• Hybrid forecasting integrates LSTM, GRU and statistical models for industrial energy demand. • Probabilistic forecasts propagate uncertainty into multi-objective optimisation. • NSGA-II reveals uncertainty-dependent cost–carbon Pareto trade-offs. • Synthetic UK industrial scenario ensures transparency and reproducibility. • Methodology supports Net Zero analysis without deployment or compliance claims. This paper presents a hybrid AI-based methodological framework that couples time-series energy forecasting with uncertainty-aware multi-objective optimisation for industrial decarbonisation analysis in the United Kingdom context. The framework integrates Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and decomposable statistical forecasting models with the Non-dominated Sorting Genetic Algorithm II (NSGA-II) to generate Pareto-optimal trade-offs between energy cost and carbon-related objectives. The approach incorporates probabilistic forecasting to explicitly characterise demand uncertainty and propagate it into multi-objective optimisation, enabling structured exploration of cost–carbon trade-offs under uncertainty. To ensure reproducibility and avoid claims of operational deployment or industrial compliance, the framework is evaluated using a synthetic, statistically calibrated UK manufacturing-type scenario. Results are reported as methodological evidence, including probabilistic forecasting behaviour and Pareto-front analysis, illustrating how uncertainty-aware forecasting can inform optimisation. The contribution is methodological, reproducible, and policy-aware, aligning with the UK Net Zero agenda without implying real-world implementation or proprietary data access.
ASTAN et al. (Mon,) studied this question.