AI And CM Intersect

Combining AI with condition monitoring can raise the benefits of predictive maintenance to new levels.

AI can drive a host of manufacturing and business operations. The key is harnessing that power.

By Michael DeMaria, Fluke Reliability

Imagine a manufacturing facility that is always on. It monitors itself in real-time, predicting maintenance needs before equipment begins to lag or break down, and increases productivity while being more sustainable than ever. This is the potential when artificial intelligence (AI), condition monitoring, and Industry 5.0 converge.

Industry 5.0, is characterized by the seamless integration of humans and digital technology such as AI, robotics, IIoT devices, and big data. Its focus is to enhance capabilities for both, empowering workers and driving innovation and growth. Allied Research states the global Industry 5.0 has the potential to reach $637.4 billion USD by 2032. Condition monitoring plays a critical role in this growth.

Advancements in AI are further enhancing the capabilities of modern condition-monitoring techniques. AI-powered tools can unlock even higher levels of operational efficiency, moving from reactive to predictive maintenance. They can uncover hidden insights within operational data, facilitate predictive modeling and scenario analysis to optimize maintenance schedules, allocate resources, and finesse production planning.

In fact, a recent survey by Fluke Reliability (Everett, WA, fluke.com) of 600 senior decision makers and maintenance professionals in the U.S., U.K., and Germany found that 93% consider AI a high business priority for the next 12 months. On average, respondents plan to allocate 44% of their 2024 technology budgets to AI. Additionally, 30% of those surveyed intend to invest 51% to 75% of their technology budget in AI this year.

Let’s take a closer look at how these technologies are driving the interconnection, process optimization, and efficiency hallmarks of Industry 5.0.

Enhancing PdM

By far the most important application for AI in condition monitoring is predictive maintenance. This strategy identifies issues with equipment before it breaks down so that maintenance activities can be scheduled proactively.

Predictive maintenance can increase asset productivity by as much as 20% and reduce overall maintenance costs by 10%, according to management consulting firm McKinsey. Here’s how AI and condition monitoring play a role in achieving these metrics:

Anomaly detection: Traditionally, experts have identified faults in machinery by recognizing specific patterns in data. However, with labor shortages, few teams have the time or personnel to conduct extensive research.

Condition monitoring uses sensors to collect data that is usually centralized in a software solution such as a computerized maintenance management system (CMMS). Machine-learning algorithms then efficiently analyze huge volumes of data to establish a baseline “normal” for each asset.

This allows teams to identify anomalies in equipment performance, even minor anomalies such as temperature and vibration shifts, and flag them as potential issues. When values and thresholds for assets are breached, automation kicks in to send an alert to engineers about the issue.

Take the Jack Daniel Cooperage, where barrels for Jack Daniel’s whiskey are created, as an example. The cooperage uses an AI-powered condition-monitoring system to automatically generate work orders when equipment exceeds set thresholds. The results have been so successful that their strategies are being adopted throughout Brown-Forman, Jack Daniel’s parent company.

Pattern recognition: By using machine learning, the data derived from condition monitoring can be analyzed in greater detail to enable precise diagnostics and early warning of impending equipment failures. Causal AI algorithms can recognize complex patterns and correlations within sensor data that may indicate specific failure modes or performance degradation.

For instance, AI-powered software can use vibration data to diagnose hundreds of faults in various systems many months before they become critical enough to be considered a failure mode. This includes examples such as pump impellers, motor stators, and electric motor winding.

AI learns from new data, improving the accuracy of fault detection algorithms over time. This iterative learning process allows the system to adapt to changing operating conditions and evolving equipment behavior, keeping diagnostics accurate and effective.

Better decision making: Leveraging insights from pattern recognition and anomaly detection, predictive-maintenance platforms can provide maintenance personnel with actionable recommendations that boost equipment performance and reliability.

Some companies are now integrating generative AI into their predictive-maintenance systems, making them more conversational and intuitive. This means that facility workers can ask questions such as, “What could be causing this equipment issue?” or, “Show me this compressor’s repair history,” to the AI tool, which can leverage asset history, data insight, and reference material such as maintenance manuals to help the workers solve problems.

Smarter Business Decisions

Just as AI and condition monitoring have advanced predictive maintenance practices, they are now redefining business processes to be more accurate and intelligent.

For example, significantly improved forecasting accuracy, based on asset-health monitoring, allows an AI-based CMMS to determine what tools or spare parts will be needed for preventive-maintenance work. Additionally, an AI-based CMMS, trained in image recognition and large-language models, can provide accurate visibility into spare-parts inventory.

Combined, this means that facility managers should be better aware of inventory levels, preventing overstocking and unnecessary spending. Concurrently, if inventory drops below set levels, automation can trigger an email reminder or a reorder process, preemptively stocking parts. This efficiency reduces downtime and delays.

The data gained from AI and condition monitoring can also be linked with other systems in a plant, such as an enterprise resource planning (ERP) system, to improve coordination and decision making across all business functions.

The future of Industry 5.0 is predictive, interconnected, and intelligent. Adoption of AI for condition monitoring is at the forefront of this transformation. However, successful deployment requires a strategic approach and commitment to system integration. A connected operations ecosystem should centralize all data, integrate asset health data with operational systems, and provide all personnel access to asset information for efficient maintenance. Embracing these innovations means less waste, smarter decisions, and a sharper competitive edge. EP

Michael DeMaria, is Director of Product Management at Fluke Reliability, Everett, WA, (reliability.fluke.com). He leads the team responsible for overseeing hardware platforms, integrations, diagnostic software, AI tools, and user portal deliverables and business metrics. With more than three decades of experience in vibration analysis, his expertise is rooted in his background in Navy nuclear engineering.