Withthe widespread deployment of intelligent terminals, mobile payment platforms, and Internet of Things devices, security systems are being progressively transformed from traditional transaction outcome analysis toward an intelligent perception paradigm centered on user behavior, device states, and environmental context. To address the challenges of multimodal data heterogeneity, non-independent and identically distributed data across nodes, and the difficulty of centralized modeling under privacy constraints in distributed scenarios, an artificial intelligence-driven federated multimodal security perception framework, namely FMS-LLM, is proposed. At its core, the framework introduces a Non-IID adaptive federated fusion mechanism that achieves dual-level alignment—structural alignment via parameter-level masks and semantic alignment via feature consistency constraints—to effectively mitigate cross-node distribution discrepancies. Additionally, an LLM-driven semantic enhancement module is developed, utilizing trend-guided token selection and inertia-suppression to map low-level sensing features into high-level risk semantic representations, thereby supporting logical reasoning and explainable decision-making. This framework takes user behavioral sensing data, device state information, environmental context data, and transaction behavior data as inputs, and constructs an integrated security analysis pipeline of “perception–collaboration–reasoning”. Experimental results on the distributed multimodal security perception task demonstrate that the proposed method achieves an Accuracy of 91.62%, a Precision of 91.04%, a Recall of 90.37%, an F1-score of 90.70%, and a ROC-AUC of 94.73%, consistently outperforming baseline methods including Logistic Regression, Random Forest, LSTM, the centralized multimodal deep model, FedAvg, FedProx, and MOON. Under strongly Non-IID conditions, when $\alpha =0.1$, the model still maintains an Accuracy of 88.47% and an F1-score of 87.11%, demonstrating stronger cross-node robustness. The ablation study further indicates that the complete model attains the best classification performance while reducing communication cost to 18.92 MB/Round. These results demonstrate that the proposed method can effectively fuse multi-source sensing information under privacy-preserving conditions and support intelligent security perception tasks with higher accuracy, stronger robustness, and improved interpretability.

In the context of increasingly multi-source and complex online financial information dissemination, traditional single-modality risk detection methods are insufficient for effectively identifying highly concealed and coordinated financial risk behaviors. To address this challenge, a multi-granularity joint financial risk detection framework is proposed, which integrates semantic information content, entity behavioral patterns, and group relational structures. The framework is constructed through an information content semantic risk modeling module, an entity behavior consistency modeling module, and a group coordinated manipulation modeling module, while a hierarchical consistency constraint mechanism is introduced to achieve unified representation and collaborative inference of multi-source risk signals. In this way, the capability of identifying complex financial risks is significantly enhanced. Experimental results on real-world multi-platform financial datasets demonstrate that the proposed method consistently outperforms existing baseline models across all evaluation metrics. Specifically, a Precision of $0.9176$, Recall of $0.8943$, F1-score of $0.9058$, and area under curve (AUC) of $0.9524$ are achieved. Compared with the strongest baseline model, FinBERT, improvements of approximately $3.12\%$, $4.24\%$, $3.70\%$, and $3.08\%$ are obtained, respectively. Further multi-granularity comparison experiments indicate that the full-granularity coupling model improves the F1-score by $3.70\%$ over the best single-granularity setting and by approximately $1.5\%$ over the best dual-granularity configuration, thereby validating the effectiveness of multi-granularity information fusion. In addition, ablation studies confirm that each core module and key mechanism contributes significantly to the overall performance. By modeling financial risks from a multi-dimensional sensing perspective, the proposed method not only improves detection accuracy and robustness, but also provides a novel and practical solution for intelligent risk perception and decision-making in complex financial environments.

With protected agriculture increasingly developing toward large-scale, multi-zone, and intelligent production, reducing water and fertilizer consumption, drainage losses, and energy waste while maintaining crop yield, product quality, and long-term economic profitability has become an important challenge for sustainable greenhouse production. To address sensor abnormalities, spatial heterogeneity among irrigation zones, competition for shared hydraulic resources, and the limited ability of conventional control strategies to jointly optimize resource efficiency and economic performance, a sensor-driven spatiotemporal graph multi-agent reinforcement learning framework, termed SensorGraph-FertiMARL, is proposed for state sensing, response prediction, and sustainable coordinated fertigation optimization in multi-zone greenhouses. Reliable zone-level states are first obtained through sensor-reliability gating, missing-state reconstruction, and multi-relational graph fusion incorporating spatial, hydraulic, and microclimatic relationships. An action-conditioned fertigation–yield–quality–profit digital twin is then constructed to jointly predict root-zone responses, resource consumption, marketable yield, product quality, and net profit associated with candidate control actions. Finally, graph-based multi-agent proximal policy optimization (Graph-MAPPO) with capacity, safety, and fairness constraints is employed to coordinate fertigation allocation among multiple zones under shared pump, main-pipeline, and nutrient-solution resource constraints. Evaluated through rigorous offline digital-twin simulation and replay testing calibrated against complete multi-stage historical dataset records from an eight-zone commercial greenhouse, experimental results show that the proposed method reduced water and fertilizer use to $75.8\%$ and $78.3\%$ (achieving water and fertilizer saving rates of $24.2\%$ and $21.7\%$), respectively, relative to the uniform timer-based control baseline, while decreasing the drainage rate to $7.2\%$. The Grade-A product rate increased to $84.6\%$, net profit reached $138.9\%$ of the baseline level (a $+38.9\%$ net-profit improvement), return on investment (ROI) reached $1.94$, and the stress-violation rate was limited to $2.9\%$. For state sensing and digital-twin prediction, the sensor-state reconstruction mean absolute error (MAE) and coefficient of determination ($R^2$) were $0.036$ and $0.918$, respectively; the water-content prediction MAE was $0.041$, the electrical conductivity ($EC$) prediction root mean squared error (RMSE) was $0.119$, and the drainage prediction mean absolute percentage error (MAPE) was $8.7\%$. In addition, the $R^2$ for marketable-yield prediction reached $0.912$, the macro-averaged F1 (Macro-F1) score for quality-grade prediction reached $0.851$, the $R^2$ for net-profit prediction reached $0.904$, the mean prediction interval width (MPIW) was $0.142$, and the prediction interval coverage probability (PICP) reached $0.931$. These results demonstrate that SensorGraph-FertiMARL can simultaneously achieve water and fertilizer conservation, drainage reduction, operational-risk control, and economic improvement while maintaining marketable production and quality, thereby providing an effective intelligent optimization approach for resource-efficient, environmentally sustainable, and economically viable greenhouse production.

Banking data centers generate heterogeneous IT operational streams from servers, network devices, edge nodes, and business terminals. Existing detectors are vulnerable to asynchronous sampling and unreliable channels, while separate root-cause and language-model modules cannot preserve an end-to-end evidence trail. This study presents a cloud–edge–end framework that couples sensor-quality-aware temporal encoding, topology- and chronology-constrained event-graph reasoning, and evidence-constrained language generation. At the edge, dynamic reliability scores suppress missing, noisy, delayed, drifting, and frozen channels. At the cloud layer, the model ranks root causes and identifies propagation relations and onset times from device topology, logs, alarms, and maintenance records. The language model then generates traceable incident interpretations from the resulting evidence package and rejects claims when support is insufficient or conflicting. Experiments use IT operational data collected from the headquarters and disaster-recovery data centers of a commercial bank between 2024 and 2025; the dataset contains no customer financial or transaction records. The proposed method achieves $95.38\%$ Accuracy, $95.06\%$ Precision, $94.71\%$ Recall, a $94.88\%$ F1-score, and a $97.41\%$ AUC for hardware anomaly detection. Root-cause localization accuracy, Top-3 Accuracy, event-relation F1-score, and temporal localization error are $91.84\%$, $97.26\%$, $90.73\%$, and $1.82$ s, respectively. The complete model attains a $94.68\%$ evidence support rate and a $4.26\%$ unsupported statement rate. These results show that reliability is propagated from sensing through system-level diagnosis to evidence-grounded interpretation.