End-to-End Trustworthy Traceability and Predictive Optimization for WEEE Recycling via Consortium Blockchain

Efficient and trustworthy traceability across the Waste Electrical and Electronic Equipment (WEEE) recycling chain remains limited by fragmented crossorganizational data, heterogeneous event recording, and the absence of mechanisms that connect provenance assurance with real-time blockchain performance control. Existing blockchain-based traceability studies largely address linear flows and static system configurations, leaving a gap in supporting WEEE's many-to-many batch transformations and in adapting ledger performance under fluctuating industrial-IoT workloads. To address these issues, this study proposes an integrated framework that jointly enables end-to-end trusted traceability and predictive, workload-aware blockchain optimization. The first component is a consortium-chain traceability architecture covering collection & logistics, dismantling, and reprocessing, where evidentiary events are immutably anchored through hash checkpoints and batch-to-batch linkages preserve full provenance continuity. The second component is a predictive tuning mechanism that couples an edge-deployed Informer model for short-term transaction-arrival forecasting with a cloud-side multi-agent reinforcement learning (MADDPG) policy that proactively adjusts key Fabric parameters, including block size and interval. Implemented on Hyperledger Fabric 2.2 and evaluated using PeMSD8, WUSTL-IoT-2021, ALPI, and Caliper benchmarks, the framework demonstrates that the Informer forecaster surpasses LSTM, ARIMA, GAN, and GridTST across MSE/MAE/RMSE metrics, while RL-based tuning increases throughput by up to 58.6% and reduces latency by 34.1% over default settings. Integrating prediction with RL yields an additional 8.2% throughput gain and 18.8% latency reduction. The results confirm that rigorous provenance assurance and adaptively optimized ledger operation can be jointly realized for scalable, industry-grade WEEE recycling systems.