pass finite impulse response filters, which have linear phase responses. The non-stationary phase approach is tested several transient events including LV side ACFs. Performance results reveal accurate and reliable detection, identification, and response to LV side ACFs with negligible sensitivity to loading level and/or ACF type (series or parallel).
Due to the influence of many local random mutation factors, the ultra-short-term prediction of distributed photovoltaic power is faced with great challenges. This study is a modeling study for accurate prediction of photovoltaic power in hazy days based on all-sky image observation data. Firstly, based on the all-sky images, digital image processing technology is used to extract radiation-related image features of haze days, including image brightness, power spectrum, smoothness and mean of singular value. Secondly, the support vector regression model is established by combining image features and radiation attenuation coefficient. Finally, via determining the suitable forecasting time horizon, the surface irradiance can be forecasted based on the support vector regression model, and the photovoltaic power ultra-short-term prediction is realized using the irradiance-power convert model. The experimental results show that the ultra-short-term prediction method of photovoltaic power based on the whole sky image features and support vector regression has a good prediction effect when the optimal prediction time is determined, which provides an important practical basis for the accurate prediction of ultrashort-term power of distributed photovoltaic power stations.
