This study proposes a weighted ensemble deep learning framework for accurately predicting the State of Health (SOH) of lithium-ion batteries. Three distinct model architectures—CNN-LSTM, Transformer-LSTM, and CEEMDAN-BiGRU—are combined using a normalized inverse RMSE-based weighting scheme to enhance predictive performance. Unlike conventional approaches using fixed hyperparameter settings, this study employs Bayesian Optimization via Optuna to automatically tune key hyperparameters such as time steps (range: 10-35) and hidden units (range: 32-128). To ensure robustness and reproducibility, ten independent runs were conducted with different random seeds. Experimental evaluations were performed using the NASA Ames B0047 cell discharge dataset. The ensemble model achieved an average RMSE of 0.01381 with a standard deviation of ±0.00190, outperforming the best single model (CEEMDAN-BiGRU, average RMSE: 0.01487) in both accuracy and stability. Additionally, the ensemble's average inference time of 3.83 seconds demonstrates its practical feasibility for real-time Battery Management System (BMS) integration. The proposed framework effectively leverages complementary model characteristics and automated optimization strategies to provide accurate and stable SOH predictions for lithium-ion batteries.

A choice in temporal discounting is often made on one preferred option between a sooner but smaller reward and a later but larger reward. In temporal discounting, a consumer undergoes an evaluation process concerning the trade-offs between time delay (i.e., a choice between sooner and later receipts) and the amount of reward (i.e., a choice between smaller and larger rewards). It has been evidenced that a surprisingly large number of consumers is inclined to choose a sooner option with a lower economic value. Those consumers tend to apply an excessive temporal discounting rate to the time delay and as a result, the sooner option with lower economic value stands out to be more attractive. However, there is another group of consumers who deliberately promote the long-term perspective in the evaluation process. Those consumers show a high tendency of choosing a later option with greater economic value. A key factor affecting the evaluation process is the sensitivity to time delay a consumer entails. The sensitivity to time delay could systematically affect the representations of hedonic pleasure being projected at different time delays and is the basis of individual differences in hedonic orientation. An anhedonia, a person with the ability to control physical pleasure, is likely to be less sensitive to time delay and apply a lower discounting rate for delayed rewards because he processes the presently available pleasure less intensively than do ordinary consumers. Thus, anhedonic consumers are more likely to choose a delay reward in temporal discounting choices. In a famous fable by Aesop, “The Ant and the Grasshopper” suggest a possible linkage between hedonic orientation and temporal discounting tendency. The story goes that the Grasshopper enjoys whatever is on its hands and does not give much thought to what is going to happen in the future, while the Ant takes on the hardship denying the current pleasure as he plans for the difficult winter time. We conducted an experiment to investigate the relationships among individual differences in hedonic orientation, and the behavioral choices in temporal discounting trials by examining the EEG theta (4-8 Hz) correlates. A hedonic orientation questionnaire was developed using twenty selected items from Physical Anhedonia Scale, Barratt Impulsiveness Scale and Impulsive Buying Scale. Using the questionnaire, the hedonic orientation of fifteen participants was analyzed. Among them, five participants were classified into anhedonia (Ant-like, AN) group and ten participants were into hedonic pleasure dependent (Grasshopper-like, GH) group. Participants conducted ten temporal discounting trials and, in each trial, a choice was made between two options, either a smaller reward today (the “now” option) or a greater reward to be received after a specified time delay (the “future-oriented” option). The absolute theta activations while viewing a future-oriented option in each trial were analyzed as the neural correlates of temporal discounting. The results indicated that the hedonic orientation of participants had a high prediction power of not only behavioral choices in temporal discounting but also EEG theta activations in the frontal brain region. In analyzing temporal discounting tendencies, the discounting rate of AN group was found to be lower than that of GH group. In addition, EEG theta analysis shows two interesting activation patterns in the frontal region. First, AN group showed higher frontal theta activations than GH group indicating its deliberate engagement to future-oriented options. Second, theta activations in the Dorsolateral Prefrontal Cortex (DLPFC), more specifically AF3 and AFz, were negatively correlated with the discounting rate of participants. Neuroscience researches often refer that DLPFC is associated with deliberate and analytic information processing. Therefore, the high theta activations in DLPFC could be considered as a distinctive neural indicator of high anhedonia traits and low temporal impulsivity.