Metal accounting data acquisition and management using real-time sensors and artificial intelligence

• Real-time sensors improve MA data frequency, resolution, and ore characterisation. • ML improves feed characterisation, error detection and metal flow forecasting. • Real-time reconciliation enables continuous process control and decision support. • Standardised data management improves cross-departmental data harmonisation. • Digital MA systems contribute to overall mine optimisation and value addition. Reliable metallurgical accounting (MA) tracks metal flows and ensures accurate reconciliation across mining operations. The accuracy of mineral resource estimates and production reports depends on the quality of input data and subsequent management procedures. Conventional MA systems are routinely challenged by sampling bias, manual data handling, siloed databases, and delayed reporting, which undermine reconciliation efforts. This study proposes a digital MA framework that combines real-time sensor networks with machine learning (ML) and structured data management to improve data quality and traceability. Sensor technologies installed along material streams generate high-frequency geometallurgical data, allowing near real-time ore characterisation. When integrated with ML, this data can be used to correlate ore composition with processing performance metrics, providing valuable insights into metal recovery and process efficiency. Compared to conventional approaches, ML algorithms increased R² from 0.651 to 0.728 and reduced RMSE from 0.489 to 0.436. The workflow is reproducible, with an explicit data schema, QA/QC rules and a transparent reconciliation algorithm. The proposed architecture supports automated reconciliation at hourly intervals, far exceeding the timeliness of monthly or quarterly cycles used in current practices. Case studies demonstrate that digital MA workflows reduce data errors, shorten reconciliation periods and improve decision-making across the mining value chain. While implementation requires infrastructure upgrades and workforce training, the benefits include faster reporting, improved process control and measurable reductions in reconciliation time. The framework builds on AMIRA P754 principles and provides a scalable, modern solution for improving MA systems in a dynamic mining industry.