Focus on: Smart Maintenance Maintenance Matters February/March 2025 www.pwemag.co.uk Plant & Works Engineering | 17 data. It offers flexible integration, allowing customers to deploy it on-premises, in the cloud, or on edge components directly within machines. This approach helps address the challenge of adhering to increasingly stringent cybersecurity policies, ensuring that customers retain ownership of their collected data and analyses while external access remains strictly controlled. When deployed on edge, Festo AX efficiently handles large data volumes with high frequency and low latency while minimising data transfer costs. A hybrid solution is also available, allowing complex computing tasks and large-scale data processing to be managed locally, while only defined outputs and algorithm model training are processed in scalable cloud installations. Standard communication protocols, such as OPC-UA and MQTT, are used whenever possible to access data. When an anomaly is detected in the machine’s operational data, the software issues a notification identifying the affected machine operation and suggesting a recommended course of action. The notification function, known as Smartenance, enables users to document, manage, and distribute anomaly detection outputs. Notifications include functionalities such as anomaly data visualisation, automatic root cause analysis, and diagnostic classification tools. The system employs an iterative ‘human-in-theloop’ process, allowing operators to classify anomalies and refine the software’s recommendations over time. This method enhances algorithm accuracy and reduces unnecessary notifications through a reinforcement process, where useful notifications are positively reinforced, and less relevant ones are refined. The more frequently anomalies are encountered, whether across identical machines or through high recurrence rates, the faster the digital model learns. This approach addresses one of the primary frustrations of predictive maintenance: false positives. By deploying Festo AX, anomalies can be detected earlier and with greater reliability, enabling maintenance teams to prepare spare parts and conduct necessary repairs without disrupting production output. Providing valuable insight As the benefits of data analysis become more widely understood, demand for greater insight continues to grow. Sands highlights that Festo AX addresses this need by offering pre-defined, easily implemented solutions using containerised ML function blocks, enabling nondata scientists to uncover valuable correlations in their data. For broader applications, Festo AX projects can be fully customised with extensive support from an implementation team. For instance, in response to abnormal machine behaviour, the system automatically generates a root cause analysis, identifying the sensors responsible for detecting the anomaly. This visualisation capability allows users to develop a deeper understanding of anomalies and recognise critical correlations. While Festo provides default parameters for AX installations, users can readily adjust these settings to match their specific requirements without requiring advanced data science expertise. Beyond monitoring machine ‘health’, Festo AX includes pre-configured modules for Predictive Maintenance, Predictive Energy, and Predictive Quality. These modules enable tailored intelligence for operators. For example, Predictive Energy analyses energy consumption, creates load profiles, and optimises usage patterns, reducing costs by detecting leaks at an early stage. Additionally, monitoring machine output data separately from the standard control architecture prevents overloading or slowing of the primary control system. This capability is particularly valuable for existing installations where replacing the control system would be costly and pose risks related to machine warranties and reliability. A key challenge for manufacturers is ensuring seamless integration of predictive maintenance tools with existing machinery and equipment. Sands notes that the Festo AX software is designed to be open and adaptable, capable of analysing and optimising machinery containing components from multiple providers, not just those supplied by Festo. Reaping the rewards as an early adopter Unplanned work stoppages cost global industries millions of pounds in lost annual revenue. In manufacturing, where time is money, even a few minutes of unexpected downtime can lead to substantial financial losses. Manufacturers implementing AI-powered predictive maintenance solutions are already experiencing significant benefits. As with any emerging technology, there is a learning curve. However, manufacturers do not need to undertake large-scale facility-wide installations immediately. A targeted approach, focusing on areas with the greatest potential for rapid improvements and returns on investment, is often the most effective strategy. Successful AIbased predictive maintenance projects frequently adopt a step-by-step methodology, initially testing and evaluating the technology before scaling up. The coming years will be an exciting period for the industry as AI technology advances and the market continues to expand. Sands expresses that Festo remains committed to playing a key role in this evolving landscape. For more information, visit www.festo.com/ax Festo AX delivers a simple AI solution for pneumatic motion.
RkJQdWJsaXNoZXIy MjQ0NzM=