Research On Demand Side Collaborative Measurement And Energy Efficient Optimization Method In Cyber-physical Energy Systems

Cyber-physical energy systems (CPESs), which are applied to smart grid and arecomposed of immense number of sensors, embedded computing devices and large scaleheterogeneous communication networks, are systems combined sensing,computing,communication and control. Cyber physical energy systems have thecapabilities of ubiquitous collaborative sensing, intelligent information processing,real-time and flexible interactive communication and dynamic control for complicatedsystems. This dissertation applied the basic theory to utilization micro girds, andresearch on demand side utilization measurement and energy efficient optimizationmethod in cyber-physical energy systems. The main contributions of this dissertationare as follows:The problem of non-intrusive load identification is also researched forcyber-physical energy systems based utilization micogrids and the solution whichexploited the cooperation of steady and transient features of electrical load are designed.The empirical Bayesian load identification algorithm is researched to decompose thesteady load. For the transient and non-steady featured load, Gaussian process methodwith feature extraction and reduction are investigated. To enhance the performances ofthe method, a multi-model collaborative load identification methods are developed.The problem of demand side load forecasting is researched in this dissertation andthe method of cyber-physical relevance based probabilistic forecasting is proposed. Toreduce the disturbance from redundant information, an cyber-physical relevancecomputing based algorithm is developed to prune data for load forecsting model. Anempirical mode decomposition based multi-resolution load decomposition method isintroduced to analyze the features of electrical load in different frequency domain. Thehierarchical feature weighted sparse Bayesian learning algorithm, which have thecharacteristic of low forecasting error and robustness, is proposed for short-term loadforecasting in utilization networks.In order to optimize the wireless communication energy consumption profile, thecomputing intelligence based method for route optimizating are introduced. Themulti-layer client/server collaborative sensing network model are designed for CPESs utilization networks sensing and computing appliances. To enchance the route energyefficiency of the porposed wireless network infrastructure, a best solution feedbackbased intelligent water drop algorithm is developed and the performances of thismethod are also researched in this dissertation.The problem of energy efficience evalutiaon and optimization for utilizationnetworks is investigated in this dissertation. A load forecsting based dynamicpeak-valley price theoretical model are designed to dynamically guide the profile ofutilization load. To evaluate the energy efficiency of the micro grid, a delay costmodification and welfare function based evaluation system are developed. To reduce thepeak load and improve the energy efficiency of utilization networks, the method ofvirtual force supervised partical swarm optimization method are introduced.A measurement system of CPESs is designed and implemented. A low power andplug-in wireless sensor node for electrical signal sensing is developed. An integratedtesting software platform is constructed, which can monitor the working status, controlthe configuration and achieve extended simulation for utilization networks energyoptimization. Based on this measurement system, the actual performances of theproposed short-term load forecasting algorithm and the proposed non-intrusive loadidentification algorithms are investigated.