internet of things: fault detection and predictive maintenance

Future of automotive servicing and preventive maintenance

Several months ago there was a news item which compared how Tesla, the electric car company addressed a potential recall situation based on fault detection, with how a traditional car company addressed a similar situation. Tesla released a fix “over-the-air” or OTA – which is essentially a software update, like the kind your smartphones go through every so often. The other company required its customers to take their truck to a dealership and wait. Being the proud owner of  conventional, non-electric vehicle which has gone through more than 12 recalls in 3 years, I am all for the OTA recall fix. Many tech bloggers and business writers talk about how the future of automotive servicing will be dramatically remade over the next couple of years, based on this example. 

Although the two problems in the above recall case appeared to be the same: fire, the root causes were different. In Telsa’s case the fix was literally a software update to slow down the battery charging rate. Tesla’s engineers perhaps identified the root cause in their labs and determined the exact corrective action to take. Because the problem was suboptimal logic (rate of charge delivered), the fix was also a logic based one which could be delivered remotely. 

As more and more of automotive functions begin to be controlled by electronics, it is probably reasonable to expect future problems to be also logical in nature rather than mechanical. Clearly such problems can be readily addressed OTA and without the need for a mechanic. 

Future of fault detection

Furthermore, the identification of such root causes is also likely to happen without the need for a lab or testing. As more vehicles begin to communicate their current status to a central server, detection of faults becomes more of an empirical exercise rather than an experimental one. As internet of things (IoT) technology matures, we can see more of this in not just the automotive space, but other industries as well. Aircraft maintenance is already leading the charge in this space. 

Real time data collection, data communication and storage technologies are rapidly maturing. For example, in our whitepaper below, we describe how a “blackbox” hardware can easily transmit vehicle parameter readings over wifi and cellular connections to a remote storage from where it can be analyzed. The challenge today is in the data analytics space. And also in developing appropriate big data technologies for enabling meaningful analytics.

Data and data science

It is easy to see why. Ford recently estimated that a typical car can generate up to 25 GB of data per hour of driving. Today’s automotive market size is in the range of 12-15 million new vehicles per year. It is not hard to extrapolate this to realize that each automotive OEM will be sitting on petabytes of data which cannot be simply thrown away. It becomes extremely important to quickly convert this raw data into some sort of information which can then be leveraged to address fault issues, potential warranty problems and the holy grail of reducing/eliminating recall situations. This is one of the reasons why many OEMs are making huge investments in big data technology. However this is only the cart – the real horse is the data science which is necessary to pull it forward.

Data science can address all of these issues: fault detection and warning, warranty and recall. The challenge is to understand the physics empirically and develop processes to systematically move from raw data to actionable information. 

Originally posted on Tue, Sep 22, 2015 @ 11:25 AM

Photo by Samuele Errico Piccarini on Unsplash

No responses yet

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

simafore.ai