Cloud–edge collaborative control for roaster furnace: A hybrid first-principles and data-driven method

Abstract

Temperature stability control of the roaster furnace is an important yet challenging problem due to the fluctuation of feed compositions. To solve this problem, a first-principles and data-driven cloud–edge collaborative control (CECC) method is proposed. Specifically, through first-principles analysis, the influencing factors of temperature fluctuation are determined, and then the feed fluctuation is divided into two parts: the batch difference (BD) caused by changes in production objectives or process conditions, and the random fluctuation (RF) caused by uncertain factors of component content or ore source mixing process. For the problem of BD, a first-principles model of the roasting process is established by integrating multisource data in the cloud. Through timely response to the batch differences, the appropriate feed rate is sent to the edge controller to address the temperature unstable problem induced by the sudden component batch changes. For the problem of RF, a fuzzy controller based on trend analysis is adopted at the edge to fine-tune the feed rate to further reduce the negative impact of component fluctuation on the temperature. Accordingly, the proposed CECC method can realize stable control of the roaster furnace temperature synergistically. Both the numerical simulation experiment and hardware-in-the-loop platform experiment results verify the effectiveness of the proposed method.