Integration of artificial intelligence, big data, and blockchain technologies in industrial production management systems

Jamshid Atamuradov, Rajabova Rano, Ahrorova Nigina, Khurshid Allaev, Farkhod Toshtemirov

2026

Closed access Metadata (CSL-JSON)
DOI 10.1117/12.3121047
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

In this study, the convergent application of artificial intelligence, big data, and blockchain technologies in industrial manufacturing management systems is examined. Analyzed are existing architectural frameworks for Smart Factory systems based on RAMI 4.0 principles with five functional data processing layers. Investigated are high-performance streaming platforms, including Apache Kafka and Apache Flink, with mathematical throughput and latency performance models. Studied are machine learning methods for predictive maintenance, including LSTM, Attention LSTM, Transformer, and CNN architectures, evaluated on the CMAPSS benchmark dataset. Determined are key quantitative performance parameters through six computational models covering throughput, time-series storage volume, economic efficiency, and blockchain transaction rates. Revealed is the superiority of Attention LSTM over classical approaches, achieving RMSE of 12.4 cycles for remaining useful life prediction with a significantly reduced rate of dangerous overestimations. Established is the practical capability of Hyperledger Fabric for permissioned manufacturing blockchains, sustaining 3,200–3,800 TPS with sub-second transaction finality. Proposed is a five-level reference architecture integrating Edge AI, fog computing, cloud analytics, a blockchain trust layer, and a business intelligence application layer. Developed is a hybrid data storage model combining blockchain cryptographic hash anchoring with traditional database management, reducing on-chain data volume by 95–99%. Justified is the use of an asymmetric loss function for remaining useful life models, which reduced the share of dangerous overestimated predictions from 31% to 14% compared to standard mean squared error training.