This study proposes a novel predictive energy management strategy to integrate the battery energy storage (BES) degradation cost into the BES scheduling problem and address the uncertainty in the energy management problem. As the first step, the factors affecting the BES calendar aging and cycle aging are linearly modelled.
The SOH and RUL prediction methods studied in this paper are based on individual lithium-ion battery. However, in the actual scenario, lithium battery equipment systems usually use battery pack, which individual batteries were in series and parallel.
Expertise, combined with statistical methods, will likely be more effective in forecasting battery safety performance. These statistical features, illustrated in Fig. 6, offer accurate calculations based on deviations and outliers of pack-level cell behavior. This can highlight potential failures from seemingly minor details.
Early and precise prediction of voltage anomalies during the operation of energy storage stations is crucial to prevent the occurrence of voltage-related faults, as these anomalies often indicate the possibility of more serious issues.
The data is collected by searching on the “Web of Science” database with the keywords “machine learning” + “energy storage material” + “prediction” and “discovery” as key words, respectively. The earliest application of ML in energy storage materials and rechargeable batteries was the prediction of battery states.
This combined system, due to its streamlined implementation, offers flexibility when confronting real-world challenges with noisy data. Considering the potential stakes caused by overcharging or over-discharging abuse, accurate prediction of battery SOC is indispensable for battery monitoring and management.