Predictive Analytics for Vehicle Fleets

Andy Slote - Director of Customer Success for ObjectSpectrum

April 1, 2023

Economies depend on fleets of vehicles supporting virtually every product or service out there. These vehicles are a significant expense as they rack up hours of use, requiring maintenance to keep them functioning and repairs when they break down. Monitoring solutions using predictive analytics can make these tasks timelier and more effective, reducing expenses, improving efficiency, and avoiding unscheduled downtime.

Beyond the essential components of a vehicle, like an engine, transmission, and other moving parts common to most, many have additional implements like lifts, booms, plows, etc. Agricultural operations use many specialized pieces of equipment like these. Industries like construction have their particular machines, as do many other businesses. The entire universe of fleets contains unique use cases for predictive maintenance, but some solutions can be applied to most of these environments.

Many aspects are already being monitored, including oil pressure, battery level, and tire pressure, with status typically reflected by onboard gauges or, at a minimum, warning lights. Capturing this data with an interface to the vehicle’s existing diagnostics, plus a means to transmit the information to a central database makes the data more useful, both for immediate attention and long-term perspective on when a component, engine, etc., may experience its next failure.

Fluids, from the basic engine oil to others like lubricants, hydraulic fluid, etc., need to be kept at appropriate levels and replaced regularly. Sensors can provide continuous readouts, eliminating the periodic (and often forgotten) manual checks for monitoring these substances, and triggering replacement when required. Better and cleaner fluids extend the useful life of the equipment and avoid breakdowns, contributing to expense reductions and ensuring vehicle availability.

What is another example of a common indicator for monitoring most vehicles and their implements? One of the most significant is vibration. Traditional, complex vibration monitoring solutions are expensive and require highly specialized design, implementation, and interpretation. A more straightforward approach that can still provide significant value involves capturing and trending the “healthy” vibration pattern and alerting when it changes, often using Machine Learning algorithms to model the healthy patterns and detect the unhealthy ones.

Noise can also indicate problems, including the sudden appearance of noise where it shouldn’t be or a marked change in the typical sound. Detecting these anomalies can flag that an issue is about to occur with a system or component. Here too, Machine Learning can be applied to categorize the subtle differences between typical and atypical noises.

These solutions should drive maintenance personnel to change fluids, replace components, or make adjustments. Taking action after something becomes an issue that decreases efficiency, shortens the useful life, or results in a breakdown is significantly more costly and time-consuming than paying someone to do something to get ahead of a potential problem.

Although not an exclusive component of a predictive solution for fleets, tracking is one of the basic capabilities critical for success. Knowing where a vehicle is at any given time allows you to better react to an alert about a condition needing attention. In addition, deciding to service it onsite before beginning work avoids a potential interruption of much greater impact.

Regardless of where the data comes from, its usefulness falls short without an outstanding user interface. By implementing an excellent dashboard with the right level of notifications, users can keep tabs on their assets without spending time staring at pretty charts and graphs all day. Also, historical data can shape future decisions about the use of the equipment, and purchases can be driven by reliability and performance data over time. As a result, operations move from reactive to predictive, ultimately driving more revenue to the bottom line.