课程文献中心 更多>>

关键词： 短期风电功率预测   概率预测   在线学习   非平稳性   机器学习  

摘要： 为解决风电出力的非平稳性所引起的预测模型参数时变问题,降低短期风电功率预测误差,提出了一种风电功率在线概率预测方法。通过构建分位数神经网络模型及动态更新参数,实现了预测准确性的提升;采用凸神经网络模型,实现了数值天气预报和风电功率间的非线性映射;引入动态局部遗憾度,将凸神经网络模型拓展为在线学习模型,并采用动态指数时间平滑随机梯度下降(dynamic exponentially time-smoothed stochastic gradient descent,DTS-SGD)算法实现参数更新。最后,结合2014年全球能源预测大赛发布的风力发电数据,验证了所提方法的预测准确性以及对超参数选择的鲁棒性,并验证了所提方法可以保证在较高计算效率的同时,能够自适应地应对参数时变问题。

关键词： fiber laser   light pressure   measurement   measurement uncertainty   traceable measurement   ultra-high power laser  

摘要： Objective Accurate power measurement is the basis of high-power laser applications, and ensuring the quality of laser-processed products is crucial. Owing to the progress in fiber-laser technology, the maximum output power afforded by industrial-grade fiber lasers has increased continually. A set of ultrahigh-power laser-measurement devices based on the principle of light pressure was established to measure the output power of a 200 kW industrial-grade fiber laser. Methods In this study, a focused laser beam was incident at a small angle to a light-pressure power meter (LPPM) after being output from the lens set;furthermore, it was absorbed by the laser dump after being output from the LPPM, as shown in Fig. 1. The LPPM was equipped with three ultrahigh-reflectance, high-damage-threshold mirrors, whose reflectance exceeds 99.995% in the 950‒1100 nm waveband. Two rectangular mirrors were mounted on the top of the LPPM at an angle of 110°, and another 90-mm-diameter mirror was connected to the weighing module. Results and Discussions We performed power measurements at different levels, as shown in Fig. 2(a). The measurement repeatability is better than 0.1% (calculated using the polar deviation method), and the response time of 0%‒100% power is less than 3 s. The full-power output is 200.50 kW, with a response time of 2.6 s, as shown in Fig. 2(b). The measurement uncertainty of the laser power was evaluated and calculated based on the guide to the uncertainty in measurement (GUM) and the Monte Carlo method (MCM), as shown in Fig. 3. The rapid decrease in the measurement uncertainty in the power range of 3‒10 kW is primarily caused by the gradual decrease in the contributions of the mass-measurement uncertainty component and resolving power to the measurement uncertainty as the measured power increases. When the laser power exceeds 10 kW, the variation in the measurement uncertainty stabilizes and is better than 0.76% (confidence factor k = 2). Conclusions We report th

关键词： 专项护理   超声   外周动静脉穿刺   穿刺成功率   护理满意度  

摘要： 目的 探讨专项护理对超声引导下外周动静脉精准穿刺患者穿刺成功率及护理满意度的影响。方法 选取2020年11月至2022年10月本院收治的87例超声引导下外周动静脉精准穿刺患者为研究对象,以随机数字表法将其分为对照组(44例)与干预组(43例)。对照组给予常规护理,干预组给予专项护理。比较两组的干预效果。结果 干预组的一次性穿刺成功率高于对照组,差异具有统计学意义(P<0.05)。干预后,两组的舒适状况量表(GCQ)评分均升高,且干预组高于对照组,差异具有统计学意义(P<0.05)。干预后,两组的抑郁自评量表(SDS)、焦虑自评量表(SAS)评分均降低,且干预组低于对照组,差异具有统计学意义(P<0.05)。干预组的并发症总发生率低于对照组,差异具有统计学意义(P<0.05)。干预组的护理满意度高于对照组,差异具有统计学意义(P<0.05)。结论 专项护理用于超声引导下外周动静脉精准穿刺患者中的效果显著,能够有效促进穿刺成功,改善心理状态,减少并发症发生,并提高舒适度、护理满意度,值得推广。

关键词： 高功率微波   线缆耦合   响应规律   影响因素   优化布线  

摘要： 高功率微波易通过电子设备间互连线缆这一主要耦合途径进入系统内部,扰乱甚至损坏敏感电路或器件。为指导工程上合理布线,提升电子系统在高功率微波环境下生存能力,采用仿真分析和试验验证相结合的方法,系统性研究了不同参数条件(线缆长度、离地高度、端接负载阻值、辐射场入射角)下高功率微波与线缆的耦合效应,获取了耦合响应规律并分析了内在原因。结果表明:耦合信号随线缆长度增加先振荡变化后逐渐趋于稳定,振荡周期与入射波波长相等;耦合信号随线缆距地高度变化呈现振荡变化,极大值和极小值分别出现在距地高度为入射波1/4波长的奇数倍以及1/2波长的整数倍时;耦合信号随端接负载阻值增加先变小后变大,当负载阻值与线缆特性阻抗匹配时,耦合信号最小;耦合信号随来波方向与线缆布设方向间夹角的增大而增大,当两者垂直时,耦合信号最大。在此基础上给出实际工程中线缆敷设优化建议。

关键词： 非线性扩张状态观测器   Fal函数滤波器   CFal函数   收敛性   稳定性   新型观测器  

摘要： Fal函数是非线性扩张状态观测器(Fal_NESO)的核心单元,其构成的滤波器(FalFilter)对于控制系统的性能具有重要影响.通过对FalFilter和Fal函数的性能分析,提出一种在定义域内连续可导且只有一个待整定参数的CFal函数,并证明了由CFal函数构成滤波器(CFalFilter)和非线性扩张状态观测器(CFal_NESO)的可行性;然后,联合CFalFilter和CFal_NESO,得到一种带滤波器的新型观测器(CFalFilter-CFal_NESO),用于处理测量环节含噪声干扰的情况;最后,通过典型实例将CFal函数与两个改进的Fal函数进行对比实验.实验结果表明,在跟踪速度趋近条件下,CFalFilter滤波效果最佳;仅改变Fal函数参数,CFal_NESO能够更好地提取噪声环境中的信号状态,同时减小因观测器阶数增加而带来的参数整定困难;在不改变参数条件下,CFalFilter-CFal_NESO进一步提高了对带有测量噪声信号的状态跟踪性能.所提CFalFilter、CFal_NESO和CFalFilter-CFal_NESO对于大多数非线性、高阶系统的滤波和观测均能适用,具有重要的学术意义和实际应用价值.

关键词： K-means聚类   LSTM神经网络   支持向量回归   差分进化法   光伏功率预测  

摘要： 为进一步提高光伏发电功率预测的准确性,提出一种基于长短期记忆神经网络(LSTM)和支持向量回归(SVR)的组合预测模型。分别利用LSTM和SVR模型对光伏功率进行预测,在此基础上采用Stacking堆叠集成的策略对2种单一模型预测结果进行线性组合,并使用差分进化算法(DE)寻找最佳组合权重。最后,对宁夏某光伏电站的真实数据进行仿真和对比研究,结果表明该方法对比LSTM和SVR模型预测误差减小约70%。

关键词： 湍流风   三风机   最优布局   风机间距   功率提升率  

摘要： [目的]在全球能源结构转型和气候变化应对的大背景下,风能作为清洁、可再生的能源之一,其利用效率的优化已成为当前能源领域研究的重点。为减小因风机间尾流效应对风电场输出功率的影响,提高风电场效率,对风机进行合理布局是很有必要的。[方法]运用Fast Farm模拟技术,分别研究了串列以及错列布置下随着风机间间距的改变对于湍流风场中3台NREL 5 MW风机输出功率及功率提升率的变化情况。[结果]当风机间间距增大时,可以明显提升后排风机的输出功率,但其提升程度有所不同,其中,当风机纵向间距由5D增加到6D及10D增加到11D时,可使风电场发电量得到较大提升;对于风机间横向间距在合理取值范围内,可使得风电场布局达到最优。[结论]文章对风机间间距进行了较为详细的划分,通过不同工况下对风电场功率影响程度分析可知在串列及错列条件下对于风机的优化布局存在差异,为风电场的布局优化提供了一定的参考。

关键词： 新型电力系统惯量   频率响应   频率分区   谱聚类   改进ARMAX模型  

摘要： 随着高比例可再生能源并网,电力系统频率的动态行为愈加复杂,准确评估系统惯量是保障电网安全稳定运行的关键。首先,通过考虑频率响应的谱聚类方法对新型电力系统进行区域划分,可以实现系统的有效分区;然后,提出改进ARMAX模型的新型电力系统区域惯量评估方法,方法采用含有遗忘因子的递推最小二乘算法,将惯量评估问题转化为对改进ARMAX模型的参数辨识问题。算例分析,在小扰动场景下对体现新型电力系统的改进10机39节点系统进行区域惯量评估误差为3.38%,对比系统不分区误差7.61%,本文方法评估惯量精确性有所提高。

关键词： Noise   Electromagnetic interference   Integrated circuit modeling   Capacitance   Frequency conversion   Mathematical models   Frequency estimation   MOSFET   Windings   Analytical models   Common mode (CM)   differential mode (DM)   electromagnetic interference (EMI)   power factor correction (PFC)  

摘要： This work presents a novel all-inclusive power electronic converter noise model comprised of both differential-mode (DM) and common-mode (CM) parasitic circuit components. Furthermore, a thorough modeling method and novel experiment-driven methodology to analyze the impact of the DM and CM circuit components on the resultant conducted emission electromagnetic interference in a single-phase power factor correction boost converter rated for 1 kW 120 VAC/400 VDC utilizing silicon-carbide mosfets is presented. This is achieved by predicting DM and CM noise corner frequencies and observing DM/CM noise corner frequencies in a novel half-bridge noise cell-based, all-inclusive converter parasitic circuit model. Frequency spectrum results find that eight DM noise corner frequencies are estimated by the proposed all-inclusive noise model with low average error of 6.45%, and the model further successfully identifies lumped CM capacitances present in the power converter system.

关键词： Oxide semiconductors  

摘要： Superjunction (SJ) technology offers a promising solution to the challenges faced by silicon carbide (SiC) Metal Oxide Semiconductor Field-Effect Transistors (MOSFETs) operating at high voltages (>3 kV). However, the fabrication of SJ devices presents significant challenges due to fabrication complexity. This paper presents a comprehensive analysis of a feasible and easy-to-fabricate semi-superjunction (SSJ) design for 3.3 kV SiC MOSFETs. The proposed approach utilizes trench etching and sidewall implantation, with a tilted trench to facilitate the implantation process. Through Technology Computer-Aided Design (TCAD) simulations, we investigate the effects of the p-type sidewall on the charge balance and how it affects key performance characteristics, such as breakdown voltage (BV) and on-state resistance (RDS-ON). In particular, both planar gate (PSSJ) and trench gate (TSSJ) designs are simulated to evaluate their performance improvements over conventional planar MOSFETs. The PSSJ design achieves a 2.5% increase in BV and a 48.7% reduction in RDS-ON, while the TSSJ design further optimizes these trade-offs, with a 3.1% improvement in BV and a significant 64.8% reduction in RDS-ON compared to the benchmark. These results underscore the potential of tilted trench SSJ designs to significantly enhance the performance of SiC SSJ MOSFETs for high-voltage power electronics while simplifying fabrication and lowering costs. © 2025 by the authors.