Research On Hierarchical Control Strategy And Application Of VRB Energy Storage System

The energy storage system of vanadium redox flow battery(VRB),an important way of energy storage,plays an important role in wind and photovoltaic power generation systems with its advantages such as flexibility,fast response,and suitability for large-scale applications.Nowadays,the research of VRB energy storage system is still in the stage of basic theory and demonstration.There are some key technical problems in the practical application of VRB energy storage system,such as accurate estimation of battery SOC,detection and control issues during battery operation under false data attacks,precise allocation of battery power,and battery power tracking and control issues.In view of above mentioned key technical problems,the three-layer architecture hierarchical control strategy was put forward in this paper for VRB energy storage system,including cyber control layer,power allocation layer,and ground control layer.The research was carried out mainly from several aspects as follows:(1)The state space model and the power control transfer function were determined by establishing the VRB equivalent circuit model.For the issue of being difficult to accurately estimate the SOC of VRB energy storage systems,the SOC estimation method for whole vanadium flow batteries HCOAG algorithm-based optimized KELM was proposed.In addition,the improved COA(ICOA)and simplified GWO(SGWO)algorithm are fused by sinusoidal crossing strategy to form HCOAG algorithm which is used to optimize the parameters of KELM model.Benchmark functions were used to test the HCOAG algorithm and compare its optimization ability with other smart algorithms.The accuracy and feasibility of this estimation method were verified through simulation and experiments on the CEC-VRB-5k W battery model.The results showed that,the estimation accuracy of the HCOAG-KELM method proposed in this paper is superior to that of algorithm models including the GWO-KELM,ICOA-KELM,KELM,EKF,and UKF.At the same time,the estimation error is within 2%,which meets the actual demand.(2)In the cyber layer of VRB energy storage system,aiming at the problem of false data injection attacks(FDIAs)in battery SOC estimation,a false data injection attack vector was constructed.The abnormal data detection method by combining long short-term memory(LSTM)and generative adversarial network(GAN),a control strategy based on adaptive fault-tolerant observer were proposed in this study.By training the recurrent network made up of LSTM network and GAN network,the LSTM neural network is embedded into the GAN framework as a generator and discriminator for the analysis on battery timing data.At the same time,abnormal losses are obtained by judging the discriminative loss error in the network and the reconstruction residual in the generative network for comprehensive judgment.Then,FDIAs with different strengths were constructed for experiments to verify the accuracy and feasibility of the detection method and control strategy.The results showed that the proposed method has higher detection accuracy,and the adaptive fault-tolerant observer compared with the RNN,RF,Auto-Encoder,and LTSM detection methods.The adaptive fault-tolerant observer is able to resist FDIAs with small perturbations.(3)In the power allocation layer of VRB energy storage system,aiming at the problem of reasonable power allocation of battery units,a power optimization scheduling model was established with the objective function of the lowest total cost and average loss rate of energy storage system and the best balance degree of charged state.Then,the adaptive adjustment of weights and simulated annealing based whale optimization algorithm(A-SA-WOA)was proposed.The power control effect of VRB energy storage system under the condition of satisfying the total objective function was analyzed by an example.According to the results,the optimized allocation strategy effectively reduces the operating cost and loss rate of battery cells,the charging and discharging times of the battery,and has better consistency of the charged state.(4)In the ground Layer control layer of VRB energy storage system,in response to the problems of time delay,low accuracy of the power controller,the power tracking control strategy based on deep deterministic strategy gradient and fuzzy PID was propose in this study.A composite controller composed of fuzzy PID and DDPG algorithm was designed with fuzzy PID as the main controller for the control on the power loop and DDPG as the auxiliary controller to compensate for power tracking error.Then,the sparrow search algorithm was used to optimize the PID parameters and fuzzy rules,and the output response of the optimized system was tested by step signals.The above proposed controller was applied to three different scenarios to verify the effectiveness and accuracy of the control strategy.The results show that the proposed control strategy can quickly track the power command value when the battery is charging and discharging.In during real-time tracking,the deviation between the tracking power value and the scheduling instruction value is less than ± 2%.When perturbed,the power deviation can be tracked accurately to meet the practical requirements.(5)Finally,the key technologies of battery management system,such as hardware platform and software architecture of on-site control,communication mode and energy management system in VRB energy storage system were summarized.The strategy of VRB energy storage system to smooth wind power fluctuation was designed.It was verified that VRB energy storage system has a good smooth effect on power fluctuation in wind power application scenarios through simulation.There are 89 figures,15 tables,and 160 references in this paper.