The mission of the condence.io solution is to make the monitoring of industrial machinery easy and the early detection of anomalies simple.
Trend analysis is the standard method used in many applications on the market. Condence has earlier introduced tools to detect emerging patterns using the Trend Revealer tool. The new tool to the kit is the Anomaly Revealer with the ability to set value ranges for anomalies automatically. This tool increases the quality of anomaly detection, sets limits automatically and allows for the user to adjust for sensitivity, while still retaining the option to set limits manually when so desired.
The traditional method is to define limits and thresholds and to set alarms for these user-defined levels. This is still a feasible method and highly recommended in many cases. However, if there are 100 sensors all delivering 13 values (wireless vibration sensor with 4 values from 3-axis and temperature), that would result in 1300 trend metrics, so if the user wants 2 alarm levels, this means 2600 alarm limits. Configuration by itself would be a big task, but to figure out the limits for each value is a different pain. This is where the Condence Anomaly Revealer makes a difference.
In short, using a method called statistical profiling, the new feature builds automatically a channel, limits, where the values are normal or acceptable and point out, or reveals, if values deviate from the calculated normal variation range. By adjusting the time perspective, the user can adjust the sensitivity of anomaly detection for those values that best benefit from this, while still retaining the possibility to set manually limits or thresholds when desired.
This method has proven to be highly powerful and reliable in detecting anomalies that are typical to use cases in condition monitoring. On top of detecting the anomalies at a higher quality, due to the approach, they can often be explained, making it easy to continue inspecting the asset with traditional methods.
The feature will keep track of recent daily averages and compare these to historical data. The length of the period, decided by the user (30-360 days), defines the sensitivity. The longer the period is, the more sensitive the detection will be. Using daily averages as a baseline, the feature eliminates most external changes that might cause false alarms.
Notifications and alarms for the Anomaly Revealer can be configured from the user interface in Condence.io.
If you are interested in learning more, contact us or book a demo.
Panu Kinnari / COO