摘要
文章基于大量历史数据,在深度学习神经网络的基础上,构建了基于各变量随时间变化的非平稳动态模型,并结合各在线实时的测量装置模型,将状态估计问题转化在Kalman滤波框架下进行;针对目前预测方法预测步数少的不足,设定较长的预测周期,并将该周期内的所有变量视为一个整体的块向量,并据此改写相适应的块状态Kalman滤波模型;建立可同时实现点点实时估计滤波器及固定预测长度的块状态预测估计滤波器;利用计算机数字仿真结果对块状态预测滤波器的有效性进行实验验证,误差比较显示,改进算法与现有的Kalman滤波方法相比,预测效果前者均好于后者。
Based on a large amount of historical data,a deep learning neural network is used to construct a non-stationary dynamic model of each variable with time,and combined with online real-time measuring device model,the state estimation problem is transformed into the Kalman filtering framework. In view of the shortcomings of the number of prediction steps in the current prediction method,set a longer prediction period,and treat all variables in the period as an overall block vector,and then rewrite the adaptive block state Kalman filtering model. Further,establish a filter that can simultaneously implement realtime point-to-point estimation and a block state prediction estimation with a fixed prediction length. Finally,the effectiveness of the block state prediction filter is experimentally verified by computer numerical simulation. The error comparison shows that the improved Kalman filtering method has better prediction effect than the existing Kalman filtering method.
引文
[1]邹德天,潘燕,吴贝科,等.比亚迪电力储能设备与电动汽车应用现状分析[J].供用电,2018,35(9):53-61.ZOU Detian,PAN Yan,WU Beike,et al.Analysis of current-t status of BYD power storage equipment and electric vehicle application[J].Distribution-n&Utilization,2018,35(9):53-61
[2]李琼慧.太阳能光热发电发展现状与市场前景[J].电器工业,2011(8):28-31.LI Qionghui.Development status and market prospect of solar photothermal power generation[J].China Electrical Equipment Industry,2011(8):28-31.
[3]黄裕荣,侯元元,高子涵.国际太阳能光热发电产业发展现状及前景分析[J].科技和产业,2014,14(9):54-56.HUANG Yurong,HOU Yuanyuan,GAO Zihan.The development status and prospect analysis of international concentrating solar power industry[J].Science Technology and Industry,2014,14(9):54-56.
[4]陈昕,范海涛.太阳能光热发电技术发展现状[J].能源与环境,2012(1):90-92.
[5]徐静,黄南天,王文婷,等.基于GA-ELM神经网络的逐时太阳辐照量预测[J].电网与清洁能源,2016,32(8):105-109.XU Jing,HUANG Nantian,WANG Wenting,et al.Hourly solar radiation forecasting based on GA-ELM neural network[J].Advances of Power System&Hydroelectric Engineering,2016,32(8):105-109.
[6]文韬,葛泉波.未建模系统基于观测值的实时分块Kalman滤波估计方法研究[J].电子学报,2012,40(10):1958-1964.WEN Tao,GE Quanbo.Research on real-time block Kalman filtering estimation methods for the un-modeled system based on output measurements[J].Acta Electronica Sinica,2012,40(10):1958-1964.
[7]BI R,LIU XF.Active power control algorithm for distributed photovoltaic power generation cluster based on operating efficiency of photovoltaic power generation unit[J].The Journal of Engineering,2017(13):2309-2313.
[8]李霄,王昕,郑益慧,等.基于改进最小二乘支持向量机和预测误差校正的短期风电负荷预测[J].电力系统保护与控制,2015,43(11):63-69.LI Xiao,WANG Xin,ZHENG Yihui.Short-term wind load forecasting based on improved LSSVM and error forecasting correction[J].Power System Protection and Control,2015,43(11):63-69.
[9]WILSON A C,ROELOFS R,STERN M,et al.The marginal value of adaptive gradient methods in machine learning[J].Machine Learning,2017(5):1-14.
[10]KUMARI S A,SRINIVASAN S.Estimation of reheater cleanliness factor based on Kalman filtering method in neural network training[C]//2017 Trends in Industrial Measurement and Automation(TIMA),6-8 Jan.2017,Chennai,India:1-4.
[11]刘辉乐,刘天琪,黄志华.基于Kalman滤波原理的电力系统动态状态估计的研究综述[J].继电器,2004,32(20):62-66.LIU Huile,LIU Tianqi,HUANG Zhihua.Research on dynamic state estimation based on Kalman theory in power system[J].Relay,2004,32(20):62-66.