101 research outputs found
<b>10m winter wheat harvested area and planted area distribution map of China for five years (2018-2022)</b>
No full text------ 1. Introductory information ------Title: Mapping 10-m harvested area in the major winter wheat-producing regions of China from 2018 to 2022Format: TIFNaming convention: " ChinaWheatMap10_P_2018.tif" means winter wheat planted area map of 2018, and “ChinaWheatMap10_H_2018.tif" means winter wheat harvested area map of 2018.Authors: Jinkang Hu, Bing Zhang, Dailiang Peng, Jianxi Huang, Wenjuan Zhang, Bin Zhao, Enhui Cheng, Zihang Lou, Shengwei Liu, Songlin Yang, Yunlong Tan, and Yulong LvKey Laboratory of Digital Earth Science, Aerospace Information Research Institute, Chinese Academy of SciencesCorresponding author: Bing ZhangContact Information: [email protected] (JK.H.); [email protected] (B.Z.)------ 2. Data specific information ------The dataset contains winter wheat maps of harvested area and planted area with 10m spatial resolution for five years (2018-2022) over eight provinces. In the datasets, the values equal to one means winter wheat.Software: ArcGIS, QGIS or ENVI are needed to read the dataset.</p
Coordination of Legal Protection of Algorithms and Intellectual Property System
No full textIn the context of the intelligent revolution, the algorithm is increasingly becoming an important tool for assisting decision-making and regulating order. Because of the professionalism and opacity of the algorithm, a series of challenges of legal rules and legal order will occur if there is no market access mechanism and post-mortem supervision. Based on the analysis of the intellectual property protection of the algorithm and the essence of the intelligent society, this paper reveals that the algorithm is the endogenous power of the intelligent society. The intellectual property protection of the algorithm is in line with the value needs of the essence of the intelligent society, which is the necessary system for the rapid development of the intelligent society in the future. The existing algorithm protection methods include copyright, trade secrets, and patent rights. The current coverage is not wide enough, the protection effect is weak, and it is easy to trigger new social problems, which can hinder the protection of social benefits and the promotion of technological progress. The authors believe that the patent law “public change protection” mechanism can not only alleviate the contradiction between “algorithm power” and public interest but also stimulate the development of algorithm technology. An algorithm is a technical solution, and it is also a rule of thinking. The algorithm has the characteristics of technical solutions and thinking rules, which is different from pure thought rules and can produce “changes in the physical state”. Therefore, it should be protected as an object of the patent law. It is necessary to determine the patent-ability standard of the algorithm as soon as possible. The algorithm acts as a new type of object protected by the patent law directly, and at the same time, it sets the algorithm value evaluation mechanism. Finally, through the system construction of algorithm protection, the intellectual property law can be used to promote the innovation of algorithms, so that the algorithm can be developed in a more rational, ethical and legal direction to boost the rapid development of intelligent society
Refined Battery Energy Storage System Modelling for Grid Dispatch
No full textBattery energy storage systems (BESSs) can be charged and discharged rapidly, which makes them capable of many tasks in the grid operation, such as frequency regulation and load management. However, BESSs generally suffer from high cost, not only from the initial investment but more importantly from the cycle loss and calendar loss due to unavoidable battery degradation. If the dispatchers do not follow suitable BESS operation plans, it may cause economic losses. Therefore, the BESS degradation models are important research topics by considering the BESS economic loss in the optimization problems. The research topic of this graduation thesis is to maximize the benefits that the BESSs bring when participating in power grid work. The following points need to be addressed to complete the topic:
1. it is valuable to identify the main degradation factors of Li-ion batteries when the BESSs are used in the power grid dispatching problems. 2. accurate and appropriate modeling of BESSs with other devices in the simulations and calculations. 3. the BESSs can be used in various power system tasks such as regulations services, reserve, OPF and contingency recovery, etc. Therefore, it is crucial to be clear how BESSs can be used in various scenarios. 4. the Li-ion battery life assessment methods can be used as the reference for measuring the accuracy of the BESS models. Therefore, it is necessary to adopt a suitable life span assessment method for the BESSs.
The following work has been done in this thesis: 1. A C-rate-based BESS linearization model is established, which can consider the change of BESS control strategies and is suitable to be used in BESS high-power scenarios. The traditional BESS models are mainly based on the DoD and are designed for low-power working conditions. The proposed C-rate-based model makes up for the vacancy of traditional models in applicable scenarios. 2. A fatigue life cycle counting method is designed. This algorithm is used for BESS life evaluation. The proposed SCCM is better than the commonly used RCM for battery cycle life assessment. 3. The author carries out linearization processing according to the SCCM, establishing a SoC-DoD-based BESS model, which separately considers the fact that the charging and discharging of Li-ion batteries have different effects on cycle life. The optimization results show that the proposed model can obtain optimization results closer to the actual optimal situation
Practical Differentially Private and Byzantine-resilient Federated Learning
No full textPrivacy and Byzantine resilience are two indispensable requirements for a
federated learning (FL) system. Although there have been extensive studies on
privacy and Byzantine security in their own track, solutions that consider both
remain sparse. This is due to difficulties in reconciling privacy-preserving
and Byzantine-resilient algorithms.
In this work, we propose a solution to such a two-fold issue. We use our
version of differentially private stochastic gradient descent (DP-SGD)
algorithm to preserve privacy and then apply our Byzantine-resilient
algorithms. We note that while existing works follow this general approach, an
in-depth analysis on the interplay between DP and Byzantine resilience has been
ignored, leading to unsatisfactory performance. Specifically, for the random
noise introduced by DP, previous works strive to reduce its impact on the
Byzantine aggregation. In contrast, we leverage the random noise to construct
an aggregation that effectively rejects many existing Byzantine attacks.
We provide both theoretical proof and empirical experiments to show our
protocol is effective: retaining high accuracy while preserving the DP
guarantee and Byzantine resilience. Compared with the previous work, our
protocol 1) achieves significantly higher accuracy even in a high privacy
regime; 2) works well even when up to 90% of distributive workers are
Byzantine
Study of volcanic ash impact onto turbine blades in jet engines.
No full textGas turbines are of great importance in industry. In the turbine section within a jet engine, thermal barrier coatings (TBCs) are utilized to protect the metal turbine blades, thus improve the efficiency of engine. However, this coating is extremely vulnerable to attack by injected particulates. This ingested particulate is often referred to as "CMAS" (Calcia-Magnesia-Alumina-Silica). Among all the CMAS materials, Volcanic Ash (VA) is the most common type which aeroengines may encounter during the flight. This type of CMAS material would melt in the combustion section by ultra high temperature and then impact the turbine blades with relatively high speed. Some of the particles would then stick on and bond with the TBC, thus cause degradation of the protecting coating. In this way, the jet engine would be permanently damaged.
In the recent years, experiments have been done by different researchers to elaborate the effect of CMAS materials on TBCs. However, there is still a lack of knowledge in the bonding mechanism and physical adhesion between the CMAS particles (especially VA particles) and the substrate. A study of VA particle impingement is required in order to understand particle impingement, phase transition and heat transfer, bonding mechanism and splat morphology in detail.
In this research, experiments were carried at Cambridge University, by Prof. Trevor William Clyne, Dr. James Dean and Dr. Catalina Taltavull to reproduce the VA-substrate impingement in jet engine. A Vacuum Plasma Spray (VPS) system was utilized to create a high-temperature, high-velocity flow field. Different types of Volcanic Ashes (VAs) were introduced into the experimental set-up. Sticking rate, Scanning Electron microscope (SEM) micrographs of deposition morphology were examined and collected. Chapter 3 elaborates the details of this experiment set-up and data collected from the experiment. This experiment set-up is utilized by the author for building numerical models and the result of this experiment is used to validate the numerical models.
Three numerical models were built to perform a systematic study. Firstly, in Chapter 4, a Computational Fluid Dynamic (CFD) model was created to simulate the steady-state of the VPS flow field. The Discrete Particle Method (DPM) model was then utilized to simulate the injection of volcanic ash particles. After calculating the BI number, non-isothermal effects within the ceramic particles were simulated by introducing the heat transfer function by a user-defined function (UDF). This model gives the temperature gradient within and velocity of the in-flight VA particles at any time during the spray. It is shown that small particles (diameter ) would easily impact the turbine blades, but would remain unmelted due to the large grain size. It is concluded that VA particles with diameter of to are the most "dangerous" particles, because these particles have both relatively high possibility to be melted, and high possibility to impact thus adhere on the substrate.
Second of all in Chapter 5, systematic study of Yttria-stabilized Zirconia (YSZ) particle impingement and deposition on stainless steel in thermal spray process has been performed. A Coupled Eulerian and Lagrangian model was developed. This model contributes to simulate the process of semi-molten particle impact. By utilizing this model, both the large deformation of liquid part and the plastic deformation of the solid part could be extracted. One fully molten and two semi-molten(solid core with liquid shell and solid shell with liquid core) cases were studied. The results of the numerical model matches well with the experiment and analytical data. Interest parameters such as velocity, temperature, fraction of liquid part were varied. The contact area, splat morphology and local contact temperature were collected and studied. It is shown that, the larger the liquid fraction is, the larger the contact area would be. Moreover, effect of roughness of substrate is also studied. It is suggested that substrate roughness whose average asperity size is higher than the 1/10 of particle size is beneficial for adhesion.
Third of all in Chapter 6, in order to simulate the impact for high-viscosity glass-state ceramic particles, Smoothed Particle Hydrodynamic (SPH) model was built. For high/ultra viscosity cases, traditional CFD method and Finite Element Method (FEM) would be extremely slow. SPH model transfers the Eulerian equations into Lagrangian equations. By utilizing this method, computational resources could be saved, and high viscosity impact could be simulated. The SPH algorithm was coded and equations for heat transfer was introduced to simulate the solidification of liquid. Systematic study were performed by utilizing this model. Viscosity, contact angle, velocity of particles were varied. Contact area, splat morphology and solidification at the contact area were examined. It is shown that, large contact angle would result in large contact area. However, particles impingement with low viscosity and high contact angle could result in the break up of the particles. The similar phenomena could be seen in experiment - small particles have lower viscosity and are approaching the substrate with a large contact angle. Therefore, the deposition of these types of particles show an obvious evidence of break up and oblique impact
Accurate Rotational Speed Measurement for Determining the Mechanical Power and Efficiency of Electrical Machines
No full textAccurately measuring the mechanical power of electrical machines is essential for determining their efficiency. Conventionally, the machines' rotational speed is measured over a short time interval with a high measurement uncertainty. In this paper, we propose a rotational speed determination based on the angle covered within a longer time interval. This method is shown to achieve a higher accuracy for the determination of mechanical power and total efficiency with negligible systematic effect.This is an author-created, un-copyedited version of a Conference contribution published in 2021 24th International Conference on Electrical Machines and Systems (ICEMS), 2021, 10.23919/ICEMS52562.2021.9634265
Extreme Masking for Learning Instance and Distributed Visual Representations
No full textThe paper presents a scalable approach for learning spatially distributed
visual representations over individual tokens and a holistic instance
representation simultaneously. We use self-attention blocks to represent
spatially distributed tokens, followed by cross-attention blocks to aggregate
the holistic image instance. The core of the approach is the use of extremely
large token masking (75\%-90\%) as the data augmentation for supervision. Our
model, named ExtreMA, follows the plain BYOL approach where the instance
representation from the unmasked subset is trained to predict that from the
intact input. Instead of encouraging invariance across inputs, the model is
required to capture informative variations in an image. The paper makes three
contributions: 1) It presents random masking as a strong and computationally
efficient data augmentation for siamese representation learning. 2) With
multiple sampling per instance, extreme masking greatly speeds up learning and
improves performance with more data. 3) ExtreMA obtains stronger linear probing
performance than masked modeling methods, and better transfer performance than
prior contrastive models.Comment: Accepted in TML
A Linearized Analog Photonic Link Based on a Single z-Cut LiNbO3 Dual-Output Mach–Zehnder Modulator
No full text
- …
