基于量测残差的改进参数估计方法

电网参数错误将影响状态估计的质量,并降低能量管理系统中其他高级应用软件的实用化程度。文中简要介绍了主要的参数估计算法,并详细分析了基于量测残差和参数误差间灵敏度关系的参数估计方法。在总体沿袭两步法参数估计方法的基础上,对该方法的近似模型进行了详细分析。为使算法具有更广的适用面,对第1步中偏差向量的计算和第2步中参数的估计方法分别进行了改进。对该算法进行了测试,结果表明,该方法具有良好的应用前景。

预测残差判别统计量抗差因子在UKF算法中的应用

在UKF滤波中,针对观测信息存在粗差的问题,基于预测残差构建了预测残差判别统计量。结合三段函数组成了一种新的抗差因子函数,并给出抗差UKF算法公式。该算法计算过程无需迭代,实现一步抗差,适合实时滤波估计。计算结果表明,该抗差因子函数不仅能够有效地控制含粗差观测值对滤波结果的影响,得到可靠的滤波解,而且可以提高滤波精度。

基于无人船的双分支解码轻量型分割网络研究

为保证水面无人艇(USVs)进行水上任务时能够顺利航行,需要对河道信息进行精确的提取,因此,对河道语义分割的网络模型进行了研究。针对河道图像分割中类间不一致和类内不一致的问题,文中提出了分割网络DBDL-Net,网络中设计双分支解码结构和双重损失函数,分别把握语义信息和空间信息;同时在编码部分设计了多尺度残差的轻量模块,一方面减少参数,一方面以不同的比例捕捉特征信息。最后在USVIn-land数据集上对模型进行消融和对比实验,实验结果表明:DBDL-Net的精确度和平均交并比最终达到了93.619%和87.682%,与其他先进分割网络相比,DBDL-Net也具有更佳的综合表现。

一种用于平抑风光功率波动的混合储能容量配置方法

风光发电功率具有强随机性和波动性,直接并网会引起电网频率不稳定,可利用混合储能系统有效平抑。文中以全钒液流电池和超级电容组成混合储能系统作为研究对象,提出一种混合储能容量配置方法。首先,采用自适应移动平均算法确定风光并网功率和储能系统功率;然后,利用鲸鱼优化的VMD分解储能系统功率,得到一系列模态分量和残差量,考虑残差量中可能包含的丰富信息,对其作相同的VMD分解,将两次分解产生的各模态分量分别作希尔伯特边际谱分析,确定分界频率,将所有低于分界频率的模态分量分配给全钒液流电池,其余由超级电容承担;最后,根据全钒液流电池和超级电容各自吸收、补偿功率情况,配置储能系统的额定功率和额定容量。算例分析表明,文中方法能够有效平抑风光输出功率波动,并实现混合储能系统额定功率、容量合理配置。

步进电机电流信号测量及其负载估算方法研究

因为感应电机、伺服电机的输出力矩大小同其电流的幅值有直接的关系,且电流信号易于采集,故可以直接由采集电流的幅值,评判电机的负载情况。但是对于步进电机而言,其驱动控制原理同上述两种电机有较大不同,采用恒流斩波控制方式,无论电机的末端负载多大,均表现为相近的电流波形。这造成仅通过采集的电流信号,无法直接表征设备的负载情况。本文通过深入分析步进电机的驱动原理,基于设备位置信号和相电流信号,得到同电机负载正相关的增量残差累加特征量。通过搭建专用的测试平台,基于实测实验验证了本文提出方法的有效性。

北斗导航系统的整周模糊度解算及周跳探测

影响北斗卫星导航系统(BDS,Bei Dou navigation satellite system)观测数据质量的重要指标包括整周模糊度的确定、周跳探测与修正。首先,从码伪距观测方程和载波相位观测方程出发,研究BDS整周模糊度解算方法。其次,利用电离层残差量作为探测周跳的手段,通过仿真实验分析该方法的可行性。实验表明:当周跳小于8周时,用电离层残差探测周跳比较适合;若周跳大于8周,需选用其它方法。

基于改进的电离层延迟的BDS周跳探测方法

针对电离层延迟变化较剧烈时,现有的北斗导航卫星系统(BeiDou Navigation Satellite System,BDS)双频周跳探测方法探测效果会降低的问题,文章提出了改进的电离层延迟法,该方法首先基于灰色理论对电离层延迟变化量进行预测,然后根据实际值与预测值之差构造的电离层残差量作为新的周跳探测量进行周跳探测;通过实验验证了该方法在电离层延迟变化较剧烈,尤其在低采样率、低卫星高度角时,能有效弥补原来方法存在的不足。

改进PCA及其在过程监测与故障诊断中的应用

提出一种改进的主元分析 (PCA)方法 ,采用主元相关变量残差 (PVR)统计量代替通常的平方预测误差Q统计量 ,用于工业过程的监测与故障诊断。改进PCA避免了Q统计量的保守性 ,能够提供更详细的过程变化信息 ,从而有效识别正常工况改变与过程故障引起的T2 图变化。通过对双效蒸发过程的仿真监测 ,与普通PCA方法进行了比较 。

基于VMD-LSTM-LightGBM的多特征短期电力负荷预测

针对目前多特征电力负荷预测精度不准的问题,为充分挖掘电力负荷数据中的时序信息、天气信息等特征信息,提出了一种基于变分模态分解(variational mode decomposition,VMD)的长短期记忆(long short-term memory,LSTM)神经网络与轻量级梯度提升机(light gradient boosting machine,LightGBM)预测模型,优化负荷数据非线性、非平稳、长记忆等问题,解决了多特征预测提取特征信息差的问题。该方法首先用VMD分解代表不同尺度的特征模态分量,降低了原始序列的不平稳度,同时分解的残差量代表负荷数据强非线性部分,通过特征性强的算法进行预测,将各模态分量通过LSTM的单特征预测,再将各个分量加入多特征利用LightGBM进行负荷预测。通过与目前多特征电力负荷预测模型进行对比实验,平均绝对误差(mean absolute error,MAE)值仅为其23%~73%,平均绝对百分比误差(mean absolute percentage error,MAPE)值能达到0.37%,具有更好的预测精度。

线性模型的在线检测

为了使检测统计量本身不受数据较大波动的影响,使用标准化残差来构造指数滑动平均统计量(Exponentially Weighted Moving Average statistic),并且采用迭代方法来更新估计参数,得到新的EWMA统计量及其分布.

室内环境下对OLOS误差具有鲁棒性的TOA无线传感器网络定位算法

针对无线传感器网络及室内环境的特点,在综合考虑节点定位所需的通信量、计算量和定位精度的基础上,提出一种室内环境下对受阻碍的视距(OLOS)误差具有鲁棒性的低计算量的残差加权无线传感器网络定位算法。该算法利用对残差优选再加权的方法,在未知信道特性和无须反复通信的条件下对距离测量值的组合进行了优化。分析表明该算法与同类算法相比具有低计算量的特点。大量仿真结果表明该算法提高了定位精度,有效地抑制了室内环境下OLOS误差。

基于小波变换的GPS周跳探测方法

研究了GPS定位理论中载波相位观测量及载波相位与伪距组合观测量的特点 ,给出了一种用小波变换探测GPS周跳的方法 ,通过分析得出了优于传统方法的结论。

基于最小二乘法的电力系统网络参数误差估计

为满足电力系统实时动态观测的要求,需要精确估计电网络参数,基于电力系统的任何网络参数误差都会影响状态估计中的量测残差,通过对量测残差的分析提出了一个基于最小二乘的电网络参数误差估计方法。分两步完成电网络参数误差的估计:首先根据动态系统的量测方程,利用加权最小二乘法(weighted least squares method,以下简称WLS),同时引入网络节点虚拟量测量作为等式约束条件建立最优化模型,利用拉格朗日乘法提出量测残差的表达式,网络节点虚拟量测量的引入加强了算法的精确性和收敛速度;第二步首先利用递推最小二乘建立量测残差方程中偏差项的的迭代公式,电网络参数误差包含在量测残差方程的偏差项中,通过迭代得出一个无偏估计的初始状态量,最后求得电网络参数误差。IEEE9节点系统的仿真验证了算法的正确性和有效性。

基于混合模型的云平台资源弹性伸缩方法

为使云资源数据最大化匹配平台主机的存储条件,保障云服务的质量水平,提出基于混合模型的云平台资源弹性伸缩方法。根据聚类混合系数指标,建立完整的初始定位条件,通过计算云数据优选残差加权量的方式,完成基于混合模型的云数据定位。在此基础上,预测平台主机的需求变化趋势,按照资源数据的采集与融合标准,定义必要的弹性扩容与伸缩扩容条件,实现基于混合模型云平台资源弹性伸缩方法的顺利应用。实验结果表明,与高斯负载型数据伸缩方法相比,在混合模型支持下,云平台主机可将资源数据最小压缩至0.125 bit,最大拉伸至16 bit,能够根据平台主机的实际存储条件,对云资源数据传输体积进行及时调整。

Characteristics of physical blocking on co-occupant's exposure to respiratory droplet residuals

Existed evidences show that airborne transmission of human respiratory droplets may be related with the spread of some infectious disease, such as severe acute respiratory syndrome （SARS） and H1N1 pandemic. Non-pharmaceutical approaches, including ventilation system and personal protection, are believed to have certain positive effects on the reduction of co-occupant＇s inhalation. This work then aims to numerically study the performances of mouth covering on co-occupant＇s exposure under mixing ventilation （MV）, under-floor air distribution （UFAD） and displacement ventilation （DV） system, using drift-flux model. Desk partition, as one generally employed arrangement in plan office, is also investigated under MV. The dispersion of 1, 5 and 10 grn droplet residuals are numerically calculated and CO2 is used to represent tracer gas. The results show that using mouth covering by the infected person can reduce the co-occupant＇s inhalation greatly by interrupting direct spread of the expelled droplets, and best performance can be achieved under DV since the coughed air is mainly confined in the microenvironment of the infected person. The researches under MV show that the two interventions, mouth covering and desk partition, achieve almost the sarae inhalation for fine droplets while the inhalation of the co-occupant is lower when using mouth covering for large droplets.

A new damping ratio identification method based on pattern search

In order to improve the effectiveness of traditional time domain identification methods in identifying damping ratios, a new damping ratio identification method based on pattern search is proposed by fluctuating the reliable natural frequency obtained through traditional time domain identification methods by about 10% to build the boundary conditions, using all the initial identification results to establish the free decay response of the system, and using the pattern search method to correct the initial identification results with the residual sum of squares between the free decay response and the actually measured free-decay signal as the objective function. The proposed method deals with the actually measured free-decay signal with curve fitting and avoids enlarging the identified error caused by intermediate conversion, so it can effectively improve the identified accuracy of damping ratios. Simulations for a room-sized vibration isolation foundation show that the relative errors of analyzed three damping ratios are down to 1.05%, 1.51% and 3.7% by the proposed method from 8.42%, 5.85% and 8.5% by STD method when the noise level is 10%.

Technical assessment of synthetic natural gas (SNG) production from agriculture residuals

This paper presents thermodynamic evaluations of the agriculture residual-to-SNG process by thermochemical conversion, which mainly consists of the interconnected fluidized beds, hot gas cleaning, fluidized bed methanation reactor and Selexol absorption unit. The process was modeled using Aspen Plus soft-ware. The process performances, i.e., CH4 content in SNG~ higher heating value and yield of SNGexergy efficiencies with and without heat recovery, unit power consumption, were evaluated firstly. The results indicate that when the other parameters remain unchanged, the steam-to-biomass ratio at carbon boundary point is the optimal value for the process. Im proving the preheating temperatures of air and gasifying agent is beneficial for the SNG yield and exergy effi ciencies. Due to the effects of COz removal efficiency, there are two optimization objectives for the SNG produc tion process： （I） to maximize CH4 content in SNC or （II） to maximize SNG yield. Further, the comparison among different feedstocks indicates that the decreasing order of SNG yield is： corn stalk 〉 wheat straw 〉 rice straw. The evaluation on the potential of agriculture-based SNG shows that the potential annual production of agriculture re sidual-based SNG could be between 555 108 - 611 108 m3 with utilization of 100% of the available unexplored resources. The agriculture residual-based SNG could play a significant role on solving the big shortfall of China＇s natural gas supply in future.

Modification of the land surface energy balance relationship by introducing vertical sensible heat advection and soil heat storage over the Loess Plateau

Little is known about the surface energy balance problem for a complex underlying surface.Taking data from the Loess Plateau Land-surface Processes Experiment(LOPEX) and investigating the characteristics of the surface energy balance over a complex underlying surface,this paper calculates the soil heat storage and vertical sensible heat advection,analyzes their contributions to the surface energy imbalance,and discusses the mechanism by which the vertical velocity and temperature gradient in the surface layer affect the vertical sensible heat advection transfer.We found that the vertical velocity in the surface layer provides the necessary dynamic power for vertical sensible heat advection,and a relatively strong temperature gradient is the energy source generating vertical sensible heat advection.Under an ascending condition,the effect of vertical sensible heat advection on the surface energy budget is more obvious.It is also found that when the soil heat storage term and the vertical sensible heat advection term are added to the energy balance equation,the imbalance significantly improves.The peak of average diurnal residuals decreases from 125.1 to 41.5 W m-2,the daily average absolute value of residuals falls from 59.0 to 26.4 W m-2,and the surface energy balance closure increases from 78.4% to 94.0%.

Residual Cusum Test for Parameters Change in ARCH Errors Models with Deterministic Trend

This paper analyzes the problem of testing for parameters change in ARCH errors models with deterministic trend based on residual cusum test. It is shown that the asymptotically limiting distribution of the residual cusum test statistic is still the sup of a standard Brownian bridge under null hypothesis. In order to check this, we carry out a Monte Carlo simulation and examine the return of IBM data. The results from both simulation and real data analysis support our claim. We also can explain this phenomenon from a theoretical viewpoint that the variance in ARCH model in mainly determined by its parameters.