SEARCH · Engineering Papers
Results for “partial power processing”
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Traction Inverter Integrated On-Board DC Fast Charging through Partial Power Processing
This paper introduces an innovative on-board integrated DC charging approach through partial power processing (PPP) in a traction inverter system. The proposed system, incorporating a series-connected partial power dual-active-bridge (DAB) converter, efficiently regulates the DC link voltage to achieve an optimal bus voltage for traction operation across a wide speed and torque range. Leveraging partial and bidirectional power processing, the battery current during charging is precisely regulated without the need for external DC-DC charging equipment, ensuring seamless integration to a DC hub with different voltage levels. This advanced integration also endows the system with bidirectional power exchange capability to/from the external DC hub, i.e., vehicle-to-DC (V2DC).
Multiport Control with Partial Power Processing in Solid-State Transformer for PV, Storage, and Fast-Charging Electric Vehicle Integration
This article proposes a multiport control method to enable partial power processing (PPP) in a medium-voltage (MV) multiport solid-state transformer (SST). MV multiport SSTs are promising in integrating low-voltage DC sources or loads such as solar photovoltaic, energy storage, and electric vehicles into smart grids without bulky line-frequency transformers. Compared to voltage-source SST, current-source (CS) SST features single-stage isolated bidirectional AC/AC, AC/DC, or DC/DC conversion using an inductive DC link. For a multiport CS SST, it is revealed in this article that the PPP capability can be enabled through the proposed control without extra hardware, different from the case of voltage-source converters where special hardware architecture is required for the PPP. With the PPP, most power exchange between LV ports is processed by only a fraction of the entire conversion stage, leading to reduced DC-link current, volume, loss, and improved efficiency. The proposed multiport PPP control scheme is analyzed to verify the advantages across a wide voltage and power range against conventional full power processing (FPP) multiport control, using the soft-switching solid-state transformer (S4T) with reduced conduction loss as an example. Comparative experimental results based on a SiC three-port S4T prototype verify the effectiveness of the proposed PPP scheme against the FPP scheme under both steady state and dynamic conditions. Here, the DC-link current reduction is measured to be more than 36%. Significantly, the proposed multiport PPP control scheme is generic and applicable to any hard-switching or soft-switching CS SSTs without extra hardware.
PV-BESS DC-Series Integration for Regulated DC Systems
DC-series integration introduces a novel approach to seamlessly integrate a solar photovoltaic (PV) array and a battery energy storage (BES) in series. This system, referred to as the PV-integrated battery energy storage system—dc series (PVBESS-DCS), simplifies integration and enhances power density by leveraging the inherent voltage-source characteristics of batteries and adopting the concept of partial power processing. However, addressing voltage variations of a PV array and a BES under different operational conditions remains a challenge to apply the PVBESS-DCS to a regulated dc system, where the dc-bus voltage is fixed. This article proposes a PVBESS-DCS solution tailored for regulated dc systems. Within this framework, we present a controllable series voltage source compensating for the voltage differences between the PV and BES sources and the dc bus. Thus, the system can perform PV maximum power point tracking and curtailment control while charging and discharging BES and connected to a regulated dc bus. Additionally, we present a single-stage multiport partial power processing dc–dc converter designed for the proposed PVBESS-DCS. It utilizes a triple active bridge dc–dc converter topology. Finally, this article presents operating principles, control strategies, simulation results, and experimental findings of the proposed PVBESS-DCS for regulated dc systems.
HyPerPy (Hydrogen Extraction and Parabolic Trough Plant Performance Models using Python) [SWR-21-53]
The HyPerPy package consists of three scripts: 1) Model for Hydrogen Tracking in Parabolic Trough Power Plants, 2) Model for Receiver Performance, and 3) Model for Hydrogen Extraction Process. The power plant model (1) tracks hydrogen generation and transport within the circulating heat transfer fluid (HTF) of the power plant. This script is a transient, initial value simulation, in which the hydrogen concentration in the circulating HTF is 0 moles per cubic meter everywhere at time 0 seconds. Hydrogen concentration is calculated at discrete locations within the circulating HTF with 4-second time resolution. During each time step, the change in hydrogen concentration due to hydrogen generation and permeation is calculated for each location according to the local HTF temperature, vessel or piping properties, hydrogen concentration and partial pressure. This model predicts hourly hydrogen concentrations for typical operating days in the spring, summer, fall, and winter seasons. Data is used to create hourly mappings of hydrogen concentrations for a typical operating year. The receiver model (2) uses a 1-hour time step to estimate getter loading and annulus pressure. For each time step, the model uses HTF and ambient temperature data to estimate absorber tube, bellows, and getter temperatures for the time step. In addition, the model uses hourly HTF hydrogen concentrations that are generated by the power plant model (1), and annulus hydrogen pressure from the previous time step. With these data, the model calculates the moles of hydrogen permeating across the absorber tube and bellows during the time step. The net change in moles hydrogen is added or subtracted from the getter loading for the previous time step, and the hydrogen pressure is re-calculated based on getter loading and temperature. The model uses this algorithm to simulate hydrogen permeation and loading 24 hours per day, 365 days per year using seasonal temperature data. The model repeats these calculations for 25 years to create a mapping of receiver getter loading and annulus hydrogen pressure for four seasons of each year. The model for hydrogen extraction (3) estimates hydrogen extraction rates for a specific separation module configuration. The rate depends primarily on membrane area, vacuum pump performance, headspace gas flowrate to the membrane, and headspace gas hydrogen partial pressure. This model has two versions. The steady-state version predicts hydrogen extraction rates when the module is operating in separation mode. The dynamic version predicts hydrogen transfer through the membrane when the module is operating in sensor mode. The steady-state version is used with the plant model (1) to predict hydrogen partial pressures in the power plant when the extraction process is operating.
Reinforcement learning for adaptive maintenance policy optimization under imperfect knowledge of the system degradation model and partial observability of system states
Maintenance policy optimization usually is faced with challenges that arise from an imperfect knowledge of system degradation models and from the partial observability of system degradation states. Here, this paper proposes a reinforcement learning method to address these two challenges for a class of maintenance problems with Markov degradation processes. The reinforcement learning approach consists of a learning component and a planning component. Using sequentially collected observations, at each step of decision-making the learning component improves the knowledge of system degradation in terms of the probability distributions of the transition rates based on sequential Bayesian inference. Using the updated transition rates, at each step of decision-making the maintenance policy optimization problem is then formulated as a partially observable Markov decision problem, and the planning component computes the optimal maintenance policy that maximizes the expected cumulative reward. The proposed method is illustrated using a numerical example with repair and inspection maintenance actions. The result shows that as more observations are collected, the learning component progressively learns the true system degradation process, and the planning component adjusts the optimal maintenance policy accordingly as well, which leads to increased reward.
NMPC for Mode-Switching Operation of Reversible Solid Oxide Cell Systems
Solid oxide cells (SOCs) are a promising dual-mode technology that generates hydrogen through high-temperature water electrolysis and generates power through a fuel cell reaction that consumes hydrogen. Reversible operation of SOCs requires a transition between these two modes for hydrogen production setpoints as the demand and price of electricity fluctuate. Moreover, a well-functioning control system is important to avoid cell degradation during mode-switching operation. In this work, we apply nonlinear model predictive control (NMPC) to an SOC module and supporting equipment and compare NMPC performance to classical proportional integral (PI) control strategies, while ramping between the modes of hydrogen and power production. While both control methods provide similar performance in many metrics, NMPC significantly reduces cell thermal gradients and curvatures (mixed spatial-temporal partial derivatives) during mode switching. A dynamic process flowsheet of the reversible SOC system was developed in the open-source, equation-based IDAES modeling framework. Our IDAES dynamic simulation results show that NMPC can ramp the SOC system between hydrogen and power production targets within short mode-switching times. Moreover, NMPC can comply with operating limits in the SOC system more effectively than PI, and only NMPC can directly enforce user-specified limits for mixed spatial-temporal partial derivatives of temperature. This allows for management of the trade-off be-tween operating efficiency and cell degradation, which is dependent on these temperature curvatures.
Measuring Nd(III) Solution Concentration in the Presence of Interfering Er(III) and Cu(II) Ions: A Partial Least Squares Analysis of Ultraviolet–Visible Spectra
Optical spectroscopy is a powerful characterization tool with applications ranging from fundamental studies to real-time process monitoring. However, it can be difficult to apply to complex samples that contain interfering analytes which are common in processing streams. Multivariate (chemometric) analysis has been examined for providing selectivity and accuracy to the analysis of optical spectra and expanding its potential applications. Here we will discuss chemometric modeling with an in-depth comparison to more simplistic analysis approaches and outline how chemometric modeling works while exploring the limits on modeling accuracy. Understanding the limitations of the chemometric model can provide better analytical assessment regarding the accuracy and precision of the analytical result. This will be explored in the context of UV–Vis absorbance of neodymium (Nd 3+ ) in the presence of interferents, erbium (Er 3+ ) and copper (Cu 2+ ) under conditions simulating the liquid–liquid extraction approach used to recycle plutonium (Pu) and uranium (U) in used nuclear fuel worldwide. Finally, the selected chemometric model, partial least squares regression, accurately quantifies Nd 3+ with a low percentage error in the presence of interfering analytes and even under conditions that the training set does not describe.
Hydrogen Mitigation Process Testing at Nevada Solar One
The National Renewable Energy Laboratory (NREL) and Acciona Solar Power (ASP) developed and installed a process that addresses the issue of hydrogen buildup in Acciona's Nevada Solar One power plant. Our method selectively removes hydrogen from the expansion tanks of the power plant to control hydrogen levels in the circulating heat-transfer fluid (HTF). During previous work, we developed a sensor that measures hydrogen partial pressure in the expansion-tank headspace gas. We demonstrated that our sensor measures hydrogen levels over a wide range of partial pressure from 10 mbar down to 0.003 mbar. More recently, we conceived and developed an integrated process module that performs both hydrogen sensing and separating functions. The sensor/separator measures hydrogen partial pressure in the headspace gas in the same way as our original sensor design. Additionally, the integrated module separates hydrogen from the headspace gas to reduce hydrogen to the level needed to maintain the performance of receivers in the collector field. Laboratory testing at NREL showed that the sensor function had an accuracy of +/-7%, and the hydrogen extraction rate for separator function was consistent with our modeling predictions. The primary benefit of this module is its simple design, both in terms of function and incorporation into the HTF subsystem of the power plant. Most recently, NREL and ASP completed installation and initial testing of a mitigation process at ASP's Nevada Solar One power plant in Boulder City, Nevada. In this paper, we report on the completed installation, initial testing, and plans to bring the process to full automation, so that it can be operated unattended on a daily schedule.
Hydrogen Mitigation Process Installation at Nevada Solar One
The National Renewable Energy Laboratory (NREL) and Acciona Solar Power (ASP) have developed and are implementing a process that addresses the issue of hydrogen buildup in parabolic trough power plants. Our method selectively removes hydrogen from the expansion tanks of the power plant to control hydrogen levels in the circulating heat-transfer fluid (HTF). During previous work, we developed a sensor that measures hydrogen partial pressure in the expansion-tank headspace gas. We demonstrated that our sensor measures hydrogen levels over a wide range of partial pressure—from 1.33 mbar down to 0.003 mbar. More recently, we conceived and developed an integrated process module that performs both hydrogen sensing and separating functions. The sensor/separator measures hydrogen partial pressure in the headspace gas in the same way as our original sensor design. Additionally, the integrated module separates hydrogen from the headspace gas to reduce hydrogen to the level needed to maintain the performance of receivers in the collector field. We demonstrated the performance of a laboratory-scale version of this module. Testing showed that the module performed as expected: the accuracy of the sensing function was ±7%, and the hydrogen extraction rate for the separating mode was consistent with our modeling predictions. The primary benefit of this module is its simple design, both in terms of function and incorporation into the HTF subsystem of the power plant. Most recently, NREL and ASP planned, specified, and designed a mitigation process that is based on the integrated module. We are currently completing installation of this process at ASP's Nevada Solar One power plant in Boulder City, Nevada, USA. The mitigation process is being installed at ground level below the HTF expansion tanks, where it draws headspace gas from the tanks, removes hydrogen, and returns the treated gas back to the tanks. In this paper, we report progress on the installation and describe some of the many design details and challenges that we addressed during the past year. We will generate initial performance data from the Nevada Solar One mitigation process in early 2020.
Process Anomaly Detection for Sparsely Labeled Events in Nuclear Power Plants
An essential aspect of online monitoring, subtle anomaly detection increases the detection lead time for equipment failure and enables a nuclear power plant (NPP) to mitigate unexpected partial or full outages, resulting in significant cost saving to the plant. Once an anomaly is detected by plant staff, its cause and severity are investigated. Because the vast majority of anomalies require some level of investigation, including some that require time-consuming examination, before they are passed over to the engineering organization for further analysis, plants are often equipped with tools to assist the staff in performing anomaly detection. Those tools operate as a black box and are often based on statistical methods that establish sensor correlations using preconfigured mathematical models and flag correlation deviations as anomalies. Due to the number of anomalies detected at a given NPP on a daily basis, a significant number of flagged anomalies usually await examination for days or weeks. A primary cause of this backlog is that the methods used by the tools generate many false positives. Though this is usually attributed to oversensitive model settings due to very narrow normal operation bands, it can also be associated with the model development being inadequate for the process being monitored, or with missing model inputs that could have explained misclassified positives. The performance of anomaly detection tools impacts their plant acceptance and utilization, especially when the effort to address false positives generated by the tool depletes the value or cost saved by using that tool. Thus, means to advance anomaly detection performance have been investigated by the Department of Energy’s Light Water Reactor Sustainability program. Previous and ongoing efforts have targeted unsupervised machine-learning (ML) methods, which do not require the labeling of any data fed into the ML model. By contrast, in supervised anomaly detection methods, every data point is labeled as either a normal or abnormal process condition, and the model is trained to replicate the classification process. Supervised methods usually outperform unsupervised methods, due to the added value in differentiating normal from anomalous states of the monitored process. An NPP’s corrective action program requires it to track and document, via a dedicated report, the resolution of any issues that occur within the plant. Once created, each report is reviewed by a plant screening committee, and several classifications and decisions are made. Recently, a collaborating NPP developed an artificial intelligence and ML-based classifier to categorize a condition report (CR) into classes that can serve to label the data as normal or anomalous. Applying CRs as labels represents a semi-supervised use case. Semi-supervised ML assumes that labels exist for some data points (i.e., labeled anomalies, in this case) but not for the rest. In this effort, semi-supervised ML methods were used to fuse data from CRs with anomaly detection methods in order to test the hypothesis that partially labeled anomalies would improve the accuracy of the anomaly detection methods. Specifically, two methods were used. The first is the deep Semi-supervised Anomaly Detection (deep SAD) method, which can handle labels ranging from fully unsupervised to fully supervised cases. The second is a newly designed ML method developed specifically for this effort and referred to as the high-order feature (HOF)-based method. To evaluate these two methods in controlled environments, synthetic data generators were developed and used. The first datasets used a spring-mass-damper (SMD) system simulator commonly found in mechanical engineering references. This was used to create two use cases: a one- and a three-mass system. Anomalies were introduced by changing the spring and damper coefficients while the system was actuated by random forces. The second datasets used the commercial Dymola-Modelica software to build a simplified nuclear reactor model. Anomalies were added in the form of corrupted sensor readings and/or control commands. The deep SAD method was tested using the SMD system, while the HOF method was tested using both datasets. Application of the deep SAD semi-supervised ML method demonstrated that labels can generate increased confidence in detecting true anomalies. This helped increase the number of true positives and decrease the number of false negatives—something that would aid in addressing the backlog of possible anomalies. Application of the HOF method demonstrated that labels can aid in down selecting from a candidate set of features to a more optimal subset in order to better differentiate between normal and anomalous conditions.
Design and Optimization of a 200-kW Medium-Frequency Transformer for Medium-Voltage SiC PV Inverters
This paper presents a design and optimization methodology for a 200-kW medium-frequency transformer (MFT) based on low-loss magnetic core (FINEMET FT-3TL). The proposed optimal design methodology consists of predesign, preliminary design, and optimal design. In the preliminary design, the parallel-concentric winding structure is selected to increase the current carrying capability and reduce the leakage inductance. Based on the parallel-concentric winding concept, a novel cooing and insulation structure with 3-D printed bobbins are proposed. The optimal designed MFT prototype achieves a power density higher than 19.23 kW/L. The electrical insulation system is tested at 12 kV ac peak voltage. In addition, the partial discharge (PD) test is conducted at 7.5 kV ac peak voltage to ensure the PD-free design. The high-frequency bipolar pulsewidth modulation voltage insulation (PD) test is the first time applied in MFT design and test process. Finally, the transformer is applied in a dual-active-bridge (DAB) converter with 200 kW rated power. The peak efficiency of the DAB converter is 99.53%, and the efficiency at 200 kW is 98.85%. The peak efficiency of MFT is 99.844%, and the efficiency at 200 kW is 99.842%.
The genetics of aerotolerant growth in an alphaproteobacterium with a naturally reduced genome
Reduced genome bacteria are genetically simplified systems that facilitate biological study and industrial use. The free-living alphaproteobacterium Zymomonas mobilis has a naturally reduced genome containing fewer than 2,000 protein-coding genes. Despite its small genome, Z. mobilis thrives in diverse conditions including the presence or absence of atmospheric oxygen. However, insufficient characterization of essential and conditionally essential genes has limited broader adoption of Z. mobilis as a model alphaproteobacterium. Here, we use genome-scale CRISPRi-seq (clustered regularly interspaced short palindromic repeats interference sequencing) to systematically identify and characterize Z. mobilis genes that are conditionally essential for aerotolerant or anaerobic growth or are generally essential across both conditions. Comparative genomics revealed that the essentiality of most “generally essential” genes was shared between Z. mobilis and other Alphaproteobacteria, validating Z. mobilis as a reduced genome model. Among conditionally essential genes, we found that the DNA repair gene, recJ, was critical only for aerobic growth but reduced the mutation rate under both conditions. Further, we show that genes encoding the F1FO ATP synthase and Rhodobacter nitrogen fixation (Rnf) respiratory complex are required for the anaerobic growth of Z. mobilis. Combining CRISPRi partial knockdowns with metabolomics and membrane potential measurements, we determined that the ATP synthase generates membrane potential that is consumed by Rnf to power downstream processes. Rnf knockdown strains accumulated isoprenoid biosynthesis intermediates, suggesting a key role for Rnf in powering essential biosynthetic reactions. Our work establishes Z. mobilis as a streamlined model for alphaproteobacterial genetics, has broad implications in bacterial energy coupling, and informs Z. mobilis genome manipulation for optimized production of valuable isoprenoid-based bioproducts.
The Genetics of Aerotolerant Growth in Zymomonas mobilis
Reduced genome bacteria are genetically simplified systems that facilitate biological study and industrial use. The free-living Alphaproteobacterium, Zymomonas mobilis, has a naturally reduced genome containing fewer than 2000 protein coding genes. Despite its small genome, Z. mobilis thrives in diverse conditions including the presence or absence of atmospheric oxygen. However, insufficient characterization of essential and conditionally essential genes has limited broader adoption of Z. mobilis as a model Alphaproteobacterium. Here, we use genome-scale CRISPRi-seq to systematically identify and characterize Z. mobilis genes that are conditionally essential for aerotolerant or anaerobic growth, or are generally essential across both conditions. Comparative genomics revealed that the essentiality of most "generally essential" genes was shared between Z. mobilis and other Alphaproteobacteria, validating Z. mobilis as reduced genome model. Among conditionally essential genes, we found that the DNA repair gene, recJ, was critical only for aerobic growth but reduced the mutation rate under both conditions. Further, we show that genes encoding the F1FO ATP synthase and Rnf respiratory complex are required for anaerobic growth of Z. mobilis. Combining CRISPRi partial knockdowns with metabolomics and membrane potential measurements, we determined that the ATP synthase generates membrane potential that is consumed by Rnf to power downstream processes. Rnf knockdown strains accumulated isoprenoid biosynthesis intermediates, suggesting a key role for Rnf in powering essential biosynthetic reactions. Our work establishes Z. mobilis as a streamlined model for alphaproteobacterial genetics, has broad implications in bacterial energy coupling, and informs Z. mobilis genome manipulation for optimized production of valuable isoprenoid-based bioproducts
Transport Properties of Magnetized High-Energy Density Plasma (Final Project Report)
This report summarizes results of project DE-SC0016159 "Transport Properties of Magnetized High-Energy-Density Plasma,'' which ran from 7/15/2016 - 7/14/2021. The primary objectives of the work, as stated in the original proposal, were to develop a theory that describes transport coefficients in magnetized high energy density plasmas and to test the theory with molecular dynamics (MD) simulations. The science challenge is that strong ion coupling, strong magnetization of electrons, and partial degeneracy of electrons, cause the system to be in a regime that is not well described by current theory. The particular processes that were to be investigated include: ion stopping power, diffusion, electron-ion temperature relaxation, thermal conduction and viscosity. All of the primary research objectives were accomplished during the course of this work. In addition, some unexpected results were discovered along the way that led to new and productive research directions. This work resulted in 17 publications in well-respected peer-reviewed journals (primarily Physics of Plasmas and Physical Review E), and 2 more are being prepared for publication. A few of these were selected as Editor's Picks and one was published as a Rapid Communication. Highlights of the research results include: Transport phase space: The first identification of the parameter space that defines fundamental transport regimes in terms of the Coulomb coupling and magnetization parameters. Mean force kinetic theory: Systematic derivation of a kinetic theory for strongly coupled plasmas based on a new expansion parameter of the BBGKY hierarchy. This provided the derivation of a theory that we had previously posed phenomenologically, and also revealed a new term that captures the equation of state properties at all coupling strengths. Transverse friction force: Discovery that the friction force on a test charge in a strongly magnetized plasma includes a component that is perpendicular to the motion of the test charge in the plane defined by the velocity and magnetic field vectors. Gyrofriction force: Discovery that the friction force on a test charge in a plasma that is both strongly magnetized and strongly coupled includes a component of the force in the direction of the Lorentz force. Kinetic theory for warm dense matter: Extension of the mean force kinetic theory to include partial electron degeneracy, so that it applies to warm dense matter.
Hardware acceleration for HPS algorithms in two and three dimensions
We provide a flexible, open-source framework for hardware acceleration, namely massively-parallel execution on general-purpose graphics processing units (GPUs), applied to the hierarchical Poincaré–Steklov (HPS) family of algorithms for building fast direct solvers for linear elliptic partial differential equations. To take full advantage of the power of hardware acceleration, we propose two variants of HPS algorithms to improve performance on two- and three-dimensional problems. In the two-dimensional setting, we introduce a novel recomputation strategy that minimizes costly data transfers to and from the GPU; in three dimensions, we modify and extend the adaptive discretization technique of Geldermans and Gillman [1] to greatly reduce peak memory usage. We provide an open-source implementation of these methods written in JAX, a high-level accelerated linear algebra package, which allows for the first integration of a high-order fast direct solver with automatic differentiation tools. We conclude with extensive numerical examples showing our methods are fast and accurate on two- and three-dimensional problems.
Uniqueness of relaxation times determined by dielectric spectroscopy
Dielectric spectroscopy is extremely powerful to study molecular dynamics, because of the very broad frequency range. Often multiple processes superimpose resulting in spectra that expand over several orders of magnitude, with some of the contributions partially hidden. For illustration, we selected two examples, (i) normal mode of high molar mass polymers partially hidden by conductivity and polarization and (ii) contour length fluctuations partially hidden by reptation using the well-studied polyisoprene melts as example. The intuitive approach to describe experimental spectra and to extract relaxation times is the addition of two or more model functions. Here, we use the empirical Havriliak-Negami function to illustrate the ambiguity of the extracted relaxation time, despite an excellent agreement of the fit with experimental data. We show that there are an infinite number of solutions for which a perfect description of experimental data can be achieved. However, a simple mathematical relationship indicates uniqueness of the pairs of the relaxation strength and relaxation time. Sacrificing the absolute value of the relaxation time enables to find the temperature dependence of the parameters with a high accuracy. For the specific cases studied here, the time temperature superposition (TTS) is very useful to confirm the principle. However, the derivation is not based on a specific temperature dependence, hence, independent from the TTS. We compare new and traditional approaches and find the same trend for the temperature dependence. The important advantage of the new technology is the knowledge of the accuracy of the relaxation times. Relaxation times determined from data for which the peak is clearly visible are the same within the experimental accuracy for traditional and new technology. However, for data where a dominant process hides the peak, substantial deviations can be observed. Finally, we conclude that the new approach is particularly helpful for cases in which relaxation times need to be determined without having access to the associated peak position.
Early Alarm: Robust Event Analysis for Power Systems using 1-D Fully Convolutional Network
This work presents a novel deep learning model for early, accurate, and robust detection, recognition, and temporal localization of multi-type events in large-scale power systems. The proposed method develops a unified 1-D fully convolutional network (FCN) model that takes time series of raw frequency signals measured from a power system as input, extracts distinguishing features, and predicts at every temporal point in the time series if an event is happening and what the type of the event is. Compared to existing methods, the proposed model eliminates the necessity for hand-crafted feature extraction or complicated data pre-processing, can flexibly handle input signals of arbitrary length, and precisely infer the event occurrence time. Most importantly, the model is intentionally trained with incomplete patterns, such that it is more robust to partial features of an event which is common in real-world online recognition, resulting in early alarm for power system failures. Extensive experimental results demonstrate that the proposed method achieves superior performance to the state-of-the-art, and also shows strong robustness to noise and system oscillations.