Hybrid AI-based energy forecasting and multi-objective optimisation for industrial decarbonisation: A UK Net Zero case study

Industrial energy management is pivotal to achieving the UK’s Net Zero commitments. This paper presents a regulatory-aligned hybrid AI framework that combines Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models for short- and long-term energy forecasting, integrated with Non-dominated Sorting Genetic Algorithm II (NSGA-II) for multi-objective optimisation of cost and carbon emissions. The framework incorporates ISO 50001 and PAS 1878/1879 compliance constraints, ensuring alignment with the UK’s Industrial Decarbonisation Strategy. Using real-time IoT sensor data, historical energy profiles, and renewable generation inputs, the system dynamically optimises load scheduling, battery dispatch, and renewable utilisation. A UK-based manufacturing case study demonstrated approximately 21% cost savings and 20% CO2 emission reductions, verified through probabilistic forecasting and scenario analysis. The proposed approach contributes a reproducible, policy-aware methodology for AI-driven industrial energy management, supporting scalable, data-driven decarbonisation across UK industrial sectors.