Reliable remaining useful life (RUL) estimation is important for the safe and efficient use of lithium-ion (Li-ion) batteries in electric vehicles (EVs) and energy-storage systems. Most data-driven RUL models are trained under controlled laboratory conditions, but their performance can weaken during field operation, where usage, sensing quality, and degradation paths are less predictable. This study proposes a physics-informed laboratory-to-field (L2F) deep learning framework for battery RUL prediction under limited or unavailable target labels. The framework combines three components: a six-channel laboratory cycle representation comprising voltage, current, temperature, cumulative charge throughput, cumulative energy throughput, and voltage derivative; a gated Transformer–Temporal Convolutional Network (TCN) Fusion backbone for modeling long-range and local degradation patterns; and a staged adaptation policy based on paired-view consistency and covariance alignment. The Fusion backbone achieved the lowest held-out XJTU laboratory root mean square error (RMSE) of 46.67, compared with 48.27 for TCN and 48.85 for the Transformer. In the Tsinghua University (Tsinghua) deployment experiment, measured target RUL labels were unavailable after preprocessing and window construction. Therefore, the direct field-side mean absolute error (MAE), RMSE, and coefficient of determination were not computed. The Tsinghua results are interpreted as an unlabeled deployment-credibility and trajectory-regularity assessment, showing operational continuity, finite vehicle-specific predicted trajectories, and reduced local trajectory volatility after staged adaptation. The S2a + S2b policy reduced Fusion RUL volatility from 8.376 to 0.632. In the XJTU laboratory source representation, Integrated Gradients showed that physics-aware channels contributed 30.98% of the attribution mass, increasing from 19.50% in early-life windows to 31.86% in late-life windows. These attributions explain the laboratory six-channel waveform model and are not used as direct evidence of Tsinghua field-feature importance.
Wildlife conservation is the practice of protecting wild plant and animal species. Wildlife plays a vital role in balancing the ecosystem and stabilizing different natural processes. In order to enable real-time tracking and monitoring of animal movements, different tracking devices can be used. A collaboration work with a telecommunication industrial partner in senior capstone design projects considers enhancing animal tracking devices. In this work, IoT-based tracking device is designed to be powered from animal's motion by using build-in harvesting unit. The device will be used to monitor and track animal's motions as well as biological signals such as temperature, pressure and heart rates in order to observe and study animal species in their natural habitat.
Accurate estimation of state-of-charge (SoC) and state-of-health (SoH) is important for lithium-ion battery operation. This study investigates an actor–critic formulation of joint SoC/SoH estimation using TD3 and SAC with a composite reward and compares the observed results with four supervised baselines. On the supplied XJTU run, the actor–critic models achieved lower observed SoC RMSE than the supervised baselines, while the best supervised model remained competitive or superior for SoH. The current evaluation uses a limited cell cohort and three random seeds, and the per-cell normalization and the OCV calibration used the records of held-out cells (the protocol is not training-only); therefore, the results are exploratory and do not establish confirmatory statistical superiority or generalization to a larger battery population. The planned Tsinghua label-retention experiment could not be completed because the supplied capacity field had no usable variance. The resulting Tsinghua analysis is limited to label-free output diagnostics and does not establish target-domain estimation accuracy or validate reward-model transfer under label scarcity. The reward terms, including a coulomb-counting term applied to SoH that is provisional and not a validated physical constraint, are exploratory, and any mechanistic explanation of the SoC–SoH difference is a hypothesis rather than an established result. The reward-component ablation, the hyperparameter-sensitivity analysis, the matched non-RL comparison, and the Tsinghua label-retention sweep remain future validation requirements.