Research On Energy Management For Electric Vehicle Battery Charge-Swap-Storge-Discharge Power Station With Multi-Energy

Renewable energy and electric vehicles (EVs) have received global attention due to environmental and energy crisis. However, EVs charging infrastructure deployment affects the promotion and development of EVs. Electric vehicles battery charge-swap-storage-discharge power station(EV-BCSSDPS) with multi-energy, serving as an important integrated charging infrastructure, integrate the charging stations, battery swap station, and energy storage power station features together. EV-BCSSDPS with multi-energy can not only provide for fast and convenient service for battery swapping, but also provide for the clean energy for electric vehicles charging. Moreover, second-life batteries can be used as back-up power storage. With the construction of smart grid and a breakthrough in battery technology, EV-BCSSDPS, serving as an important part of the smart grid power plants, rational use of its energy storage characteristics will make great contribution to stabilizing power system load fluctuations, accepting intermittent renewable energy sources and improving the efficiency of grid operation. In this paper, basic theory and design methods to predict energy and energy management for EV-BCSSDPS with multi-energy are researched, including photovoltaic (PV) power forecast, charging load demand forecast, economic and optimal operation strategies of EV-CSSDPS with multi-energy and cost and benefit analysis.Accurately predicting the output of PV systems has a very important significance to properly dispatch the battery in EV-BCSSDPS, second-life batteries to charge or discharge and power grid transaction. For the uncertainty power output of PV system, a short-term prediction framework of PV system output power is established. Firstly, through the theorial analysis and historical data analysis, the travel distance is used to quantify the correlation between meteorological factors and PV power. And considering the insufficient solar radiation stations available and poor performance of solar radiation forecasting in China, the ambient temperature and humility are chosed as input parameters of neural network. Then the network structure design method containing the hidden layer nodes is proposed and a quantitative assessment criterion of the accuracy of the model is also given. Hence, a PV power forecasting model based on back propagation (BP) neural network is initiallyproposed. In order to enhance the model capacity of adapting the sudden weather changes, cloud cover is used to recognize the weather type, and the self-organizing feature map (SOM) is used to cluster the future weather type. Then, PV power generation in each weather type could be forecasted from its corresponding forecast network. Therefore, the over fitting drawbacks of single network model issue can be addressed. In addition, the anti-disturbance of weather foreacasting errors is also validated.Due to the randomness of batteries’ swapping&charging patterns, the load demand of EV-BCSSDPS has stochastic nature. It is necessary to investigate the charging load characteristics of EV-BCSSDPS to guide the coordinated battery charging for the purpose of mitigating the impact of disorderly charging behaviors on distribution network to guarantee the safe operation of the power gird. Under the uncontrolled swapping&charging scenario four variables are essential:hourly numbers of EV for battery swapping, the battery swapping&charging start time, the travel distance, and the charging duration. The BP and Radial Basis Function (RBF) nueral network and statisitical modeling methods are used to build the model of hourly numbers of EVs for battery swapping. Then the charging start time model is built based on the model of hourly numbers of EVs for battery swapping. According to statistical data of EVs travel distance, the Gaussian mixture model (GMM) is used to set up a statistical model of distance which can indirectly reflect the initial state-of-charge (SOC0). Besides, considering the charging characteristic of batteries and chargers, the charging duration model is also proposed. Taking above factors into account, a novel model based on Monte Carlo Simulation is presented to eliminate the impact of random sampling and correctly estimate uncontrolled charging power consumption of EV-BCSSDPS for forecasting horizon of24h ahead due to transportation patterns. Then a generic non-parametric method for the estimation of prediction uncertainty is introduced.This paper establishes the energy management models and optimal operation strageties for EV-BCSSDPS with multi-energy to eliminate the adverse effect based on the PV power forecasting and battery swapping and charging load forecasting. The models of PV, batteries, and second-life batteries are analyzed and three optimal energy management models to optimally schedule charging load of swapped batteries and storage in second-life batteries are presented. The three energy management models atempt to optimize the operation of EV-BCSSDPS with multi-energy with the obejectives to minimize only the charging cost, the distribution system load variance, and the both. These energy management models take into account of the battery swapping demand, battery charging requirements, power balancing constraints, battery circulation constraints, and second-life batteries power constraints. Mixed integer programming and quadratic programming model can be solved by CPLEX sofeware, and the optimal results include battery, second-life batteries, power grid, and PV output combinations. Furthermore, the optimal results of the three models are compared with the uncoordinated charging model by the quantitative assessment of the charging cost and system load variation. Besides, in the actual operation, the battery swapping demand forecasting will inevitably introduce errors, this paper also quantitatively assesses the impact of battery demand forecasting error on the optimization on EV-BCSSDPS.Nowadays, EV-BCSSDPS for electric vehicles has become an important infrastructure contributing to the development and popularization of EVs in China. The cost-benefit model of EV-BCSSDPS is crucial for its commercialization. In this paper, the battery leased mode is chosen as the operation mode of EV-BCSSDPS. On the ground of the chosen mode, the cost-benefit model of the EV-BCSSDPS based on the time-varying economic assessment index including net present value is built which considers the whole capital costs, the maintenance and operation fees, employee fees and the service for swap and charging income et al. Finally, taken one real intelligent demonstration of EV-BCSSDPS based on leased mode in China as an example, the whole life cost-benefit model is presented and the key elements affecting the station’s revenue are found out through sensitivity analysis. The model of EV-BCSSDPS and its sensitivity analysis can provide some basis for assessment of commercial operation of BCSSDPS for electric vehicles.