In every production facility of any capital-intensive industries, unexpected failure and downtime of equipments can bring the whole operation to a halt, and cause major business problem.
Being able to predict equipments or machines failure before it happens will help company to reduce maintenance and repair cost significantly. And it also help them to plan better for sourcing crucial parts.
We collect and integrate maintenance-related data from multiple sources like inspection data (oil sampling, visual inspection report, etc.), sensors data (vibration, pressure, temperature, voltage, RPM), and maintenance post-event data (damage classes, maintenance types, planned v.s. unplanned events, maintenance actions, and maintenance subcomponents. Then with the help of machine learning model, we combine these data to produce estimation of failure mode probability, distribution of repair cost and severity of failure, and prediction of detectability and time to failure.