Applying AI Capabilities in SAP Asset Performance Management

1. Predictive Maintenance: The use of data analysis tools and techniques to detect anomalies and failure patterns to predict equipment failures before they happen.
2. Machine Learning Engine: A core component of AI that processes data and learns from it to make informed decisions, infused into SAP APM to provide insights.
3. Data Science-Driven Approach: A method that utilizes data analysis and statistics to create predictive models and make data-informed decisions in condition monitoring.
4. Extension Framework: A flexible system within SAP APM that allows for the development of industry-specific or customer-specific models and analytics.
5. Operational Visibility: The ability to see and understand the workings of your assets in real-time, enabled by AI to predict and prevent potential failures.
6. Anomaly Detection: An AI function that identifies patterns in data that do not conform to expected behavior, used to identify early signs of potential problems.
7. Anomaly Score: A metric indicating the degree of deviation from the normal behavior in data, which helps in assessing the significance of an anomaly.
8. Weibull Analysis: A statistical analysis tool used in reliability engineering for predicting failure rates and understanding life data analysis, integrated into SAP APM for predicting future failures.
9. Probability of Failure Curve: A graphical representation that shows the likelihood of failure over time, helping in decision-making for maintenance planning.
10. Maintenance Strategies: Approaches combining predictive, prescriptive, and preventive maintenance, formulated using AI capabilities within SAP APM.
11. Equipment-as-a-Service (EaaS): A business model where the OEM provides the equipment's functionality as a service, with reduced risk and cost through AI capabilities.
12. First Visit Fix Rate: The ability of service managers to correctly identify and resolve issues on the first visit, improved by AI's predictive insights.

This lesson covers the forthcoming artificial intelligence functionalities within SAP Asset Performance Management (SAP APM), which play a pivotal role in facilitating predictive maintenance for assets to proactively prevent failures.

AI capabilities integrated into SAP APM will contribute to achieving the following asset management goals:

• Facilitates a data science-driven approach to condition monitoring.
• Offers a flexible extension framework for customers to develop industry- or customer-specific models and analytics.
• Employs a scalable Machine Learning Engine to infuse data science insights into our business processes.
• Provides adaptable visualizations for equipment structures.
• Supports end-to-end process integration, encompassing alerting, discovery, and remediation.

Leveraging machine learning and artificial intelligence capabilities, maintenance strategies within SAP APM can be formulated by combining predictive, prescriptive, and preventive maintenance approaches.

The utilization of engineering and machine learning models is increasing in prominence. The forthcoming AI capabilities in SAP APM will bring about improved operational visibility and early detection of failures, benefiting both Original Equipment Manufacturers (OEMs) and operators alike.

OEM Aspect

• Reduced risk and cost executing an equipment-as-a-service business model.
• Timely identification of design issues that can impact the number of future warranty costs.
• Higher first visit fix rates as Service Managers can identify the required skills and spare parts prior to arriving at the job site.

Operator Aspect

• Reduced number of reactive maintenance events.
• Better utilization of maintenance resources since planned maintenance events can be more flexibly planned and packaged.
• More effective use of scheduled asset downtime.

Forthcoming AI Capabilities in SAP APM will be:

• Generic issue detection
• Anomaly detection
  • Across various indicators (time series sensor data), assess the degree of anomaly ("anomaly score") in the most recent data in comparison to historical or reference data.
  • Additionally, provide explanations indicating which specific indicator(s) contributed significantly to a high anomaly score.
• Purpose-built functions
  • Utilize the well-established "Weibull" algorithm to compute a "probability of failure" curve based on historical breakdown data, projecting future failure probabilities.
  • Recommend adjustments to regular maintenance intervals based on the calculated probability of failure.