My PhD research project proposes an Autonomous Operator (AO) based on Machine Learning to analyse and identify Cyber-Physical Attacks (CPA) in Mining Process Plants. Supervisory Control and Data Acquisition (SCADA) systems monitor and controls physical processes that can be attacked physically or virtually. These attacks trigger alarms in the SCADA system to alert the human operator to respond, which could be misleading he or she to stop the process plant unnecessarily possibly causing financial losses and, in some extreme cases, even fatalities.
The AO is proposed to aid the human operator to identify if the SCADA alarm triggered is a result of a CPA or component or system malfunction. The AO will not be a cyber expert because the human operator is not. The AO will learn how the process plant works and when the alarm goes off the AO will investigate through independent access to the key points of the process plant to determine the cause of the alarm.
The AO will be trained with two sets of data, one to teach the normal behaviour and one for the abnormal behaviour of the process plant. After the training, the process plant will be under simulations of different types of CPA to create the abnormal behaviour, so the AO can be tested in identifying the CPA and distinguish them from an ordinary system or equipment failure.
Past Research employment history