Another setting considers , which is a multi-class classification task grouping batteries into lifetime. Given a training dataset , the goal of modeling is to learn the nonlinear mapping from the early-cycle raw battery data to the battery lifetime group, which is expressed in (1). (1)
The developed method is effective and robust to different battery types. The battery quality classification accuracy can reach 96.6% based on data of first 20 cycles. Lithium-ion batteries (LIBs) are currently the primary energy storage devices for modern electric vehicles (EVs).
The new Battery Installation Standard (MIS 3012) outlines the requirements for MCS certified installers who supply, design, and install electrical energy storage or battery systems. It covers installations up to 50kW and Electrical Energy Storage Systems (EESS) classes 1 – 4.
Average results of 20 splits are listed in Table 8. As shown in Tables 8 and in the multi-class battery classification task, the proposed RLR model still presents the best performance. The four metrics are all higher than considered benchmarks, which are 87.6%, 70.8%, 73.4%, and 72.1%, respectively.
Battery storage systems come in numerous forms, so for the purpose of this new standard MCS has adopted a classification system aligned with the four EESS classes: Class 1 – all the components in the same enclosure, or multiple enclosures from the same manufacturer but with no visible direct current (DC) cable.
Finally, an RLR model integrating battery nominal and operational parameters was developed to classify battery into different lifetime groups. Computational studies were conducted on datasets containing LIBs of three different chemistries and tested under multiple conditions.