Predicting Life of Electric Vehicle and Load-Leveling Lead Acid Batteries from Initial Acceptance Test Data by Use of Pattern Recognition Analysis.
Abstract
A novel approach to battery lifetime prediction was applied to life-cycling data for 108 ESB EV-106 6-V golf cart batteries (tests conducted by TRW for NASA-Lewis). Computerized pattern recognition methods were used to examine initial cycling measurements and to classify individual batteries into long-lived or short-lived classes with greater than 85% accuracy. Results of this study were used to design a fabrication and test program for 340 GNB lead-acid cells in a 500 kWh load-leveling battery to be tested at the Battery Energy Storage Test Facility. Preliminary observations on initial test data are reported here. Keywords: Multivariate analysis.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1984
- Accession Number
- ADA168853
Entities
People
- S. P. Perone
- W. C. Spindler
Organizations
- Lawrence Livermore National Laboratory