Energy Management Strategy Of Wayside Supercapacitor Energy Storage System Based On BP Neural Network Energy Prediction

With the rapid development of urban rail transit in China,the urban rail transit system represented by subway has the advantages of large traffic volume,high safety,fast speed,punctual efficiency,low pollution and so on.The power consumption of train operation in urban rail transit system is large.The absorption and utilization of regenerative braking energy is one of the effective measures to reduce the energy consumption of train operation.In recent years,the ground super capacitor energy storage system has been widely used as an important way to absorb and utilize the regenerative energy.It can effectively recycle the regenerative braking energy,reduce the energy consumption of train operation,reduce or eliminate the regenerative failure.In this paper,based on the traditional voltage and current double loop control strategy,the energy management strategy of the super capacitor energy storage system is optimized considering the train operation state and the regenerative energy absorbed by the super capacitor.First of all,in order to realize the synchronous power flow simulation of urban rail transit power supply system,this paper sets up the mathematical models needed for the power supply simulation of urban rail transit system,such as traction substation,super capacitor energy storage system,urban rail train,traction network,etc.On this basis,the simulation platform of urban rail transit power supply system is established,and the simulation analysis of Beijing Metro Batong line is carried out.Secondly,based on BP neural network model,a real-time prediction method of regenerative braking energy is proposed.By acquiring the real-time information of the position and power of each train on the line,the model can predict the regenerative braking energy that the energy storage system needs to absorb,and judge whether the energy storage system will be full in the charging process.When the energy storage system can not fully absorb the regenerative braking energy,properly increasing the charging threshold can delay the opening time of the braking resistor and reduce the output energy of the substation in some cases,so as to achieve certain energy saving effect.In this paper,the typical cases of single energy storage device and multi energy storage device model are selected for analysis,and the simulation verification is also carried out according to the actual line data of Batong line.Taking the minimum output energy of the substation as the optimization objective,the particle swarm optimization algorithm with adaptive weight is used to optimize several control parameters of the proposed strategy offline to achieve the best energy-saving effect.Finally,this paper uses the power hardware-in-the-loop experimental platform to verify the energy management strategy based on BP neural network energy prediction.With the help of RT-LAB real-time simulator,the platform realizes the hardware-in-the-loop simulation of the complex system of multiple traction substations and multiple trains.By comparing the improved strategy proposed in this paper with the traditional dual loop energy management strategy,the feasibility and effectiveness of the improved strategy are proved.