With the rise of Internet of Things (IoT) and Artificial Intelligence (AI), collecting data has become relatively simple. Drones, sensors, RFID tags, acoustic emission, inspection trains etc. all collect immeasurable quantities of track data. Put simply, collecting data has become the ‘easy’ part.
Having access to all this data can, however, seem a bit overwhelming. How can you make sense of what’s useful and what’s not? And, how can you turn all this data into smart maintenance? Do you have a clear view of what information you want to extract? Yes. Translating data into maintenance actions is not an easy task. However, it can become much more manageable if you follow these five steps:
Data processing represents the conversion of data into a useable and desired from.
The conversion is done by synchronizing the gathered data into a software of your choice. The software will then help you to filter out the most useful information out of it. This enables you to extract the most relevant information from the collected data. Depending on the software, the processed data can be obtained in a desired format. This could be a graph, table, charts or any other format.
Data analytics and interpretation
The once overwhelming volume of disparate data has now been processed into something much more manageable. However, you need more from your data:
- You need to find out whether the processed data is the right data for answering your initial questions (the reason why you started collecting data)
- You need to use this data to draw accurate conclusions
- You need data that informs your decision-making process
Data analysis enables you to analyze the relationships and correlations between data sets. It also helps you to identify patterns and trends which you can then interpret.
There are several types of data analysis techniques and the next question is which one to choose. It depends on the purpose: why did you initiate the data collection process to begin with? In smart maintenance, the primary data analysis techniques are:
- Diagnostic analysis:
Why did it happen? This is what you set out to explore when you choose to conduct a diagnostic analysis. This type of analysis is useful to identify different behavioral patterns of data.
- Predictive analysis:
By using previously collected data, predictive analysis can help you predict what is likely to happen. Based on current or past data, predictive analysis makes predictions about future outcomes. The level of accuracy of this analysis depends on the amount of detailed information you have and how much you dig in it.
Data analysis is about uncovering patterns and trends in datasets. Data interpretation is when you set out to explain those patterns and trends. In this process, you must try to discern the differences between the three C’s: correlation, causation and coincidences. It’s also important that you ask yourself the following questions:
- Does the data answer my questions and how?
- Does the data help my defend against any objections and how?
- Are there any angles you haven’t considered?
If your data interpretation can withstand your questions and considerations, you’ve likely reached a productive conclusion. What’s left now is to translate your results into maintenance actions.
Translating data to maintenance actions
Now you can begin the process of turning your data into value.
Sometimes, the process is straightforward. The results of your data analysis/interpretation may show you exactly which actions to take. You can then take these into account when planning your future maintenance strategy.
In other cases, it may be more difficult to understand how you should go about translating data to action. In this case, you can use A/B testing to determine which actions improve the overall performance of your assets. The testing will then be based on insights gathered from the data analytics.
Make maintenance predictable
Translating data to action is not an easy task. Even if you have the right people, right skills, and right culture to make it happen, it will take time to implement. However, the benefits far outweigh the costs. By providing you with a unique insight into the actual status of your assets, your maintenance can become predictable. Predictive maintenance helps you to improve safety, sustainability, availability and cost efficiency.
- Dillard, J. (2017). The Data Analysis Process: 5 Steps To Better Decision Making
- Hausman, A. (2016). Translating Analytics to Action: Right Metrics, Right Time