A Multioutput Convolved Gaussian Process for Capacity Forecasting of Li-Ion Battery Cells
Abdallah Chehade, Ala A. Hussein
Abstract
A latent function decomposition method is proposed for forecasting the capacity of lithium-ion battery cells. The method uses the multioutput convolved Gaussian process (MCGP), a machine learning framework for multitask and transfer learning. The MCGP decomposes the available capacity trends from multiple battery cells into latent functions. The latent functions are then convolved with optimized kernel smoothers to reconstruct and forecast the capacity trends. The latent functions capture nontrivial cross correlations between the capacity trends of the available battery cells. The MCGP also provides uncertainty quantification for its predictions. These two merits make the proposed MCGP a very reliable and practical solution for applications that use battery cell packs. The MCGP is derived and compared to benchmark methods on two experimental lithium-ion battery cells datasets. The results show the effectiveness of the proposed MCGP for long-term capacity forecasting.