Cloud-edge-end collaborative multi-process dynamic optimization for energy-efficient aluminum casting

Abstract

Casting is a crucial, but energy-intensive aluminum processing technology. To achieve carbon neutrality goals, it is essential to reduce casting energy consumption without compromising productivity. Optimizing operational parameters in aluminum casting is an effective strategy, yet two main challenges remain: understanding the complex relationship between operational parameters and energy consumption, and adapting the optimization process to production dynamics. This paper introduces a cloud-edge-end collaborative predictive-reactive scheduling approach to tackle the second challenge, based on our understanding of the first challenge. Specific dynamic adjustment measures for four common dynamic events, that is, alterations in production plans, fluctuations in pass rates, production interruptions, and deviations from implementation, were proposed. A cloud-edge-end collaborative dynamic adjustment framework is then designed to implement these measures. The proposed approach was tested in a die-casting factory to validate its performance. The results demonstrate that the data-driven approach can generate adjustment measures for detected dynamic events in near-real-time, with the longest response time being less than one minute. These measures significantly reduce casting inventory and energy consumption, achieving a 19.5 % reduction in energy cost during a planned production interruption. The proposed dynamic optimization approach shows promise for energy conservation in the aluminum casting industry.