The mining industry is undergoing a transformation. Margins are being squeezed as mines advance deeper, face more challenging geologies, and adapt to societal pressures.
In order to remain competitive, innovative miners are leveraging digital technologies to improve recoveries, increase throughput, and reduce energy consumption.
Digital technologies, such as advanced analytics, provide a powerful set of tools to improve and optimise process control within beneficiation plants.
Part of this opportunity stems from the fact that many plant operators lack an optimised control methodology to help them select appropriate process set points.
Read more about base metals
These operators rely solely on their years of experience, and on their gut-feel, to make control decisions. Their decisions are based on a loose approximation of the intricacies of the process. The result is sub-optimal performance, and increased day-to-day variance in the plant’s metrics.
Read more about gold
While theoretically possible, the exercise of optimising a beneficiation process based on an accurate and complex mathematical model is expensive and time-consuming – if not impracticable.
Read more about platinum group metals
It requires a team of domain specialists to painstakingly model the intricate relationships between process parameters and output metrics. Many of the second and third-order inter-dependencies involved, would, in all likelihood, not be documented in the scientific literature.
Read more about coal
The field of advanced analytics has opened up effective alternative methods of process optimisation that unlock the potential of historical process data.
These include expert systems, which yield insights from snapshots of historical data, as well as the more recent and sophisticated expert execution systems (EES), which inductively discover the complex relationships between process parameters and plant metrics in order to generate prescriptive remedial actions.
The older expert system is a form of advanced analytics that, for years, has assisted plant operators by inferring improved control decisions from historical plant data.
While expert systems have yielded significant efficiencies over the years, in industries that range from manufacturing to mining, they also have important limitations.
In particular, expert systems sought to provide operators with inputs that had been predetermined from a set of observable prior conditions. These were inherently analytically modeled from a set of first-order equations reflecting the state of knowledge about the system at the time.
Even when these systems were augmented with some learned decision logic, they remained inflexible around the domain of the observed science. Some expert systems took incredibly long to compile, as the underlying models were excessively complex.
Accurate, dynamic, analytical expressions were absent for many of the processes. Meanwhile, engineers needed time to interpret the expert system’s results, adding to their inability to keep up with changing plant dynamics.
Expert execution systems
The EES overcomes the limitations of classical expert systems in two important ways. First, an EES harnesses advanced machine learning algorithms to extract a deeper and more holistic model of the industrial process in a short space of time.
Second, the EES leverages this model through an intelligent interface that delivers prescriptive remedial actions to the plant operator, thereby optimising the process ahead of real-time
DataProphet PRESCRIBE is an award-winning EES that enables beneficiation plants to improve plant metrics through advanced, real-time prescriptions.
DataProphet have won a number of awards such as being recognised as a 2019 Tech Pioneer by the World Economic Forum, listed as one of the 100 AI Startups of 2020 by CBInsights, winner of the Best Innovation in Deep Learning by AIconics as well as recognised by Frost & Sullivan as the winner of their Technology Innovation Leadership Award 2020, amongst many others.