Upgrading, Renovation and Engineering Practice of Core Equipment for Digital-intelligent Fuel Management of 1 000 MW Ultra‑supercritical Peak‑regulating Units
Digitalization and Intelligentization|更新时间:2026-09-10
|
Upgrading, Renovation and Engineering Practice of Core Equipment for Digital-intelligent Fuel Management of 1 000 MW Ultra‑supercritical Peak‑regulating Units
POWER EQUIPMENTVol. 40, Issue 5, Pages: 387-393(2026)
ZHANG Guoshan,MA Jun,HAO Yong,et al.Upgrading, Renovation and Engineering Practice of Core Equipment for Digital-intelligent Fuel Management of 1 000 MW Ultra‑supercritical Peak‑regulating Units[J].POWER EQUIPMENT,2026,40(05):387-393.
ZHANG Guoshan,MA Jun,HAO Yong,et al.Upgrading, Renovation and Engineering Practice of Core Equipment for Digital-intelligent Fuel Management of 1 000 MW Ultra‑supercritical Peak‑regulating Units[J].POWER EQUIPMENT,2026,40(05):387-393.DOI: 10.19806/j.cnki.fdsb.2026.05.009.
Upgrading, Renovation and Engineering Practice of Core Equipment for Digital-intelligent Fuel Management of 1 000 MW Ultra‑supercritical Peak‑regulating Units
To address the three major drawbacks of "low precision, slow response and difficult traceability" in the fuel management of coal-fired peak-regulating units under the new power system and the "dual carbon" goals, the upgrading of the digital-intelligent fuel management system for the 4×1 000 MW units of a power plant was focused on, and a three‑in‑one framework of "multi-source perception-intelligent algorithms-blockchain certification" was proposed. A fully intelligent management system across the entire process was finally achieved by integrating optical character recognition (OCR)-QR code recognition, attention‑mechanism‑improved back propagation (BP) neural networks, lightweight fuel large models, and consortium blockchain technology. The results show that the system has achieved three major breakthroughs. The accuracy rate of mine ticket/vehicle recognition is over 99.85%. The sampling qualification rate has been raised to 98.6%, and the prediction error of procurement cost is less than 3.2%. The settlement dispute rate has dropped from 8% to 1%, and the failure rate has decreased to 0.8 times per month. The research provides systematic solutions for equipment selection, algorithm paths and blockchain modes in digital- intelligent fuel management systems.
Wu J h , Jin Z Y , Li G S , et al . Design of vehicle certification schemes in IoV based on blockchain [J]. World Wide Web , 2022 , 25 ( 5 ): 2241 - 2263 .
Zhu M , Min W D , Li J H , et al . Constructing a smoothed Leaky ReLU using a linear combination of the smoothed ReLU and identity function [J]. Neural Computing and Applications , 2025 , 37 ( 9 ): 6465 - 6478 .
Petersen P , Voigtlaender F . Optimal approximation of piecewise smooth functions using deep ReLU neural networks [J]. Neural Networks , 2018 , 108 : 296 - 330 .
Yao D Z , Shao Y X . A data efficient transformer based on Swin Transformer [J]. The Visual Computer , 2024 , 40 ( 4 ): 2589 - 2598 .