The accurate prediction of autonomous vessel CO2 emissions is critical for achieving IMO 2050 carbon neutrality and optimizing low-carbon maritime operations. Traditional models face limitations in real-time multi-source data analysis and dynamic cross-variable dependency modeling, hindering data-driven decision-making for sustainable autonomous shipping. This study proposes a Multi-scale Channel-aligned Transformer (MCAT) model, integrated with a 5G–satellite–IoT communication architecture, to address these challenges. The MCAT model employs multi-scale token reconstruction and a dual-level attention mechanism, effectively capturing spatiotemporal dependencies in heterogeneous data streams (AIS, sensors, weather) while suppressing high-frequency noise. To enable seamless data collaboration, a hybrid transmission framework combining satellite (Inmarsat/Iridium), 5G URLLC slicing, and industrial Ethernet is designed, achieving ultra-low latency (10 ms) and nanosecond-level synchronization via IEEE 1588v2. Validated on a 22-dimensional real autonomous vessel dataset, MCAT reduces prediction errors by 12.5% MAE and 24% MSE compared to state-of-the-art methods, demonstrating superior robustness under noisy scenarios. Furthermore, the proposed architecture supports smart autonomous shipping solutions by providing demonstrably interpretable emission insights through its dual-level attention mechanism (visualized via attention maps) for route optimization, fuel efficiency enhancement, and compliance with CII regulations. This research bridges AI-driven predictive analytics with green autonomous shipping technologies, offering a scalable framework for digitalized and sustainable maritime operations.
Family structure has long been regarded as an important determinant of household saving, yet the empirical evidence for developing economies remains limited. Using the 2018–2022 panels of the China Family Panel Studies (CFPS), a nationwide survey that follows 16,519 households across three waves, the present study investigates how family size, the elderly share, and the child share jointly shape saving behavior. A household fixed effects framework is employed to control for time-invariant heterogeneity, followed by a sequential endogeneity strategy: external-shock instruments are tested and rejected, lagged two-stage least squares implement internal instruments, and a dynamic System-GMM model is estimated to capture saving persistence. Robustness checks include province-by-year fixed effects, inverse probability weighting for attrition, balanced-panel replication, alternative variable definitions, lag structures, and sample filters. Family size raises the saving rate by 4.6 percentage points in the preferred dynamic specification (p < 0.01). The elderly ratio remains insignificant throughout, whereas the child ratio exerts a negative but model-sensitive association. A three-path mediation analysis indicates that approximately 26 percent of the total family size effect operates through scale economy savings on quasi-fixed expenses, 19 percent is offset by resource dilution pressure, and less than 1 percent flows through a precautionary saving channel linked to income volatility. These findings extend the resource dilution literature by quantifying the relative strength of competing mechanisms in a middle-income context and showing that cost-sharing economies dominate child-related dilution for most households. Policy discussion highlights the importance of public childcare subsidies and targeted credit access for rural parents, whose saving capacity is the most constrained by additional children. The study also demonstrates that fixed effects estimates of family structure can be upward-biased unless dynamic saving behavior and internal instruments are considered.