Data Scientist, Vale, Australia
The application of Predicting Fractures to Optimize Rails Maintenance
Transferring the mined material from mine to port is one the most important operations in mine value chain. The rail maintenance not only is important to reduce the total mine product cost but also is a critical parameter to increase the safety and sustainability. The main part of mine material producers working in big countries same as Australia, China, Brazil, etc. where we need huge and complex rail networks. These networks need highly complex equipment, consisting of a variety of integrated subsystems, assembled to provide mine material transportation.
There are plenty of classical methods to sense or estimate rail failures. This presentation demonstrates an innovative predictive model working based on advanced analytics technology to predict failures that potentially will happen in the next 30 days. This technology has been developed by researchers in VALE company in Brazil. A special control car has been equipped in this project to observe and collect data from the all rail components using high-quality cameras and laser detection system. The collected data will be completed by other data sets coming from telemetry, GPS, ultrasonic diagnostic system and historical failures data sets. A smart maintenance prediction model works based on the gathered and cleaned data in data lake using Big Data Technologies, Statistical Models and Machine Learning. The application of the developed technology gives this capability to the maintenance team to predict the rail failures in a small segment with a length of about 25 cm. The developed technology has been validated and tested by VALE maintenance team in Brazil. The application of this technology helps VALE to optimize maintenance process, minimize the risk of accidents, unwanted interruption, reduction of total cost and make better decisions to use of mine assets. This presentation illustrates the developed technology, and some achieved practical results in VALE mine assets in Brazil.
Dr. Ali Soofastaei is a Research Developer at the Artificial Intelligence Centre of Excellence at Vale and an Adjunct Senior Fellow at the University of Queensland (UQ), Australia. Vale is a multinational corporation engaged in metals and mining. It is one of the world’s foremost producers of iron ore and the largest producer of nickel. UQ has been ranked in the world’s top 50 universities and is one of Australia’s leading research and teaching institutions.
Dr. Soofastaei uses innovative models based on artificial intelligence (AI) methods to improve safety, productivity, and energy efficiency, and to reduce maintenance costs. He holds a Bachelor of Engineering in Mechanical Engineering and has an in-depth understanding of energy management (EM) and equipment maintenance solutions (EMS). The extensive research he conducted on AI and value engineering (VE) methods while completing his Master of Engineering also provided him with expertise in the application of advanced analytics in EM and EMS.
In the past 15 years, Dr. Soofastaei has conducted a variety of research studies in academic and industrial environments. He has acquired in-depth knowledge of energy efficiency opportunities (EEO), VE and advanced analytics. He is an expert in the use of DL and AI methods in data analysis to develop predictive and optimization models of complex systems.