Adaptive self-learning mechanisms for updating short-term production decisions in an industrial mining complex

Abstract

A mining complex is an integrated value chain where the materials extracted from a group of mineral deposits are sent to different processing streams to produce sellable products. A major short-term decision in a mining complex is to determine the flow of materials that first includes deciding which handling facilities to send the extracted materials and then determining how to utilize the processing facilities. The flow of materials through the mining complex is significantly dependent on the performance of and interaction between its different components. New digital technologies, including the development of advanced sensors and monitoring devices, have enabled a mining complex to acquire new information about the performance of its different components. This paper proposes a new continuous updating framework that combines policy gradient reinforcement learning and an extended ensemble Kalman filter to adapt the short-term flow of materials in a mining complex with incoming information. The framework first uses a new extended ensemble Kalman filter to update the uncertainty models of the different components of a mining complex with new incoming information. Then, the updated uncertainty models are fed to a neural network trained using a policy gradient reinforcement learning algorithm to adapt the short-term flow of materials in a mining complex. The proposed framework is applied to a copper mining complex and shows its ability to efficiently adapt the short-term flow of materials in an operational mining environment with new incoming information. The framework better meets the different production targets while improving the cumulative cash flow compared to industry standard approaches.

Cognitive manufacturing: definition and current trends

“…Based on the identified disturbance, adaptive scheduling was applied to a semiconductor manufacturing system (Qiao et al, 2020). To adapt to current conditions, a Reinforcement learning mechanism was developed which updates flow of material based on real time sensor data and other monitoring devices (Kumar et al, 2020). Another control mechanism that was implemented involves an Analytic Hierarchy Process with expert rules applied to a dispatching problem in an assembly process to adapt to disturbances in production (Attajer et al, 2022).…”

Deep reinforcement learning in production systems: a systematic literature review

“…Although the agents received detailed local state information, they not only selected the fastest local dispatching actions but also improved the global logistics performance. Further, the distributed agents enabled real-time responses, a feature also emphasised by Kumar, Dimitrakopoulos, and Maulen (2020) for the short-term value stream adaptation in a copper mining complex. Based on the current mining process and component data, the single-agent framework allowed to deliver continuous updates regarding the latest plant status and increased the expected net present value by 6.5%.…”

Deep reinforcement learning methods for structure-guided processing path optimization

“…Recently published work include the optimization of process control in sheet metal milling (Veeramani et al 2019), polymerization reaction systems (Ma et al 2019), laser welding (Günther et al 2016) and in deep drawing (Dornheim et al 2019). Operational optimization objects are amongst others material flow in industrial mining (Kumar et al 2020), preventive maintenance scheduling of flow line systems (Wang et al 2016) and job shop scheduling (Kuhnle et al 2020).…”