| Energy is an important material foundation for the continuation and development of human society.In order to guarantee the energy safety,expedite the innovation of energy resource structure,establish a low-carbon and high-efficient energy system,the renewable energy sources are drawing more and more attention.However,with the influence of the randomness,uncertainty and uncontrollability of the renewable energy sources,the energy efficiency,power quality and stability would be reduced if the power system is simply supplied by a renewable energy generation.Therefore,with the help of coordination of renewable energy sources and energy storage techniques,as well as the modeling,estimation and management methods from information science and control science,micro-grid(MG)with the integration of power generation system,power transmission system,power distribution system,energy storage system and load is an effective solution to solve the aforementioned problems.Due to the vital function of the battery energy storage system(BESS)in the process of renewable energy conversion,storage and utilization,in order to guarantee a safe and reliable operation of the BESS in the MG,the description mechanism and state estimation methods should be conducted based on the differences of nonlinearity,complexity,and uncertainty of the BESS under different time scales,the acquisition and management of valid information in the system.Furthermore,in order to achieve the coordination of renewable energy sources and energy storage systems,the optimization objectives and constraints of the complex MG should be fully investigated,the optimal energy management mechanism of the MG should be studied,and the formulating and solving methods of the optimization problems with multi-dimension limits should be proposed.Therefore,in this thesis,on the basis of the analysis for battery dynamic characteristics,the model for battery dynamic behavior description has been built,the short time scale modeling and estimation method for the battery under dynamic operational conditions has also been presented.Based on the strong capability of neural networks for nonlinear fitting,the data-driven approaches for behavior description and long time scale state estimation of battery aging process have been developed.Moreover,the mathematical model and energy dispatching strategy to optimize the economic objective of the MG has been given based on the description and expression of the dynamic behaviors of the BESS,and the analysis of the uncertainty of the renewable energy sources.The multi-objective management strategy for the hybrid system with human,electric vehicle(EV)and MG has been proposed with the help of behavior modeling of the MG users and behavior description of battery degradation.The results showed that the developed methods in this thesis can deal with the issue of multi-timescale modeling and state estimation for the BESS,and the issue of the cooperation between renewable energy sources and BESS.The main works and contributions of this thesis can be summarized as follows:1)To deal with the state of charge(SoC)estimation problem,a hysteresis voltage based battery model(HVBBM)has been proposed by analyzing the various characteristics of the lithium-ion battery with dynamic currents and different SoC.Hence,an accurate simulating result for battery terminal voltage under dynamic current has been obtained.Meanwhile,in order to achieve an accurate SoC estimation result,an adaptive extended Kalman filter has been applied in this thesis to suppress the impacts of the measurement errors of the sensors in actual applications.The comparison results indicate that the proposed method can accurately estimate lithium-ion battery SoC under dynamic currents.2)To deal with the problem of battery remaining useful life(RUL)estimation,battery terminal voltage curves under different cycle life during constant current charge process have been investigated for RUL definition,and the relationship between RUL and battery charging curve has been simulated by a feed-forward neural network(FFNN)for online implementation.According to the nonlinearity of battery charging curve,the importance sampling(IS)has been utilized to extract the input of the FFNN.Thus,a method based on FFNN and IS has been proposed to online estimate the RUL.Experimental results based comparisons verified that the proposed approach was accurate for actual operation.3)To address the issue of battery state of health(SoH)estimation,variables have been selected according to the differential geometric analysis of battery terminal voltage curves to depict battery SoH.Moreover,the relation of the differential geometric properties and battery SoH has been modeled by the group method of data handling(GMDH)polynomial neural network.Afterward,battery SoH has been estimated by GMDH with inputs of voltage curve properties.Experiments with different types of lithium-ion batteries have been conducted to verify the effectiveness and universality of the developed methods.4)To deal with energy scheduling problem of MG based on the cooperation of renewable energy sources and the BESS,the impacts of energy storage efficiency,degradation degree of the BESS on energy management of the MG are studied in this thesis.The mathematical expression has been obtained to depict the efficiencies under both charge and discharge based on the experimental data and mechanism analysis.To minimize the operational electricity cost of the MG in the next 24 hours,a nonlinear programming with discontinuous derivatives(DNLP)solver is applied based on the proposed objective function and constraints.Additionally,to balance the power flow of MG and overcome the predicting error of the PV power,a two steps MG management strategy is therefore developed based on the scheduled power.The experimental results show that the proposed EMS can effectively improve the energy efficiency and reduce power costs.5)To address the issue for the management of“human-EV-MG" hybrid system,RUL of the electric vehicle(EV)in the MG were analyzed as the breakthrough point.An RUL forecasting method for the batteries in EV has been proposed based on the analysis of effects of different operating SoC intervals on EV cycle life.Afterward,factors that could influence the comfort level of the users,e.g.,state of charge of the EV,RUL of the EV,were investigated.The comfort model of the users and a comfort level considered EMS have been developed.Experimental results indicated that the proposed EMS could promote the usage of renewable energy,optimize the power scheduling and effectively utilize the energy of the residential MG while meeting different requirements of the users. |
PDF Full Text Request