Journal of Resource Management and Decision Engineering In Press 2027 07 01 Multi-Objective Trust-Aware Service Selection Modeling in Dynamic Cloud Manufacturing Using Blockchain, Adaptive Learning, and Sustainability Constraints Multi-Objective Trust-Aware Service Selection Modeling in Dynamic Cloud Manufacturing Using Blockchain, Adaptive Learning, and Sustainability Constraints 1 20 EN Journal Article 2026 05 27 <p>This study aimed to develop and evaluate a multi-objective trust-aware service selection model for dynamic cloud manufacturing by integrating blockchain-based transaction verification, adaptive learning, reliability-oriented optimization, and explicit sustainability constraints. A quantitative computational modeling and simulation design was used. The simulated cloud manufacturing environment included 1,200 heterogeneous manufacturing services distributed across eight functional categories and 240 service providers, with 4,800 composite manufacturing requests generated under low-, medium-, and high-dynamic conditions. Service profiles incorporated cost, processing time, availability, reliability, transaction success, trust, behavioral consistency, energy consumption, carbon emissions, and defect rate. Blockchain-based transaction histories were used to construct tamper-resistant trust evidence, while adaptive learning continuously updated provider performance estimates. The service selection problem was formulated as a constrained multi-objective optimization problem and evaluated against conventional QoS-based, trust-aware, and trust-plus-adaptive-learning comparison models through repeated simulations, robustness analysis, sensitivity analysis, convergence assessment, scalability testing, and ablation analysis. Inferential comparisons showed significant differences among the competing models for aggregate reliability, provider trust, manufacturing request success, service failure, energy consumption, carbon emissions, and the proportion of feasible service compositions. Pairwise comparisons indicated that the complete proposed model significantly outperformed the conventional QoS-based and trust-only models across the principal trust, reliability, and sustainability outcomes. The complete model also significantly outperformed the trust-plus-adaptive-learning configuration for environmental indicators, confirming the added contribution of sustainability constraints. Ablation analyses further demonstrated significant performance deterioration when blockchain-based trust or adaptive learning was removed, whereas removing sustainability constraints significantly increased energy consumption and carbon emissions despite preserving short-term operational success.</p> https://www.journalrmde.com/index.php/jrmde/article/download/412/387