Numerous benchmarks have been established to assess the performance of foundation models on open-ended question answering, which serves as a comprehensive test of a model's ability to understand and ...generate language in a manner similar to humans. Most of these works focus on proposing new datasets, however, we see two main issues within previous benchmarking pipelines, namely testing leakage and evaluation automation. In this paper, we propose a novel benchmarking framework, Language-Model-as-an-Examiner, where the LM serves as a knowledgeable examiner that formulates questions based on its knowledge and evaluates responses in a reference-free manner. Our framework allows for effortless extensibility as various LMs can be adopted as the examiner, and the questions can be constantly updated given more diverse trigger topics. For a more comprehensive and equitable evaluation, we devise three strategies: (1) We instruct the LM examiner to generate questions across a multitude of domains to probe for a broad acquisition, and raise follow-up questions to engage in a more in-depth assessment. (2) Upon evaluation, the examiner combines both scoring and ranking measurements, providing a reliable result as it aligns closely with human annotations. (3) We additionally propose a decentralized Peer-examination method to address the biases in a single examiner. Our data and benchmarking results are available at: http://lmexam.xlore.cn.
In order to meet the higher demand of dynamic reactive power support is put forward in the Ultra High Voltage Direct Current(UHV DC) power transmission project, the new-generation synchronous ...condensers are applied in the UHV DC-AC. However, in order to provide bidirectional transient reactive power characteristics during AC and DC grid fault condition, especially to reduce the commutation failure phenomenon at the inverter station, there are great changes in the main body and excitation control system of the new-generation synchronous condensers compared with the same-capacity generators. In this paper, the limiting model of excitation system for the new-generation synchronous condensers under low excitation and strong excitation conditions are established. Then, the expression of sub-transient reactive power for the new-generation synchronous condenser is derived and the reactive power characteristics considering the influence of excitation regulation are also obtained. In addition, the transient reactive power demand of the high-voltage direct current(HVDC) transmission system is also analyzed. Finally, the simulation model of the new-generation synchronous condensers access to UHV DC power grids are built based on the PSCAD software. The simulation results verify the support effect of the new-generation synchronous condensers on DC transient reactive power.
Analysis of Technical Losses in Hubei Power Grid Wang, Wenna; Cai, Defu; Liu, Haiguang ...
2021 IEEE 4th International Electrical and Energy Conference (CIEEC),
2021-May-28
Conference Proceeding
The power system loss is a very important technical and economic indicator in power enterprises. In a system, there are two types of losses: technical losses and non-technical losses (commercial ...losses). Out of the total system losses, technical losses can be reduced more directly and effectively. This paper focuses on the analysis of the variation and contributing factors of technical losses of each voltage class in Hubei power grid. As a result, main reasons and solutions for reducing technical losses are figured that provide a better understanding of weakness and loss composition of Hubei power grid, and are useful for power enterprises that intend to reduce electric losses as a benchmark and as a background in the planning purposes.
The low-temperature magnetotransport properties of manganite thin films are characterized by the occurrence of resistivity minima, rho sub(min), below 30 K whose origin and especially role of ...disorder has not yet been explored in detail. In order to contribute to the clarification of the physical mechanism giving rise to the resistivity minimum in these systems, an appropriate concentration (3%, 6%, and 20%) of nanoscaled nonmagnetic ZrO sub(2) particles are introduced as a secondary phase into La sub(2/3)Sr sub(1/3)MnO sub(3) thin films. As the volume density of ZrO sub(2) precipitates increases, the films show a more pronounced resistivity upturn for T < T sub(min). The measured temperature and magnetic field dependence of the resistivity of our samples is in good agreement with a combination of the theory of three-dimensional weak localization and electron-electron interactions. We show that within this frame the observed features of the scattering-related resistivity minimum at low temperature in correlated electron systems can be explained, including its spin dependence, its scattering parameters, and its variation with increasing nonmagnetic disorder.
The "two-tickets" system of electric power enterprises is an important organizational measure to ensure operation safety of electric power production site. In the actual electric power production, ...the operation safety accidents caused by "two tickets" signed not in accordance with regulations or by others occur from time to time. At present, electric power enterprises mainly rely on manual inspection to verify the authenticity of "two tickets" signature, which has the shortcomings of more human resources investment and low efficiency. In this paper, a signature identification method of "two-tickets" based on siamese convolutional neural networks is proposed. The proposed method introduces image processing, max-pooling, batch normalization, and dropout technologies, and effectively combines convolutional neural network and siamese network, which can effectively improve the accuracy of "two-tickets" signature identification. The results show that the proposed method is more accurate than k-nearest neighbor method, naive bayes method, decision tree method and support vector machine method. The proposed method can be applied to the "two-tickets" signature identification in electric power enterprises.
The intersection of physics-based vision and deep learning presents an exciting frontier for advancing computer vision technologies. By leveraging the principles of physics to inform and enhance deep ...learning models, we can develop more robust and accurate vision systems. Physics-based vision aims to invert the processes to recover scene properties such as shape, reflectance, light distribution, and medium properties from images. In recent years, deep learning has shown promising improvements for various vision tasks, and when combined with physics-based vision, these approaches can enhance the robustness and accuracy of vision systems. This technical report summarizes the outcomes of the Physics-Based Vision Meets Deep Learning (PBDL) 2024 challenge, held in CVPR 2024 workshop. The challenge consisted of eight tracks, focusing on Low-Light Enhancement and Detection as well as High Dynamic Range (HDR) Imaging. This report details the objectives, methodologies, and results of each track, highlighting the top-performing solutions and their innovative approaches.
Many power utilities have problems with the quality of data about distribution network topology. This affects the operation and maintenance of smart grid, including outage management and line loss. ...The operation data, such as voltage data whose similarity has been analyzed to verify distribution network connectivity. The most commonly method used to evaluate similarity is correlation analysis. In this paper, correlation analysis and morphology similarity distance (MSD) have been compared when used in distribution network connectivity verification. The results show that the data should be normalized when MSD is used. Four case studies have been carried out in order to compare the performance of the two methods. Results show that when the voltage loss of 10kV feeder is large, the performance of correlation analysis is better than MSD. When the variety range of the voltage curves is small, the performance of MSD is better than correlation analysis. When the voltage curves have catastrophe points, the performance of MSD is better than correlation analysis.
This paper discusses electric field under overhead lines with a nearby building. Surface charge method is used. Planer triangle surface charge element is chosen. Electric potential and electric field ...strength produced by a planer triangle surface charge are presented. And analytical expression is deduced. Then electric field of 220kV double circuit overhead lines with a nearby building is calculated. Results show that electric field above the building increases by its distortion effect and electric field nearby the building decreases due to its shielding effect.