Research On Optimal Operation Of Islanded Microgrid System Including Electrochemical Energy Storage

It is a major strategic decision made by the CPC Central Committee to comprehensively promote “Carbon Peak and Carbon Neutrality”.Building a new type of power system with new energy-based is one of the main measures to achieve China’s“carbon emission peak and carbon neutrality” strategic objectives.At present,microgrid is widely used in islands and remote areas as an important means to make full use of renewable energy and realize the complementarity of multiple energy sources.However,with the large-scale grid connection of renewable energy such as wind and solar,the security of microgrid,and new energy accommodation demand are facing the challenges of randomicity and fluctuation of renewable energy.Under this background,how to make full and efficient use of renewable energy,realize the stable and economic operation of the microgrid,and reduce the adverse impact of renewable energy connected into it,is the core and key issue of this dissertation.Therefore,this dissertation takes the microgrid as the research object,and studies its island operation control and optimization strategy.It mainly involves three levels: operational control,prediction and optimization.The operational control level is used to realize the stable operation of the system,including the modeling and control of distributed generation and the modeling and control of energy storage system.The prediction level provides relevant decision data for the optimization level,including short-term load forecasting and short-term renewable energy output power forecasting in the microgrid.The optimization level improves the economic and environmental benefits of the islanded microgrid by reasonably allocating the output of each micro source,which involves the research on the operational optimization strategy of the islanded microgrid.The main works are as follows:1)With the increasing proportion of renewable energy such as wind and solar in the islanded microgrid,its output power presents the characteristics of randomicity and fluctuation,which makes the overall modeling and stability analysis of the system with great challenges.In this dissertation,the mathematical models of wind turbines,photovoltaic and diesel generators commonly used in the islanded microgrid are described in detail.And the control strategies of each generation unit connected to the system are studied,which provides theoretical support for the subsequent stability control of the islanded microgrid system.2)In order to maximize the utilization of renewable energy such as wind and solar,give full play to the advantages of energy storage,and realize the stable operation of the island microgrid,this dissertation studies the modeling and control strategy of the electrochemical energy storage system in the islanded microgrid based on the realization of the output control of microsources such as wind turbines,photovoltaic,diesel generators,etc.Firstly,the mathematical models of four kinds of electrochemical energy storage batteries are established.Secondly,the mathematical models of bidirectional energy storage converter in V/f control mode and PQ control mode under dq coordinate are established.At the same time,aiming at the nonlinear problem of the system,the state feedback method is used for linearization decoupled,and on this basis,the two-stage charging control strategy of the storage battery of the energy storage system and the output control strategy of the energy storage converter based on improved sliding mode control are proposed.Thus,the stable operation of the islanded microgrid is realized.3)A short-term load forecasting method based on decomposition prediction and error correction is proposed to address the short-term load forecasting problem in the optimization process of microgrids,to overcome the challenges brought by the obvious fluctuation and nonlinear characteristics of load series caused by various factors such as holidays and meteorological factors.Firstly,a decomposition loss evaluation criterion is established,and the variational mode decomposition optimal decomposition parameters under the evaluation criterion are determined based on bald eagle search(BES)algorithm to decompose the load data into subsequence components of different frequencies.And the corresponding convolutional bi-directional long short-term memory(CNN-Bi-LSTM)network prediction models are established for each modal component to reduce the complexity of load time series and improve prediction performance.Then,the impact of various modal components and short-term factors on the prediction error is comprehensively considered,an error correction model based on CNN-Bi-LSTM network is established,by mining the effective information hidden in errors to reduce the error of the prediction model.This method effectively improves the accuracy of short-term load forecasting of the microgrid,and provides a load forecasting algorithm foundation for subsequent microgrid optimization.4)A renewable energy output power prediction method based on multi-model fusion stacking ensemble learning method is proposed to address the problem of traditional microsource output power prediction methods being difficult to cope with random fluctuation data and limited ability to process time series in the optimization process of microgrid.Firstly,a density-based spatial clustering of applications with noise algorithm is used to clean the raw data of microsource output power.Then,a microsource output power prediction model with multiple machine learning algorithms embedded in the stacking architecture is built,making full use of the advantages of each algorithm to improve the prediction accuracy and generalization ability of the model.The stacking base learners include CNN-Bi-LSTM model which has the advantages of deep architecture feature extraction and gives consideration to data timing and nonlinear relationship,as well as e Xtreme gradient boosting(XGBoost)and other tree ensemble learning models which were suitable for complex data modeling.The meta learner corrects the prediction bias of base learners to prevent overfitting.Taking wind turbines as an example,this method effectively improves the prediction accuracy of the output power of wind turbines in the microgrid,and provides a power forecasting algorithm foundation for subsequent microgrid optimization.5)Aiming at the uncertainty factors such as the randomicity and fluctuation of the output power of high proportion of renewable energy in the islanded microgrid and the uncontrollability of the power load,which bring challenges to the stable operation of the microgrid.To maximize the utilization of renewable energy and improve the economic benefits of islanded microgrid,an operational optimization strategy of microgrid based on BES algorithm was proposed.Firstly,based on the basic data of renewable energy output power forecast,load demand forecast,etc,the appropriate operation control mode is selected.Secondly,for each operation control mode,an operation optimization model of islanded microgrid is constructed,which aims at minimizing the comprehensive operation cost of the system including fuel consumption cost,system operation and maintenance cost,etc,and combines the power balance constraints,distributed generation unit operation constraints and other constraint conditions.Then,a model solution scheme based on BES algorithm is proposed.Finally,the optimal operation of the islanded microgrid was realized.