• Written by Azhar Aliyar, Director

Can we really trust our maintenance records to an AI system, or will it just become another source of conflicting information, asked the maintenance manager during the meeting with the IT department. This question gets to the heart of the challenge of integrating AI systems with the existing systems of record, particularly in maintenance, where accuracy and consistency are crucial. The difficulty lies in connecting the AI system to the various sources of maintenance data without creating a second, potentially conflicting, source of truth.

Read paths, write paths and the need for care

The integration of AI systems with maintenance records involves both read paths and write paths. Read paths are used to extract data from the existing systems, such as the maintenance log or the asset history, while write paths are used to update these systems with new information. However, the write paths require more care, as they can potentially introduce inconsistencies or errors into the system. Most people get wrong the idea that write paths are just as simple as read paths, but they are wrong, as the consequences of incorrect data can be severe.

A shift supervisor checking the maintenance log, for example, relies on the accuracy of the data to plan and schedule maintenance activities. If the AI system introduces errors or inconsistencies into the log, it can have serious consequences, such as delayed or missed maintenance, which can lead to equipment failures or even safety risks. Therefore, it is essential to carefully design and test the write paths to ensure that they update the systems correctly and consistently.

Reconciling definitions that differ between systems

One of the challenges of integrating AI systems with maintenance records is reconciling the different definitions and formats used in various systems. For instance, the maintenance log may use a specific format to describe an asset, while the asset history uses a different format. The AI system must be able to reconcile these differences and provide a consistent view of the data. This requires careful analysis and mapping of the data formats and definitions used in each system.

The category manager comparing two suppliers, for example, needs to ensure that the data from each supplier is consistent and comparable. If the AI system can reconcile the differences in definitions and formats, it can provide a more accurate and comprehensive view of the data, enabling better decision-making.

Maintenance data and its components

The integration of AI systems with maintenance records relies on several key components of maintenance data, including work orders, asset history, spare parts inventory, and condition data where it exists. Work orders provide a record of the maintenance activities performed, while the asset history provides a detailed record of the asset's maintenance and repair history. The spare parts inventory is essential for planning and scheduling maintenance, as it ensures that the necessary parts are available when needed.

Condition data, where it exists, provides real-time information on the asset's condition, enabling predictive maintenance and reducing the risk of unexpected failures. The AI system can analyze this data to identify patterns and trends, enabling more effective maintenance planning and scheduling. For example, a maintenance planner can use the AI system to analyze the condition data and schedule maintenance activities accordingly, reducing downtime and increasing overall efficiency.

Impact on reactive and planned work

The integration of AI systems with maintenance records can have a significant impact on the balance between reactive and planned work. Reactive work, such as emergency repairs, can crowd out planned work, such as routine maintenance, if not managed properly. The AI system can help to identify and prioritize planned work, ensuring that it is not neglected in favor of reactive work.

Additionally, the AI system can help to identify and address the root causes of recurring issues, reducing the likelihood of the same asset recurring under different descriptions. For example, a maintenance technician can use the AI system to analyze the maintenance history and identify the root cause of a recurring issue, enabling them to perform a more effective repair and reducing the likelihood of future occurrences.

Measuring success

The success of the integration of AI systems with maintenance records can be measured using several key metrics, including the planned versus reactive ratio, mean time between failures, and spares availability. The planned versus reactive ratio provides a measure of the balance between planned and reactive work, while the mean time between failures provides a measure of the overall reliability of the assets. Spares availability provides a measure of the effectiveness of the spare parts inventory management.

By tracking these metrics, maintenance managers can evaluate the effectiveness of the AI system and make adjustments as needed. For example, if the planned versus reactive ratio is too low, it may indicate that the AI system is not effectively prioritizing planned work, and adjustments can be made to the system to address this issue. To learn more about how AsscherAi can help with production, visit our production page.

Limitations and when not to use

The integration of AI systems with maintenance records is not a solution for every maintenance-related problem. In some cases, it may be more effective to use other approaches, such as manual data analysis or traditional maintenance scheduling methods. For example, if the maintenance data is limited or of poor quality, the AI system may not be able to provide accurate or reliable results.

In such cases, it may be more effective to focus on improving the quality and availability of the maintenance data before attempting to integrate an AI system. If you have questions about how AsscherAi can help with your maintenance needs, or to discuss your specific use case, please contact us to learn more.