An interactive architecture for industrial scale prediction: Industry 4.0 adaptation of machine learning

Abstract

According to wiki definition, there are four design principles in Industry 4.0. These principles support companies in identifying and implementing Industry 4.0 scenarios, namely, Interoperability, Information transparency, Technical assistance, Decentralized decisions. In this paper we have discussed our work on an implementation of a machine learning based interactive architecture for industrial scale prediction for dynamic distribution of water resources across the continent, keeping the four corners of Industry 4.0 in place. We report the possibility of producing most probable high resolution estimation regarding the water balance in any region within Australia by implementation of an intelligent system that can integrate spatial-temporal data from various independent sensors and models, with the ground truth data produced by 250 practitioners from the irrigation industry across Australia. This architectural implementation on a cloud computing platform linked with a freely distributed mobile application, allowing interactive ground truthing of a machine learning model on a continental scale, shows accuracy of 90% with 85% sensitivity of correct surface soil moisture estimation with end users at its complete control. Along with high level of information transparency and interoperability, providing on-demand technical supports and motivating users by allowing them to customize and control their own local predictive models, show the successfulness of principles in Industry 4.0 in real environmental issues in the future adaptation in various industries starting from resource management to modern generation soft robotics.