Online Energy Scheduling And Cost Optimization For Energy Internet Microgrids With Distributed Generations And Energy Storages

Energy Internet is the interconnected,digital,and intelligent evolution of the energy system,providing high-efficiency energy support for the development of the Internet of Things(Io T).Microgrid is a regional energy internet composed of renewable energy resources,distributed generations and energy storage systems,combined heat and power(CHP)systems,etc.It can provide energy supply to areas such as residences,buildings,streets,etc.,and can operate in grid-connected or islanded mode.The energy internet microgrid is a type of cyber-physical system.We use information flow to regulate its energy flow.It first adopts the Io T technology to collect system states and external environment data,shares and exchanges information with the external energy internet through the communication network simultaneously,then computes and optimizes the current energy scheduling strategy,and finally controls the system power output,voltage,and other operating parameters.Due to the distributed,decentralized,and time-varying network structure of the microgrid,this thesis studies the online energy scheduling strategy that does not depend on any future information,focusing on tackling the key challenges of the randomness of renewable energy outputs,the heterogeneity of distributed energy storages,the coupling in CHP system,the real-time balance between supply and demand,the diversity of users’ demands,the inconsistency between scheduling timescales,and the uncertainty of worst-case performance,aiming at minimizing the long-term operating cost and improving the economy,stability,and scalability of microgrids.We present an efficient online algorithm framework for microgrid energy scheduling by formulating optimization problems,designing online algorithms,theoretically analyzing performance,and evaluating by simulation experiments.The main works and contributions of this thesis are as follows:1.Due to the heterogeneity of the capacity and the maximum charging/discharging power of the distributed energy storage system and the coupling between the power and heat supply in the CHP system,we formulate the long-term cost optimization problem of a CHP based microgrid with distributed electricity and heat energy storage systems and electric vehicles(EVs),considering the random renewable energy outputs,electricity market prices,electricity and heat demands,and electric vehicle arrival and departure times.We propose an online energy scheduling algorithm based on the stochastic network optimization approach that optimizes the problem by virtualizing all multiple energy storages into a single battery,thermal tank and EV.We prove that the performance gap between the proposed algorithm and the optimal solution is smaller than that of the conventional Lyapunov optimization approach when the capacity and the maximum charging/discharging power of the energy storage system are not identical,which addresses the problem that the performance of the existing online algorithm used in the scenario of multiple energy storages is limited by the minimum capacity of the energy storages.Simulation compares the proposed algorithm with the traditional online algorithm under real-world electricity price and demand data,and demonstrates that the proposed online algorithm can further reduce the average cost of the microgrid.2.Due to the diversity of electricity demands and quality of service(Qo S)requirements and the high-dimensional and computational complexity of the strategy caused by the joint scheduling on both the supply and demand sides in the multi-microgrid multi-user energy internet,we formulate a total system cost optimization problem,considering the multi-microgrid coordinated scheduling and multi-user flexible demand management models.By transforming the Qo S constraints of users’ flexible demands into the power outage backlog queues,we propose a centralized online energy scheduling strategy and provide the deterministic performance gap with the optimal solution.Based on the Alternating Direction Method of Multipliers(ADMM),we further propose a distributed algorithm that first solves the amount of supply for flexible demand independently by the users on the demand side and then solves the energy scheduling strategy by the microgrids in a distributed way on the supply side.Moreover,under the proposed distributed algorithm,the microgrids are not required to share their individual information and only need to output their energy scheduling strategy until the solution iteratively converges to the optimum,which protects their data privacy.Simulation shows the effectiveness of both algorithms,and demonstrates that the performance gap between the distributed and centralized algorithms is tiny and can be ignored.3.Due to the inconsistency of the scheduling timescale between the charging/discharging strategy of the energy storage system and the generator start-up/shut-down strategy,we introduce a second slower timescale to schedule the generator start-up/shutdown strategy.We also consider the start-up/shut-down cost of local generator in the microgrid operation cost as one of the optimization objectives.We present a two-timescale energy scheduling model of a microgrid with an energy storage system and a local generator and formulate a long-term cost optimization problem.First,we improve and propose a multi-time-slot Lyapunov Optimization approach and design a two-timescale online energy scheduling algorithm.Then,we derive the closed-form expression of the average start-up/shut-down cost under the proposed algorithm and its upper bound.Based on the Central Limit Theorem,we give an estimation method for its approximate value.Finally,simulation shows that the two-timescale energy scheduling algorithm can effectively reduce the cost of the microgrid and the number of starts and stops of the generator,and it also demonstrates that the proposed algorithm can outperform the existing sliding-window algorithm and reinforcement algorithm.4.We study the online start-up/shut-down decision problem of generators,considering the start-up cost,no-load cost,and shortest uptime limits.We formulate a microgrid cost optimization problem with a multi-time-slot coupling start-up/shut-down constraint.First,we propose a deterministic online algorithm,and derive its closedform competitive ratio.We prove that its competitive ratio is always less than 4 under any system parameters,which characterizes the worst-case performance bound of the proposed online algorithm.Then,based on the deterministic online algorithm,we propose a random online algorithm,which is enable to further reduce the competitive ratio.Finally,we propose two variants tailored for average-case inputs and time-varying parameters,respectively.Simulation shows the effectiveness of the proposed online algorithm framework and verifies that it can guarantee the performance gap to the optimal offline solution in the worst case.