ODiN Predictive Maintenance uses revolutionary machine learning methods to generate ongoing knowledge on the state of health of equipment. The Industry 4.0 technique uses recorded sensor data and enables reliable predictions about likely time to failure.
Traditional condition monitoring systems are associated with value-based analysis, whereas ODiN employs a model-based approach. For the first time, technology is helping to advance the process from pure state monitoring to truly predictive analysis and data-driven, anticipatory maintenance.
A machine learning algorithm determines a normal healthy state for each component, based on information from sensor signals taking readings such as pressure, flow rate and temperature. This initial phase may last for a few days (if the system in question carries out the same functions all the time) of a longer period of time.
Following this ‘learning’ period, ODiN defines a ‘health index’ for each component being monitored. If an individual measured value temporarily deviates from the tolerance range, an error warning may not necessarily be generated. However, if the index deteriorates based on the data from multiple sensors, ODiN will warn operators of an impending problem.
“Diagnosing wear and tear in industrial applications is an extremely complex task,” said Tapio Torikka of Bosch Rexroth. “Statistically, there is only a 13% chance of an issue being detected by chance, while an expert monitoring the system by traditional means has a 43% chance of detecting it. Our system has a detection rate of 99%.
“The system acquires all the necessary information from the sensor data and machine learning methods, then converts this into knowledge. The health index therefore not only shows the state of the assembly currently being monitored, but also gradual changes to upstream and downstream mechanical or hydraulic systems.
“Even ODiN cannot fully eliminate the risk of plant downtime, but we can reduce the risk so significantly that the costs for the system are generally already recouped after the first prevented downtime.”
