针对传统小波包风电波动平抑方法存在的过度平抑问题,提出一种基于时频分析的改进小波包平抑方法.首先,找出功率最大变化范围超过波动标准的时刻对集合,利用小波包对风电功率进行小波分解得到各频段分量;然后,从高频到低频依次对各频段分量在时刻对集合上进行局部削峰填谷,直到风电功率满足波动标准,从而得到平抑后的风电并网功率和储能功率指令;最后,对波动平抑效果进行评价.算例仿真表明该方法在满足风电波动标准的基础上,减小了储能功率指令的累积功率和充放电切换次数,有利于降低储能配置成本、延长储能寿命.

Based on the final assessment in 2018 of The Air Pollution Prevention and Control Action Plan, the paper has made an analysis on the pattern of policy topic changes in view of providing a decision-making basis for policy promulgation in next stage. Using 128 national air pollution control policies in 2005-2018 as samples, the text keywords correlation network is constructed, the change path of policy topics in different historical stages obtained by subgroup recognition and topic similarity calculation, and the pattern of policy topic changes and future trends analyzed. It is found that the patterns of topic evolution and integration appear in the adjacent stage of topic change paths, and no same topic is found; policy topics are changed from general to specific; the measures from single to compound;and the role of technological innovation is constantly strengthened. The key topics of air pollution control in the late 13th Five-Year Plan and even longer period will be the ＂joint control of interregional defense in areas and periods of heavy pollution＂, ＂further upgrading of energy and other key industries＂ and＂R&D and application of energy saving and emission reduction technology and equipment＂.

针对目前预测居民消费价格指数(CPI)所依赖的数据源单一,并且大多与CPI无本质关联的问题,提出了基于细分行业用电数据预测CPI的方法.该方法认为电力数据在一定程度上体现了社会生产的状况,因此利用电力数据预测CPI理论上能够提升模型预测精度.首先对所使用的时间序列数据进行了平稳性调整,将非平稳数据调整为平稳的数据.然后采用Pearson时延系数、带时延的KL-Divergence对不同行业用电数据对CPI的影响是否存在不同时延进行判断,其次用赤池信息量对不同行业时延进行最终选择.最后利用岭回归模型构建CPI预测模型,即利用计算而得的不同行业的最优延时,并采用岭回归限制模型维度,预测了未来的CPI.用某省真实的细分行业用电数据和CPI数据对模型效果进行了验证,相比目前最好的自回归积分移动平均(ARIMA)模型,该模型预测误差率下降50.8％,模型稳定性提升62.3％.通过挖掘用电量数据与CPI之间更细粒度的关系,该方法成功提高了预测结果的精确度和稳定性,能够更好地预测CPI,也进一步说明了用电信息是预测经济指标的一个重要依据.

Aiming at the heavy and over load issue in operation of distribution network,a heavy overload forecasting model to achieve short-term forecast of heavy and overload events was established.Using the actual data of the business system,the proposed method was validated.The results show that the proposed method can describe the heavy and overload events more systematically and comprehensively,and the heavy overload forecasting model based on association rules performs well in the hit rate and accuracy rate.This method provides a new technical means that has certain practical value to enhance the distribution network management which could be seen as important experience of attempts in grid big data as well.

随着智能电网的发展,配电系统部署了众多的监控和管理系统,每时每刻都在产生大量的数据;如何应用这些数据,提高配电系统的智能化水平,既是机遇也是挑战.文章首先分析配电系统大数据的特点;其次总结配电系统大数据的关键应用场景和应用价值;然后综述国内外在配用电大数据领域的重点研究成果;最后分析大数据发展面临的挑战,并提出应对措施.

设备重过载运行是引起故障停电的重要因素之一,开展重过载事预警与风险评估,对于及时消除故障隐患、指导设备升级改造具有重要作用.作为大数据应用研究的一个切入点,中国电力科学研究院与国网山东省电力公司合作研发了台区重过载预警与风险评估系统.文章从系统总体架构、系统功能以及系统实现3个方面详细介绍了系统的设计和实现方法.该系统具有历史重过载多维分析、重过载关联因素分析、台区重过载预警和风险评估4个功能,同时采用城市电力地图的理念,依托电网GIS和多种可视化工具,实现对重过载台区的精准定位,并通过对多维度分析结果的展示,探寻重过载的原因、规律和影响的用户,为日常运维检修和重要时期保供电提供支撑.

异常值检测是数据挖掘的一个重要分支.在电力行业,异常值检测可用于电网故障检测、设备故障检测、用电异常检测等领域.文章根据电力大数据的特点,研究适用于电力大数据的异常值检测算法.针对快速密度峰值聚类算法用于异常值检测时未考虑数据的局部特点以及局部密度依赖于截断距离选取的不足,利用KNN(K-Nearest Neighbors)思想重新定义局部密度和距离,提出了基于KNN的快速密度峰值异常值检测算法,从而实现更加准确的异常值检测,并基于某省配电变压器的日负荷数据异常检测仿真实验证明了该算法的有效性.

随着新能源规模的快速扩张,社会需求的不断演变,传统电力系统的发展已跟不上时代的步伐.同时,中国新一轮的电力体制改革已经推出,这将给电力系统的安全性带来新的挑战.在此背景下,国家“互联网+”行动计划应运而生.文章在介绍互联网思维和技术及其对第三产业产生的影响的基础上,进一步解读“互联网+”智慧能源的内容,分析其可能产生的影响;综述了国内外能源互联网研究和实践成果;结合目前能源互联网的发展及应用案例分析,对未来的研究方向提出建议.

The comprehensive system for new energy integration displays the research and applying results in our country new energy fields.The system mainly introduces the realization of forecasting and controlling the power output of wind generation and photovoltaic generation.Through the united controlling function of energy storage gathered with wind power and photovoltaic power in Shanghai Caoxi energy storage base,finally analyzes the hardware of the system and software structure with which the comprehensive system for new energy integration is realized.

The statistical method for predicting photovoltaic generation power was studied in this paper.Based on the analysis of various meteorological factors,such as solar radiation,temperature and cloudiness,the BP neural network was adopted to establish the statistical model,which realized the power prediction of a photovoltaic station for the next 24 hours.The proposed method was then used to establish the power prediction systems of the main photovoltaic stations in Shanghai.The results suggested that this method had higher prediction precision,and could meet the requirements of the practical engineering application.

文章结合谐波分析法介损测量的特点和要求,从频域分析的角度对采样频率、采样位数等决定采样系统性能的主要参数进行了深入的分析,得出了采样参数对介损测量结果误差的影响规律.在此基础上,文章给出了确定最佳采样参数的原则和方法.这些结论为合理、有效地设计采样系统提供了理论依据.

Wind power is an important alternative energy. It is of great importance for energy supply and environmental protection to develop wind power in a large scale in China. This paper analyzes, via full and accurate data, the status of wind power development in China, and lists the total wind power installed capacity in provinces by the end of 2003. It described briefly the basic requirements and biding issues of wind power concession projects organized by the National Development and Reform Commission in 2003 and 2004, and introduces the background and principles of the Law of People's Republic of China for Renewable Energy. In addition, aiming at the problems emerged during the development of wind power in China, technical supports needing to be resolved urgently are discussed in aspects of measurement and evaluation of wind resources, micro-siting of wind farms, grid-connection of wind units, as well as certification and test of wind generation equipment and so on.

Lightning stroke is one of the important factors affecting the secure operation of wind farm. During the lightning stroke process the induced overvoltage may appear on the sensors, communication line and even the power line which are configured within the stander and barrel tower when lightning current passes through them, the normal operation of wind turbines will be affected by these transient overvoltage. In this paper the model of wind turbines is established and the transient overvoltage caused by lightning stroke is calculated and analyzed. The calculating results show that the maximum of the overvoltage between the barrel tower and the transmission line will appear at the up end and the lower end. It can reduce the capacitive coupling overvoltage by setting metal shielding layer. Moreover, the shielding layer can flat the voltage fluctuation.

Based on the full and accurate statistical information, the status and existing technical problems of wind power in China are analyzed in terms of yearly installed capacity, single unit capacity, regional distribution,investment as well as the market shares and the localization of wind power units etc.The paper believes that wind power industry in China will grow by leaps and bounds in near future.

雷击是影响风力发电场运行安全的主要因素之一.在雷击过程中,雷电流沿塔筒注入风电场接地网,将会引起接地网的地电位升高.这些暂态过电压将对风力发电机组及其输变电设备的正常运行造成影响.通过建立相关的风电场接地网模型,对雷击暂态过电压进行了计算分析.

Along with the enlargement of capacity of wind turbine and the scale of wind farm, the attention is paid to the secure operation of wind power increasingly. In the factors influencing the secure operation of wind farm, the lightning stroke is an important one. According to the research achievements of lightning protection for wind turbine generators, the lightning stroke process and damage mechanism of wind turbine generators and its lightning protection technology are discribed comprehensively.

Dielectric loss measurement is an important index in assessing the insulation performance of power equipment. With the development of condition-based maintenance, the on-line diagnostic technologies for dielectric loss measurement are regarded more and more. Many methods have been developed in this field in which harmonic analysis is prevalent due to its perfect performance of resisting disturbance and stability relative to others Harmonic analysis includes measuring the operating voltage and the current of the specimen by some sensors firstly, then transforming the analog signal into digital signal. Finally the dielectric loss angle could be calculated by means of digital frequency analysis. According to the digital spectrum analysis theory, accurate analysis can be conducted if the sampling length is the integral multiple of the signal cycle. Actually, as the frequency of power system is often fluctuant, the condition of the sampling of integral multiple is difficult to satisfy. Therefore the frequency fluctuation is an important factor which greatly influences the precision of dielectric loss detection based on harmonic analysis. Discrete Fourier transform (DFT) is the most fundamental method in digital spectrum analysis but is too difficult to use conveniently. So a new method, quadrature filtering algorithm, is used in this paper. Then, two preprocessing methods to improve the precision of dielectric loss detection are put forward

提出了基于神经网络的特征参数选择方法,并利用该方法对所给出的表征局部放电脉冲特性的特征参数进行了选择.模式识别的结果表明,所提出的特征参数的选择方法不仅是有效的,而且是必要的.使用该方法时,对样本进行预处理是必须的,因它直接影响最终的结果,甚至会使该方法失效.本文给出了针对局部放电脉冲的去噪方法,并提出对样本数据进行处理时,宜综合应用两种归一化方式,局部归一化和全局归一化.

The anti-interference analysis situation in PD test of power equipments at home and abroad is summarized.Its trends is introduced.The present anti-interference measures are pre-sented.

In this paper, the effect of PD on the inner structure of samples is shown by scanning electron microscope and the experimental results are analyzed and interpreted. It is found that after discharge, different material has different damaging areas and different developing trend of aging. As different material can bear different intensity of discharge and short time discharge in a certain extent of intensity, even it is rather strong,it has little effect on the materials. So the aging effect of PD is a cumulative process over a long period mostly.