Using predictive model output data
The results from each of the five the predictive maintenance models in Maximo® APM - Predictive Maintenance Insights SaaS provide you with various data that you can evaluate and use to determine which maintenance actions to take for assets.
Failure Probability
By using the Failure Probability model results, you can view the probability that the asset will fail within a specified time period, such as 30 days, the amount that the probability has increased or decreased since the previous assessment, and the average failure probability for all assets in the group.For example, if the model is predicting a high probability of failure in the next 60 days, you might view upcoming maintenance and determine whether scheduled work or preventive maintenance can address the problem. You can update existing work orders with the observations from the model. Alternatively, if no upcoming work orders or preventive maintenance is scheduled, you might create an inspection work order to investigate likely problems.
Failure Contribution Breakdown
By using the Failure Contribution Breakdown model results, you can view a breakdown of the asset’s attributes to see how the attributes contribute to the failure probability of the asset. You can also see how attributes contribute to failure in an analysis tree model.You can use the Failure Contribution Breakdown results or analysis tree to review which asset attributes predict a failure to help technicians or inspectors focus on likely problems. For example, an extremely high probability of failure might require corrective or emergency work to be planned as soon as possible.
Predicted Failure Dates
By using the Predicted Failure Dates model results, you can view the date that the asset is predicted to fail, the scheduled next maintenance date, and the date that the asset was assessed. You can navigate to view more information in the maintenance logs.If you apply a predictive maintenance strategy to your assets, you can use this model to determine when to create and schedule work to address issues. You can accurately plan the work and organize the necessary labor, spare parts, and inventory ahead of the predicted failure date, and you can work with your maintenance team to plan and schedule the work during already scheduled downtime. By using a predictive maintenance strategy, you can reduce the costs that are associated with unnecessary preventive maintenance.
Anomaly Detection
By using the Anomaly Detection model results, you can view the latest anomaly score versus the anomaly threshold and the date of the previously detected anomaly. You can navigate to view more information in the anomaly score history details.You can use the results from this model to monitor the behavior of your assets. If an anomaly is detected, it indicates that there is some deviation from normal behavior for this asset. If you know the attributes of the asset, such as the meter or IoT sensor outputs that this model is analyzing, you can determine what cause the anomaly. Frequent anomalous behavior for an asset might suggest that some underlying issue requires further investigation by the maintenance team. A refurbishment or replacement might be necessary, depending on the age of the asset or where the asset exists on the failure probability curve.
Failure Probability Curve
The Failure Probability Curve shows the probability that an asset will fail at certain points in time. The curve measures retirement and failure rates of similar assets against the age of the selected asset.
You can use the curve to estimate how likely your asset is to fail as it gets older. The risk of failure grows over time, so you might decide to replace or refurbish the asset, especially if you notice that its health is also deteriorating. You can also compare information in the curve with other failure prediction models. If the information and the models align, the predicted failures are most likely related to the age of the asset.