Research And Implementation Of Time-Efficient State Evaluation System For Echelon Utilization Battery

With the rapid development of the new energy vehicle industry,the disposal problem of retired lithium-ion batteries,also known as echelon utilization batteries,is becoming increasingly serious.Currently,there are attempts within the industry to apply echelon utilization batteries to communication base station backup power storage scenarios.Compared to new batteries,the stability of echelon utilization batteries is poorer,and there is greater variation in discharge properties among different echelon utilization batteries.Therefore,when evaluating the various indicators of echelon utilization batteries for reuse,there is often a higher requirement for timeliness.To solve the reuse problem of echelon utilization batteries,this thesis designs a time-efficient echelon utilization battery state evaluation system,which guarantees time-efficient from two aspects of system architecture design and state estimation method.In terms of system architecture design,based on edge computing technology and cloud computing technology,the system builds an echelon utilization battery status evaluation system.The edge gateway of the system has storage and computing capabilities,which can avoid cloud-edge communication delays and ensure the timeliness of data collection,data analysis,status evaluation,and battery control.In terms of state estimation method,to estimate the SOC of echelon utilization batteries more accurately,this thesis proposes an echelon utilization battery SOC estimation method based on transfer learning.This method includes an echelon utilization battery SOC estimation framework based on transfer learning and a gate recurrent unit with attention mechanism.It can reduce the required amount of data and computing power for model training,and greatly reduce the time required for training data accumulation.This method can be directly applied to the edge gateway to quickly complete the training of the echelon utilization battery SOC estimation model.Compared with the method of transfer LSTM and GRU models using the ordinary fine-tuning transfer learning method,the echelon utilization battery SOC estimation method based on transfer learning proposed in this thesis requires a smaller training data scale and lower computing power during training.At the same time,the training time used by the method is shorter and the SOC estimation result is more accurate.This thesis first introduces the research background and related technologies of the time-efficient echelon utilization battery state evaluation system.Next,based on the survey of various battery evaluation management systems in the industry and the characteristics of echelon utilization batteries,the demands of the echelon utilization battery status evaluation system are analyzed.Then,an echelon utilization battery SOC estimation method based on transfer learning is proposed.After that,the design of the time-efficient echelon utilization battery state evaluation system is introduced in detail,including the overall architecture design and component design of the central management platform and edge gateway.Then the system realization is described.Finally,the usability of the system is verified through a series of functional tests.