Predictive Maintenance Part 1 of 5: Predictive Maintenance in the Automotive Industry

Man working on an HP ElitePad 900 with an ElitePad Expansion Jacket in a factory.
Predictive maintenance is a hot buzzword in the automotive industry, but what is it really, what does it take to implement it, and what are its benefits? We’ll be exploring these questions on this blog over the next few weeks, but let’s start with a definition.

First, it’s helpful to take a look at how maintenance has been done traditionally. Whether you’re maintaining a vehicle on the road or a machine on the factory floor, traditionally, maintenance has been done on a fixed schedule, such as replacing tires every 50,000 miles. This planned maintenance schedule is based on what is essentially “tribal knowledge” and does not take into consideration the unique circumstances of the machine or vehicle being maintained. By following a general purpose schedule, you risk wasting money if you replace a part that’s still good, or equipment failure if you’ve waited too long – which could have disastrous results.

Predictive maintenance is about finding the sweet spot that lets you get the most life out of your equipment while minimizing the risk of failure. It involves gathering large quantities of data – such as maintenance records and data from sensors on the equipment – analyzing the data, and creating a predictive model to determine the optimal time for maintenance tasks to be performed on each individual piece of equipment.

Steps to implement

To implement a predictive maintenance program, you need to follow these basic steps:

First, it’s important to define the business issues you are trying to solve. You may want to reduce downtime of robots in the plant or improve the customer experience by providing a vehicle that spends minimal time being serviced. (Over the next couple of weeks, we’ll discuss in more depth the business benefits of predictive maintenance on the factory floor and in the final product.)

Next, you’ll go through the data discovery phase to identify the data that will support addressing your business issues. Some of this will be existing data, such as maintenance or warranty records. Often, this data is structured data – found in databases or spreadsheets. You may determine that your existing data is sufficient to get started, or you may identify additional data sources to add now or add later – such as sensors on the equipment, free text in maintenance records, design specs, test data from failed equipment, or even comments on social media or Google searches, which may help identify problems early. Some of this data will be unstructured – free text or even voice or video data.

The next step involves using data analytics to make sense of the data, and then figuring out predictive rules that will become the basis for your predictive maintenance model. Once the model is built, you’ll test it and then operationalize it in a production environment – with a continuous learning loop to update the model based on ongoing results.

HPE can help you through this process every step of the way.

We’ll be discussing this topic in more depth over the next few weeks, so please follow us to learn more. For more information, read about our “Avoid unexpected downtime by predicting failures” brochure.

Related Links:


KenKen Elliott has over 25 years of experience delivering analytic solutions. He has been with Hewlett Packard Enterprise for the past 12 years. He is currently the Global Director of Analytics within Enterprise Services. Ken holds a Ph.D. in Industrial Psychology and is based in Austin, Texas.