State of Charge Estimation for Commercial Li-Ion Battery Based on Simultaneously Strain and Temperature Monitoring Over Optical Fiber Sensors

The combination of artificial intelligence methods and multisensory is crucial for future intelligent battery management systems (BMS). Among multi-sensing technologies in batteries, simultaneously monitoring the strain and temperature is essential to determine the batteries' safety and state of charge (SoC). However, the combination still faces a few challenges, such as obtaining multi-sensing parameters with only one simple and easy-to-fabricate sensor, and how to use artificial intelligence and measurement parameters such as strain and temperature for effective modeling. To address these, we propose a novel sensing technique based on a compact dual-diameter fiber Bragg gratings (FBGs) sensor capable of being attached to the surface of a working lithium-ion pouch cell to simultaneously monitor the battery's surface strain and temperature. Then, based on the collected data of strain and temperature, we have constructed deep artificial neural network (DNN) models with different inputs to realize accurate battery SoC estimation with high resistance to electromagnetic interference. Based on our DNN models, the experimental results show that strain and temperature information can be used as supplementary parameters for improved SoC estimation (accuracy increased from 97.40% to 99.94%). Meanwhile, we also find that by just using the strain and temperature information obtained by the optical fiber sensor, the SoC estimation can be achieved without the voltage and current inputs. This new optical fiber measurement tool will provide crucial additional capabilities to battery sensing methods, especially for the future intelligent BMS.