极端天气下基于时间卷积神经网络的新能源出力评估策略

新能源高渗透率背景下,极端天气造成的出力骤降对电力系统安全运行构成严重挑战,准确评估极端天气下的新能源出力深度是制定电力系统运行方式的重要依据。本文提出一种基于改进的深度时间卷积神经网络(deep temporal convolutional network,DeepTCN)的新能源出力评估方法。通过设计具有动态输入权重的时间卷积神经网络架构,准确量化估计了极端天气对新能源出力的影响,从而为制定极端天气下的电力系统运行方式提供风险程度信息。基于实际电力系统新能源出力数据集的算例结果表明,与常规的时序预测方法相比,所提方法的标准化均方根误差、对称平均绝对百分比误差和平均绝对标度误差3个指标分别最高可提升1.2,0.1和0.22。所提方法能够实现对极端天气下新能源出力的更精准评估。     

基于PIC16F946的CO气体检测仪

介绍了一种便携式CO气体检测仪的开发,论述了仪器的硬件原理图、一氧化碳气体传感器的参数模型分析以及软件的程序流程图,该仪器可实现检测空气中CO的含量,并可自动进行声、光以及震动报警,可通过按键设定二级报警值,并且可以对仪器进行标定,仪器以美国microchip公司的微控制器PIC16F946作为核心控制单元,分辨率为lppm;误差为士3%,采用3V的锂电池可以连续工作9000h以上,可广泛用于工业生产、环保、汽车、家庭等一氧化碳泄漏和不完全燃烧的检测报警.     

国家投资体制改革形势下的工程项目管理模式探索

上海市高级人民法院工程项目管理工作是运用建设、经济、技术、法律方面知识和经验实施管理的一项成功的工程咨询业务实例,同时也对国家投资体制改革形势下的工程咨询和项目管理模式进行了有益的探索。     

SNCR烟气脱硝系统多模型GPC-PID串级预测控制

针对SNCR烟气脱硝系统的大惯性、大迟延与时变等特性,提出了一种应用于不同典型工况下的GPC-PID串级预测控制方法.首先采用系统辨识的方法建立各典型工况下的SNCR脱硝系统的局部模型;其次依据二次最优控制原理设计各典型工况下的GPC-PID控制器参数;最后对各典型工况下的控制方法进行仿真研究,并研究各典型工况下GPC...     

Optical coherent dot-product chip for sophisticated deep learning regression

Optical implementations of neural networks (ONNs) herald the next-generation high-speed and energy-efficient deep learning computing by harnessing the technical...     

肉苁蓉低聚糖提取工艺的研究

以内蒙古阿拉善盟出产的肉苁蓉为实验材料,对肉苁蓉低聚糖的提取工艺进行了研究,通过L9(34)正交实验,确定最佳工艺参数为浸提液pH为11,时间为90min,温度60℃,料液比为1∶25,浸提次数为2次。      

Studying interconnections between two classes of two-stage fuzzy optimization problems

In this paper, we study two types of risk aversion two-stage fuzzy optimization problems. The first type is called two-stage fuzzy minimum risk problem (FMRP), while the second type is referred to as two-stage fuzzy value-at-risk problem (FVRP). In order to facilitate the solution of the two optimization problems, it is required to study the properties of FMRP and FVRP as well as their relationships. For this purpose, we first discuss the semicontinuity about the recourse function of two-stage FMRP. After that, we discuss the interconnections between optimal objective value of FMRP and that of FVRP, and the relationships between optimal solution of FMRP and that of FVRP. Using the obtained results, it would be possible to solve one two-stage optimization problem indirectly by solving its counterpart.     

基于面向时间过程的静态安全分析

针对传统静态安全分析基于单一时间断面、无法细致反映系统状态变化的不足,本文提出了一种面向时间过程的静态安全分析方法。首先介绍了面向过程方法的基本概念。充分利用丰富的电力系统测量和负荷预测数据信息,提取了描述每个时间断面的特征向量;结合电力系统负荷连续变化的特点,通过改进的K-means聚类分析技术,对负荷曲线进行了时间过程的划分。定义了提取时间子过程最大特征断面和最小特征断面的指标,并对特征断面的显著代表性进行了研究。在保证电网安全稳定的前提下,本文方法可一定程度上提高传输线路的利用率。最后,IEEE30节点算例表明面向时间过程静态安全分析方法的合理性和有效性。     

Spray-drying of alumina powder for APS: effect of slurry properties and drying conditions upon particle size and morphology of feedstock

In this paper, effects of variables concerning slurry properties and drying conditions on the particle size and morphology of feedstock were discussed. The results indicated that, the increase of powder loading increased D ₅₀ and volume content but reduced recovery rate slightly. Other variables, such as inlet temperature, showed a maximum on the desired property. The particle size of resulting feedstock increased with the rise of feeding rate of slurry, however, decreased with the increase of atomization pressure based on the premise that atomization pressure was enough to force slurry to form the liquid membrane. For preparation of spherical feedstock, optimized variables were as follows: 60 wt.% powder in slurry, 2 wt.% PVA in slurry, inlet temperature of 60°C, atomization pressure of 13 × 10³ KPa and feeding rate of 12 ml min^?1. The volume content of feedstocks with 10–45 μm was above 65% and for D ₅₀ it was 15·42 μm. In addition, according to the heat and mass transfer equation, the drying mechanism was established, which was well in agreement with the experimental results.     

基于微博重复发送的垃圾用户甄别

针对微博平台上的垃圾用户甄别问题,本文提出了基于微博重复发送行为的垃圾用户行为建模和甄别算法。在真实微博垃圾用户数据分析的基础上,本建模方法综合考虑了微博垃圾用户的行为信息、社交网络信息和文本信息,从不同的角度对垃圾用户进行了分析和建模。在真实数据集上的实验证明了方法的有效性,并且对模型中若干参数进行了优化,同时也分析了垃圾用户行为信息、社交网络信息和文本信息对模型的影响程度。