In the last two decades,motor operation monitoring tools have become a necessity,and many studies focus on the detection and diagnosis of motor electrical faults.However,at present,a core obstacle that prevents the di...In the last two decades,motor operation monitoring tools have become a necessity,and many studies focus on the detection and diagnosis of motor electrical faults.However,at present,a core obstacle that prevents the direct comparison of such classification techniques is the lack of a standard database that can be used as a benchmark.In view of this,we offer here a public experimental data-set that has beendesigned specifically for the comparison of synchronous motor electrical fault classifiers.The data-set comprises five types of motor electrical faults:open phase between inverter and motor;short circuit/leakage current between two phases;short circuit/leakage current in phase-to-neutral;rotor excitation voltage disconnection;and variation of rotor excitation current.In addition,each fault has been recorded as a four-dimensional signal:three phase voltages;three phase currents;motor speed;and motor current.The package includes two deep-learning reference classifiers that are based on a convolutional neural network(CNN)and long short term memory(LSTM).Due to the good performance of these classifiers,we suggest that they can be used by the community as benchmarks for the development of new and better motor electrical fault classification algorithms.The database and the reference classifiers are examined and insights regarding different combinations of features and lengths of recording points are provided.The developed code is available online,and is free to use.展开更多
Shaping and controlling electromagnetic fields at the nanoscale is vital for advancing efficient and compact devices used in optical communications,sensing and metrology,as well as for the exploration of fundamental p...Shaping and controlling electromagnetic fields at the nanoscale is vital for advancing efficient and compact devices used in optical communications,sensing and metrology,as well as for the exploration of fundamental properties of light-matter interaction and optical nonlinearity.Real-time feedback for active control over light can provide a significant advantage in these endeavors,compensating for ever-changing experimental conditions and inherent or accumulated device flaws.Scanning nearfield microscopy,being slow in essence,cannot provide such a real-time feedback that was thus far possible only by scattering-based microscopy.Here,we present active control over nanophotonic near-fields with direct feedback facilitated by real-time near-field imaging.We use far-field wavefront shaping to control nanophotonic patterns in surface waves,demonstrating translation and splitting of near-field focal spots at nanometer-scale precision,active toggling of different near-field angular momenta and correction of patterns damaged by structural defects using feedback enabled by the real-time operation.The ability to simultaneously shape and observe nanophotonic fields can significantly impact various applications such as nanoscale optical manipulation,optical addressing of integrated quantum emitters and near-field adaptive optics.展开更多
This paper presents a non-isolated DC-DC boost converter using voltage lift techniques. The proposed structure offers continuous input current. Substantial voltage gain along-side low voltage stress across semiconduct...This paper presents a non-isolated DC-DC boost converter using voltage lift techniques. The proposed structure offers continuous input current. Substantial voltage gain along-side low voltage stress across semiconductors in this topology contribute to utilizing an MOSFET with lower RDS-ON and devices with low nominal voltage values. Consisting of a switch makes the control procedure of the converter uncomplicated. Voltage analysis for the converter has been done in continuous and discontinuous conduction modes (CCM)-(DCM). Furthermore, current calculation, design process and comparison study are provided. Dynamic performance of the proposed circuit is scrutinized by applying the state space average technique and small signal model. In order to accredit performance of the converter, an archetype with 26V input and 222V output voltages and approximately 196W power level at 50 kHz switching frequency has been built and tested.展开更多
Despite widespread adoption and outstanding performance, machine learning models are considered as ‘‘blackboxes’’, since it is very difficult to understand how such models operate in practice. Therefore, in the po...Despite widespread adoption and outstanding performance, machine learning models are considered as ‘‘blackboxes’’, since it is very difficult to understand how such models operate in practice. Therefore, in the powersystems field, which requires a high level of accountability, it is hard for experts to trust and justify decisionsand recommendations made by these models. Meanwhile, in the last couple of years, Explainable ArtificialIntelligence (XAI) techniques have been developed to improve the explainability of machine learning models,such that their output can be better understood. In this light, it is the purpose of this paper to highlight thepotential of using XAI for power system applications. We first present the common challenges of using XAI insuch applications and then review and analyze the recent works on this topic, and the on-going trends in theresearch community. We hope that this paper will trigger fruitful discussions and encourage further researchon this important emerging topic.展开更多
Advanced machine learning(ML)algorithms have outperformed traditional approaches in various forecasting applications,especially electricity price forecasting(EPF).However,the prediction accuracy of ML reduces substant...Advanced machine learning(ML)algorithms have outperformed traditional approaches in various forecasting applications,especially electricity price forecasting(EPF).However,the prediction accuracy of ML reduces substantially if the input data is not similar to the ones seen by the model during training.This is often observed in EPF problems when market dynamics change owing to a rise in fuel prices,an increase in renewable penetration,a change in operational policies,etc.While the dip in model accuracy for unseen data is a cause for concern,what is more,challenging is not knowing when the ML model would respond in such a manner.Such uncertainty makes the power market participants,like bidding agents and retailers,vulnerable to substantial financial loss caused by the prediction errors of EPF models.Therefore,it becomes essential to identify whether or not the model prediction at a given instance is trustworthy.In this light,this paper proposes a trust algorithm for EPF users based on explainable artificial intelligence techniques.The suggested algorithm generates trust scores that reflect the model’s prediction quality for each new input.These scores are formulated in two stages:in the first stage,the coarse version of the score is formed using correlations of local and global explanations,and in the second stage,the score is fine-tuned further by the Shapley additive explanations values of different features.Such score-based explanations are more straightforward than feature-based visual explanations for EPF users like asset managers and traders.A dataset from Italy’s and ERCOT’s electricity market validates the efficacy of the proposed algorithm.Results show that the algorithm has more than 85%accuracy in identifying good predictions when the data distribution is similar to the training dataset.In the case of distribution shift,the algorithm shows the same accuracy level in identifying bad predictions.展开更多
基金This work was supported by the Natural Science Foundation of Jilin Province,China(20210101390JC).
文摘In the last two decades,motor operation monitoring tools have become a necessity,and many studies focus on the detection and diagnosis of motor electrical faults.However,at present,a core obstacle that prevents the direct comparison of such classification techniques is the lack of a standard database that can be used as a benchmark.In view of this,we offer here a public experimental data-set that has beendesigned specifically for the comparison of synchronous motor electrical fault classifiers.The data-set comprises five types of motor electrical faults:open phase between inverter and motor;short circuit/leakage current between two phases;short circuit/leakage current in phase-to-neutral;rotor excitation voltage disconnection;and variation of rotor excitation current.In addition,each fault has been recorded as a four-dimensional signal:three phase voltages;three phase currents;motor speed;and motor current.The package includes two deep-learning reference classifiers that are based on a convolutional neural network(CNN)and long short term memory(LSTM).Due to the good performance of these classifiers,we suggest that they can be used by the community as benchmarks for the development of new and better motor electrical fault classification algorithms.The database and the reference classifiers are examined and insights regarding different combinations of features and lengths of recording points are provided.The developed code is available online,and is free to use.
基金supported by the Russel Berrie Nanotechnology Institute(RBNI)at the Technion,and the Israel Ministry of Innovation,Science and Technology,grant number 2033419provided in sample fabrication by the photovoltaic laboratory and the Micro-Nano Fabrication unit(MNFU)at the Technion.S.T.acknowledges generous support from the Adams fellowship of the Israeli Academy of Science and Humanities+4 种基金the Yad Hanadiv foundation through the Rothschild fellowshipthe VATAT-Quantum fellowship by the Israel Council for Higher Educationthe Helen Diller Quantum Center postdoctoral fellowshipthe Viterbi fellowship of the Technion-Israel Institute of Technologysupport by the Israeli Council for Higher Education scholarship programme.
文摘Shaping and controlling electromagnetic fields at the nanoscale is vital for advancing efficient and compact devices used in optical communications,sensing and metrology,as well as for the exploration of fundamental properties of light-matter interaction and optical nonlinearity.Real-time feedback for active control over light can provide a significant advantage in these endeavors,compensating for ever-changing experimental conditions and inherent or accumulated device flaws.Scanning nearfield microscopy,being slow in essence,cannot provide such a real-time feedback that was thus far possible only by scattering-based microscopy.Here,we present active control over nanophotonic near-fields with direct feedback facilitated by real-time near-field imaging.We use far-field wavefront shaping to control nanophotonic patterns in surface waves,demonstrating translation and splitting of near-field focal spots at nanometer-scale precision,active toggling of different near-field angular momenta and correction of patterns damaged by structural defects using feedback enabled by the real-time operation.The ability to simultaneously shape and observe nanophotonic fields can significantly impact various applications such as nanoscale optical manipulation,optical addressing of integrated quantum emitters and near-field adaptive optics.
文摘This paper presents a non-isolated DC-DC boost converter using voltage lift techniques. The proposed structure offers continuous input current. Substantial voltage gain along-side low voltage stress across semiconductors in this topology contribute to utilizing an MOSFET with lower RDS-ON and devices with low nominal voltage values. Consisting of a switch makes the control procedure of the converter uncomplicated. Voltage analysis for the converter has been done in continuous and discontinuous conduction modes (CCM)-(DCM). Furthermore, current calculation, design process and comparison study are provided. Dynamic performance of the proposed circuit is scrutinized by applying the state space average technique and small signal model. In order to accredit performance of the converter, an archetype with 26V input and 222V output voltages and approximately 196W power level at 50 kHz switching frequency has been built and tested.
基金supported by Israel Science Foundation,grant No.1227/18.
文摘Despite widespread adoption and outstanding performance, machine learning models are considered as ‘‘blackboxes’’, since it is very difficult to understand how such models operate in practice. Therefore, in the powersystems field, which requires a high level of accountability, it is hard for experts to trust and justify decisionsand recommendations made by these models. Meanwhile, in the last couple of years, Explainable ArtificialIntelligence (XAI) techniques have been developed to improve the explainability of machine learning models,such that their output can be better understood. In this light, it is the purpose of this paper to highlight thepotential of using XAI for power system applications. We first present the common challenges of using XAI insuch applications and then review and analyze the recent works on this topic, and the on-going trends in theresearch community. We hope that this paper will trigger fruitful discussions and encourage further researchon this important emerging topic.
文摘Advanced machine learning(ML)algorithms have outperformed traditional approaches in various forecasting applications,especially electricity price forecasting(EPF).However,the prediction accuracy of ML reduces substantially if the input data is not similar to the ones seen by the model during training.This is often observed in EPF problems when market dynamics change owing to a rise in fuel prices,an increase in renewable penetration,a change in operational policies,etc.While the dip in model accuracy for unseen data is a cause for concern,what is more,challenging is not knowing when the ML model would respond in such a manner.Such uncertainty makes the power market participants,like bidding agents and retailers,vulnerable to substantial financial loss caused by the prediction errors of EPF models.Therefore,it becomes essential to identify whether or not the model prediction at a given instance is trustworthy.In this light,this paper proposes a trust algorithm for EPF users based on explainable artificial intelligence techniques.The suggested algorithm generates trust scores that reflect the model’s prediction quality for each new input.These scores are formulated in two stages:in the first stage,the coarse version of the score is formed using correlations of local and global explanations,and in the second stage,the score is fine-tuned further by the Shapley additive explanations values of different features.Such score-based explanations are more straightforward than feature-based visual explanations for EPF users like asset managers and traders.A dataset from Italy’s and ERCOT’s electricity market validates the efficacy of the proposed algorithm.Results show that the algorithm has more than 85%accuracy in identifying good predictions when the data distribution is similar to the training dataset.In the case of distribution shift,the algorithm shows the same accuracy level in identifying bad predictions.
