Remaining useful life prediction of lithium battery using convolutional neural network with optimized parameters

Dongdong Li, Lin Yang

20202020 5th Asia Conference on Power and Electrical Engineering (ACPEE)24 citationsDOI

Abstract

Accurately predicting remaining useful life (RUL) of lithium battery with nonlinear characters is essential for ensuring safety of applications. However, the diverse aging mechanism gives the challenge for present technologies. In this paper, a convolutional neural network (CNN) model is constructed for RUL prediction of lithium battery. For reducing the training time of CNN model, the orthogonal method is applied for optimizing model parameters. Then, the proposed is validated by a large dataset. And the accuracy of RUL prediction exceeds 90.9 percent while root mean square error and mean absolute error are limited to 35.1 and 13.7, respectively. The proposed method is suitable for RUL prediction of lithium battery applied in electric vehicles and energy storage devices.

Topics & Concepts

Battery (electricity)Convolutional neural networkComputer scienceArtificial neural networkMean squared prediction errorMean squared errorLithium batteryLithium (medication)Mean absolute errorNonlinear systemReliability engineeringArtificial intelligenceAlgorithmEngineeringStatisticsMathematicsMedicineIonIonic bondingPhysicsQuantum mechanicsEndocrinologyPower (physics)Advanced Battery Technologies ResearchReliability and Maintenance OptimizationAdvancements in Battery Materials