GE SmartSignal — predictive maintenance for military aircraft
Health and usage monitoring systems go a long way towards making the maintenance of military platforms more intelligent based on typical behaviour and live monitoring of threshold values. GE claims its SmartSignal product goes a step further, predicting a failure weeks or months in advance using patented algorithms that monitor the behaviour of an individual piece of equipment.
Having aircraft and vehicles out of action costs the military dearly, both financially and in terms of availability, not to mention failing to identify a potential breakdown could lead to equipment breaking down mid-mission, risking the lives of personnel. Health and usage monitoring systems (HUMS) are now a well-established method of adding intelligence to the maintenance process, using sensors to track the behaviour of different parts and raising an alert if it reaches a critical point -a threshold helicopter engine temperature, for example.
However, once that critical point is reached, that helicopter has to be taken out of use immediately and repaired out of its normal service schedule. Ideally, a system could spot that the outage is likely to happen days or weeks in advance, so maintainers could focus their resources on preventative action during scheduled services rather than checking systems that are not broken or likely to break.
General Electric (GE) believes its SmartSignal tool, which has already proven successful in the gas, energy, civil aviation and mining industries, could offer the military community just that ability.
Platform-specific maintenance models
Todd Stiefler, senior product manager for embedded services with GE Intelligent Platforms, says SmartSignal is very different from other condition-based maintenance systems.
"Most of the work that's been done in this field has looked at big populations of assets," he says. "Previous systems would take the Challenger tank, for example, and look at all the maintenance data for every tank in the field for five years, and generate equations and hypotheses about how this equipment behaves over time. So we know with 95% confidence that a Challenger tanks transmission has this particular failure mode which happens 1.5% of the time, or every thousand hours."
Stiefler explains that while that type of information is useful for figuring out failure modes, setting maintenance budgets and figuring out how many people are needed to staff the depot, it does not identify which specific platform is going to have which exact issue and when.
"SmartSignal generates an individualised model for each asset, so each helicopter has its own model," he says. "We start with a similar general template for that platform, but we pull data from that specific helicopter or tank. Each asset has its own history and its own deployment schedule, and has operated in different environments. After a while, they all become unique instances."
Elon Musk has rattled the US military space launch market, challenging the monopoly of United Launch Alliance by suing the US Air Force for the right to compete with his company SpaceX.
SmartSignal treats every asset as an individual, taking around two weeks to monitor its behaviour though standard on-board sensors to create a multi-dimensional picture of how those values relate to each other when the asset is operating in a healthy state.
"It says if the engine temperature's at this level, then the axial pressure should be here, and the bearing temperature should be here; it relates all those variables in space," explains Stiefler. "It's not saying that this helicopter should never have an engine temperature of 300 degrees; it's saying given all the other values in the system, the engine temperature right now should be 175 degrees, but it's really 185, that's weird, there's something going on we need to investigate."
While the military is embracing big data, SmartSignal champions small data. Each platform generates a lot of data from its sensors but it is used to learn about that specific item rather than being used to hypothesise about a general case. That is not to say big data is ignored altogether - GE uses it to build its templates and upgrade its models.
"We have analysts looking at the output of the models for thousands of assets," says Stiefler. "They start to get smart at a big data scale because they're seeing so many different assets, and they code that intelligence into the model."
Industrial internet and embedded intelligence
This fits into GE's grander scheme to create an 'industrial internet' of the machines all the different parts of its business build - oil and gas, energy, mining, aviation - all of which offer embedded intelligence and computing capability. While these have been generating masses of data for years, it is only now that analytical tools such as SmartSignal can enable customers to get better value from them; value that could transfer horizontally to high-value military assets.
According to Stiefler, one major commercial airline customer reports that since implementing SmartSignal, the number of times a pilot has left an airport, realised something was wrong and had to land somewhere other than its destination has gone down by 50%. In military terms this would translate into not having to abort a mission, or having to execute it with one fewer helicopter than should have been available.
The Swiss Government has agreed to spend billions of dollars replacing the air force’s 1970s-era fighters with modern Swedish jets.
However, switching to a military market for SmartSignal would represent something of a step-change for GE, and it could take at least a year before the company is ready to start bidding for military business.
"Our vision of what we want to achieve is to start co-creating a solution with the end user and discover how it delivers value for them," says Stiefler. "We're having discussions about how to quantify the value that SmartSignal would bring and we want and expect to have to prove that in a field trial with the government - what we call a test drive."
Following these test drives, which would ideally take place on at least one each of naval, ground, and rotary and fixed-wing air systems, GE would then be in a position to consider a number of options for deployment. As well as potentially bidding directly for government contracts, the company could also license the software to the government and train them how to use it, or partner with a prime contractor and make that part of its own offer to the government.
GE believes than three factors in its favour in an era of commercial off -the-shelf procurement is that SmartSignal is not tied to any one manufacturer's equipment, it does not need dedicated on-site analysts, and in most cases it works using the existing sensors on the platform.
"Those are big complaints we've heard from the military about previous approaches," says Stiefler.
"Either it's tied to someone's proprietary solution, it requires engineers to put a bunch of additional expensive or heavy hardware on a system where there isn't the budget or weight allowance, or they need to hire legions of on-site data analysts to make sense from the data. You don't need any of that stuff for SmartSignal."
If these test drives go GE's way, SmartSignal could prove to a new military audience that, through intelligent maintenance, prevention really is better than cure.