Predictive maintenance is a term widely used by reliability, maintenance and operation groups. Shape’s proposal for predictive models considers the development of artificial intelligence tools, ingesting signals generated by field sensors.
Integrate available data into a single source of truth
Transform data into insights that address business needs
Close the gap between digital solutions and front liners
From the identification of patterns and their correlation with past events of anomalies and failures, predictive models make inferences about the health status of monitored equipment. The PdM module has an overview of the plant on a heat map, monitoring both asset health and signal quality. This visualization filters signals so that engineers and operators direct attention to equipment and failure modes that effectively requires.
The alarm center gathers all notifications of anomalies generated by predictive models, in cockpits optimized for efficient decision making of reliability, maintenance and operation engineering groups. On this same screen, the user can feed back the machine learning models so that they always remain at the best possible performance.
The alarm center concentrates:
·
Active alarms that require attention from reliability, maintenance and operation teams;
·
Alarms that are in the process of analysis regarding the identification of root cause and actions to be taken;
·
Alarms already closed, which make up a base of lessons learned. The alarm center gathers all notifications of anomalies generated by predictive models, in cockpits optimized for efficient decision making of reliability, maintenance and operation engineering groups. On this same screen, the user can feed back the machine learning models so that they always remain at the best possible performance
SHAPE’S TAKE ON PdM EVOLUTION
Predictive Maintenance (PdM) models aim to detect issues early, when they are still minor. Offering a large reaction time and minimal wear to equipment.
PdM 1.0 | Condition-based maintenance
PdM 2.0 | Equation-based predictions
PdM 3.0 | Fit-for-purpose analytics suíte
PdM 4.0 | Asset-wide analytics system
MACHINE LEARNING
AND AI
The model is trained with a supervised approach over a period, it learns what is pre-failure behavior and what is not.
It is then tested – blind test – in a different period, alarming the failure events of that period.
OBJECTIVE
Anticipate failure events, to give enough time for the maintenance teams to act and prevent downtime.
HANDS ON THE PRODUCT
The possible interface of the Condition Monitoring dashboard at the equipment level are all included in the PdM solution.
Heatmap
The heatmap screen provide managers a bird’s-eye view of the current condition of multiple plants, helping them allocate resources to diagnosis and treatment.
Alarm Manager
The Alarm manager enables users to quickly identify equipment with high-risk failure mode and act to prevent it. Notifications are sent to stakeholders to ensure action in time.
Asset Dashboard
The asset-level dashboard is the one-stop shop for all information about each asset, including the alarms generated by all predictive and simulation models.
Performance
Integrated visualization of value captured by the solutions in downtime avoided historically. Includes asset performance benchmarking and failure events breakdown.
SCHEDULE A DEMONSTRATION WITH ONE OF OUR SPECIALISTS