The Smart Grid Artificial Intelligence-of-Things (SG-AIoT) uses SCADA (Supervisory Control and Data Acquisition) and AMI (Advanced Metering Infrastructure) to improve electricity distribution. However, the current SG-AIoT system faces challenges such as inaccurate energy forecasting, insecure data computation, information leakage, and data CIA (Confidentiality, Integrity, and Availability) issues. Smart-meter data has been provided to drive initiatives such as the UK’s "Low-Carbon-London" project, where the need for accurate forecasting and grid data security is felt. To address these issues, this paper proposes a secure federated learning framework that combines LSTM (Long Short-Term Memory) neural networks with AES (Advanced Encryption Standard) cryptographic blockchain across three-tier grid layers (Edge smart meters, Fog SCADA servers, and Cloud server). An AES-based blockchain framework is developed to secure data communication across all grid layers—edge, fog, and cloud. At the IoT smart meter edge nodes, a double-layered LSTM and dropout neural network is used for accurate energy demand forecasting. Forecasted data from each edge node is encrypted and aggregated at the fog layer using the AES blockchain. The proposed CBFedAggSum (Cryptographic Blockchain with Federated Learning Aggregation through Summation) method enables secure and efficient model aggregation at fog and cloud nodes. This end-to-end process strengthens grid data security and improves forecasting accuracy, effectively addressing key challenges in SG-AIoT systems. We compared our approach with six existing methods (FedAvg, FedSGD, SecAgg, 1D-CNN-GRU, PP-CE-FL, and Edge-Cloud-AIoT-FL). We found that our model outperformed them in forecasting accuracy and privacy preservation.
The use of supervisory control and data acquisition (SCADA) and advanced metering infrastructure (AMI) systems in smart grid-based Industrial Internet of Things (SG-IIoT) networks for proper energy supply is noteworthy. Inaccurate energy load forecasts, cyber threats, and energy load-based sustainability issues in smart grids hinder SG-IIoT operations. To mitigate these challenges, a federated learning approach is developed by integrating long short-term memory (LSTM), the transformer-large language model (LLM)-based foundation models, and advanced encryption standard (AES) cryptography. The proposed approach is named the responsible federated foundation model (ResFedFM). To ensure secure federated learning computation and data security at the edge (smart meter), fog (SCADA-based substation grid), and cloud (grid cloud server) layers of the SG-IIoT, a self-parent key-based cryptography method has been developed by combining AES with hash-based message authentication code (HMAC). A load forecasting algorithm called LSTM-LLM-GenResAI-Forecasting has been designed for computation at each end node of the federated learning process. The edge node forecast outputs are encrypted and aggregated at the fog node. At the fog node, the data are decrypted, and the aggregation algorithm of the federated learning process is used to generate the overall load forecasting of each sub-station grid. Again, the forecast data from these fog nodes are aggregated in an encrypted state at the cloud level, and overall load forecasts are generated for multiple fog nodes. The result of the proposed approach provides responsible forecasting (high accuracy, green computing-based energy demand, optimization of AI-hallucination, and grid data security), demonstrating enhanced performance over seven significant models.
The most broadly perceived relative directions are right, left, up, down, backward and forward. This research paper presents a new computational technique to learn human's relative directions, where one intelligent computer can learn any human's right, left, up, down, backward and forward or different relative directions. The present paper portrays models describing the essential structures of relative direction learning process between human and intelligent machine. We developed two proficient algorithms for solving this approach. In our experiment we propose Human Relative Direction Learning (HRDL) algorithm for learning human's relative directions and Human Direction Identification (HDI) algorithm for tracking any human position and identity human's relative directions from different direction points.