Multi-period planning of networked integrated hydrogen-based microgrids: A chance-constrained information-gap decision approach

In the context of global decarbonization and rising renewable penetration, integrated hydrogen-based microgrids (IHMGs) are pivotal for inter-seasonal energy balancing and cross-sectoral flexibility, yet large-scale deployment is hindered by deep non-stochastic structural uncertainty in demand evolution. Existing approaches either risk technological lock-in via static paradigms or conflate stochastic variability with structural demand shifts, leaving adaptive capacity under fixed budgets unquantified. This paper proposes a multi-period planning framework for networked IHMGs (N-IHMGs) that explicitly maximizes the adaptive robustness horizonα—the maximum admissible demand growth under budget Λ0. A chance-constrained programming–information-gap decision theory (CCP–IGDT) architecture decouples heterogeneous uncertainties: CCP enforces probabilistic reliability under short-term fluctuations, while IGDT immunizes long-term investments against non-stochastic structural demand shifts without distributional assumptions—with α serving dual roles as the strategic objective and stress parameter scaling worst-case demand by (1+α). To solve the resulting large-scale MISOCP problem, an adaptive predictive-correction Benders decomposition (APC-BD) algorithm is proposed, featuring unified relaxed subproblems and cost-weighted trust-region stabilization guided by predictive fidelity to eliminate dual-extraction barriers and SOCP-induced oscillatory convergence. Case studies on the IEEE 33-bus system show that the framework reduces costs by 13.46%–18.19% over static planning, reaching 40.6% under high-growth scenarios. Relaxing CCP risk tolerance enhances the robustness horizon by 17.3%, quantifying the reliability–robustness trade-off. APC-BD achieves 0.30%–4.18% cost improvements over standard Benders methods within a 0.5% optimality gap.