Screening Method of Liquid Metal Batteries Based on Multi-feature Extracted From Discharging Curve and Combined Cluster Algorithm
Efficient battery screening technology is crucial for ensuring safety and economy in battery system integration applications.To achieve consistent results for integrated liquid metal batteries(LMBs),this paper proposes a combined clustering method to effectively cluster the feature indicators obtained during the constant current discharge of the battery to quickly achieve liquid metal battery screening.First,a feature extraction framework is proposed to accurately characterize the battery discharge curve.It consists of four main stages:data acquisition,data pre-processing,feature indicator generation,and screening indicator generation.Three screening indicators are generated,i.e.the second change point voltage of the battery curve,the corresponding time,and the discharge energy of the battery before the first change point voltage.Then,a combined clustering method based on density-based noise applied spatial clustering and mean shift is proposed,and a consistency difference indicator is proposed to quantitatively evaluate the clustering effect.Finally,through the analysis of the clustering optimization effect of 212 200 Ah-level liquid metal batteries,it is shown that the proposed method significantly reduces the influence of cells with large variability on cell screening,and it can simultaneously achieve outlier detection and fast and accurate cell adaptive screening.The dynamic profile test verifies that the battery screening method in this paper can effectively improve the response consistency of the battery pack.
battery screeningliquid metal batteries(LMBs)feature extractioncombined clustering methodoutlier point processing
万方数据
维普
