Research On Energy Storage Inverter Control Of Flow Battery Based On Deep Neural Network Optimization

Under the research background of large-scale wind energy storage inverter control,the deep belief network optimized by improved genetic algorithm is applied to the research of PI numerical optimization control based on the modeling of zinc-bromide flow battery and its inverter control.With the optimization goal of improving the stability of the PCC grid-connected voltage,the inverter’s double closed-loop PI parameters are adjusted in real time according to the grid-side load and DC bus voltage changes to enhance the rapidity of the battery inverter response speed and the robustness of the control accuracy in the case of wind turbine output fluctuations and PCC load sudden changes.In order to establish the equivalent circuit model of zinc-bromine flow battery,the mathematical expressions of related variables of flow battery are derived,and the correctness of the model is verified by charging and discharging tests.In order to grasp the remaining capacity of the battery during charging and discharging,and switch the charging and discharging state in time,IGA-DBN(Improved Genetic Algorithm-Deep Belief Network)is used to estimate the state of charge of zinc-bromine flow battery,and the estimated results can accurately fit the test data.In order to carry out the simulation experiment of wind power grid-connected,the wind power grid-connected simulation model based on DFIG(Doubly-Fed Induction Generator)is established.The simulation experiment shows that when the wind speed changes suddenly,the grid-connected voltage of wind turbines will fluctuate greatly,and the voltage and current distortion rate will be large.When the wind turbines are directly connected to the grid,they will have a certain impact on the grid.Based on this,an energy storage system based on zinc-bromine flow battery is established to suppress the impact of wind turbine output fluctuations on the stability of the power grid.In order to improve the response rate of the energy storage converter,a PI control strategy optimized by improved genetic algorithm is proposed.By using the autonomous optimization ability of genetic algorithm,the suitable PI parameters under different working conditions can be obtained.In order to adapt to different working conditions,this paper proposes to train the deep belief network by using the data obtained by the improved genetic algorithm,so that the system can adjust the PI parameter value combination in real time according to the actual situation,so as to achieve the effect of rapid response of converter under the premise of taking into account the stability of the system.The training process based on the deep belief network is highly dependent on the initial value.Random selection may cause the DBN training time to increase and fall into local optimization problems.An improved genetic algorithm is proposed to iteratively optimize the initial value of DBN.The system simulation model is built to realize the function of suppressing voltage fluctuation of energy storage device,and the optimized control strategy is debugged to obtain the simulation results under sudden changes of wind speed.The experimental platform of zinc-bromine flow battery energy storage inverter is built to verify the proposed optimal control strategy.Experimental analysis shows that the research done can reduce the wind power grid-connected voltage fluctuation rate to less than 5%,and the voltage and current distortion rate can be reduced to 4% and 8% respectively(meet the national GB/T-14549-93 standard requirements),and the response period of the inverter is less than 20 ms.This work can provide reference for the serious problem of wind curtailment(wind curtailment rate 30%)that needs to be dealt with urgently.