信息物理能源系统需求侧协作测量与能效优化方法研究

发布时间：2018-06-08 13:52

本文选题：信息物理能源系统 + 需求侧用电能效优化　；参考：《清华大学》2014年博士论文

【摘要】：信息物理能源系统是由大量感知测量节点、嵌入式计算设备和大尺度异构通信网络等构成的应用于智能电网的闭环感知联控系统，其具有协作测量，信息分析处理，实时灵活交互通信，复杂系统动态控制等功能。本论文研究基于信息物理能源系统的需求侧协作测量与能效优化方法，论文的主要研究内容包括：研究了瞬态稳态特征协作的非侵入式负荷辨识测量方法，解决了需求侧复杂动态的用电状态背景下的负荷辨识问题。面向负荷平稳特征提出了贝叶斯经验学习辨识测量方法。面向负荷非平稳瞬态特征提出了基于高斯过程的概率辨识测量方法。为解决单辨识模型易受干扰问题，提出了多辨识模型协作的非侵入用电负荷辨识测量方法，提高了负荷辨识测量的准确性和鲁棒性。为了实现对需求侧用电负荷的高精度可靠预测，提出了信息物理关联用电负荷概率预测方法。为降低冗余信息对预测模型干扰，研究了基于信息物理相关性加权的数据约简方法；针对需求侧用电负荷异频段特征，研究了基于经验模态分解的用电负荷多分辨率分析方法；提出了分层特征加权稀疏贝叶斯负荷概率预测方法，实现了对用电负荷的概率分布预测，预测精度高，算法鲁棒性好。面向信息物理能源系统需求侧协作测量网络，提出了分层客户端-服务器协作测量网络模型。研究了智能水滴无线数据传输能效优化算法，解决协作测量网络无线数据传输能耗降低问题。为了进一步降低协作测量网络无线通信能耗，提出基于最优解反馈的半监督式智能水滴路由优化算法，提高了协作测量网络能效性。提出了虚拟力监督粒子群能效性优化方法，解决需求侧用电网络的用电时间能效优化问题和负荷削峰填谷问题。面向未来开放的电力市场，，提出了基于负荷预测的峰谷动态电价理论计算模型，研究了基于延时成本修正和福利函数的用电能效性评价指标。针对需求侧用电时间优化问题，提出了虚拟力监督粒子群能效性优化算法，实现了对峰值用电负荷的平滑，优化了能效性评价指标。研制了测量电能信息的低功耗电能信息测量节点，设计了具有网络组态监测、测量数据显示、数据预处理、用电统计信息查询与负荷预测等功能的需求侧协作测量软件平台。基于设计需求侧测量系统软硬件平台验证了需求侧用电负荷在线监测、非侵入式负荷协作辨识测量和用电负荷预测在实际工作环境中的性能。
[Abstract]:The information physical energy system is a closed loop sensing system which is composed of a large number of sensor nodes, embedded computing devices and large scale heterogeneous communication networks, which is applied to the smart grid. It has cooperative measurement, information analysis and processing, etc. Real-time flexible interactive communication, complex system dynamic control and other functions. In this paper, the demand-side collaborative measurement and energy efficiency optimization methods based on information physical energy system are studied. The main contents of this paper are as follows: the non-invasive load identification and measurement method based on transient steady-state characteristic collaboration is studied. The problem of load identification under the background of complex dynamic power consumption on the demand side is solved. A Bayesian empirical learning identification method for load stationary features is proposed. A probabilistic identification and measurement method based on Gao Si process is proposed for load nonstationary transient characteristics. In order to solve the problem that single identification model is vulnerable to interference, a multi-identification model cooperative non-invasive load identification method is proposed. The accuracy and robustness of load identification and measurement are improved. In order to achieve high accuracy and reliability prediction of demand side load, a probability forecasting method of information physics correlation is proposed. In order to reduce the interference of redundant information to the prediction model, the data reduction method based on the weight of information physics correlation is studied, and the multi-resolution analysis method based on empirical mode decomposition is studied for the different frequency band characteristics of the demand side load. A hierarchical feature weighted sparse Bayesian load probability forecasting method is proposed to predict the probability distribution of power load. The prediction accuracy is high and the algorithm is robust. A hierarchical client-server collaborative measurement network model is proposed. The energy efficiency optimization algorithm of intelligent water droplet wireless data transmission is studied to solve the problem of reducing the energy consumption of cooperative measurement network wireless data transmission. In order to further reduce the energy consumption of cooperative measurement network, a semi-supervised intelligent water droplet routing optimization algorithm based on optimal solution feedback is proposed to improve the energy efficiency of collaborative measurement network, and a virtual force supervised particle swarm optimization method is proposed. To solve the problem of energy efficiency optimization of demand side network and load peak filling. For the future open power market, a theoretical calculation model of peak-valley dynamic electricity price based on load forecasting is proposed, and the evaluation index of power efficiency based on delay cost correction and welfare function is studied. A virtual force supervised particle swarm optimization algorithm for energy efficiency optimization is proposed to solve the problem of time optimization of power consumption on the demand side, which can smooth the peak load. The energy efficiency evaluation index is optimized, the low power consumption power information measurement node is developed, and the network configuration monitoring, measurement data display, data preprocessing are designed. Demand-side collaborative measurement software platform for power statistics information query and load forecasting. Based on the software and hardware platform of the DSM system, the performance of DSM, non-intrusive load identification and load forecasting is verified in the actual working environment.
【学位授予单位】：清华大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM715

【参考文献】