RFID–Edge–Federated Learning Decision Support for Retired Power Equipment: Recovery, Dismantling, and Recycling in the Circular Economy
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
With the continuous expansion and upgrading of global power infrastructure, a large number of power equipment units are entering the retirement phase. Traditional scrap disposal models face challenges such as information opacity, low efficiency, and insufficient resource recovery rates. This study proposes an intelligent decision support system based on Radio Frequency Identification-Edge Computing-Federated Learning fusion architecture to address circular economy decision-making for retired power equipment. The system constructs a three-layer architecture: the RFID perception layer achieves equipment lifecycle information collection through passive tags, the edge computing layer deployed at recycling sites completes localized data processing and preliminary decision-making, and the federated learning decision layer coordinates multi-site collaborative training of global models. The system innovatively integrates differential privacy protection mechanisms into the Federated Averaging (FedAvg) algorithm, transmitting only model parameters rather than raw data, effectively addressing data silos and privacy leakage issues faced by traditional methods. Experimental validation is conducted based on a real dataset of 5000 units covering transformers, circuit breakers, and switchgear. Results demonstrate that the system achieves a decision accuracy of 94.3%, improving upon traditional centralized methods by 4.6 percentage points and single edge node methods by 9.1 percentage points. System processing latency is reduced to 45 milliseconds and response time shortened to 0.8 seconds, representing a 75% reduction compared to traditional methods. Regarding communication overhead, cumulative communication volume for 100 training rounds is only 230MB, representing an 85.4% reduction compared to centralized methods. Economic benefit evaluation shows that recycling sites using this system achieve an average revenue increase of 18.5% and resource recovery rate improvement of 12.3%, with decision scheme execution success rate reaching 91.2%. This study provides a feasible technical pathway for retired power equipment circular economy management, holding significant importance for promoting intelligent transformation and sustainable development in the power industry.
Fields of Study & Subjects
Recycling and Waste Management Techniques
IoT and Edge/Fog Computing
Sustainable Supply Chain Management
Publication details
- Journal / source
- IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
- Pages
- 1097–1103
- Publication date
- 28 January 2026
- Language
- EN