The Inductive Monitoring System (IMS) is a new computer program that monitors gyroscopes that keep the International Space Station properly oriented in space. IMS detects warning signs that precede a gyroscope's failure. Engineers will add the new software to a group of existing tools to identify and track problems related to the gyroscopes. If the software detects warning signs, it will quickly warn the space station's mission control center. David Iverson developed the software at NASA Ames.
NASA Tech Briefs: How long have been with NASA and what were some of the projects you worked on prior to getting involved with the Inductive Monitoring System?
David Iverson: I started with NASA as a co-op student back in 1985, so I've been here a while. I spent a lot of that time working in health monitoring, automated diagnosis, artificial intelligence, things like that. And I've been involved, on and off, in space applications. I did some space shuttle payload monitoring applications a while back. Aeronautical applications, looking at helicopters and some of the jet aircraft that NASA has, and mission control types of applications. A little bit of everything that NASA is involved with that might benefit from system health monitoring.
NTB: How does the Inductive Monitoring System work?
Iverson:It's a tool that's supposed to help, say, a mission controller or someone who's in charge of making sure that something is running properly. It keeps track of how well things are working. It lets them know if something is not working as expected. The Inductive Monitoring System is actually a tool that looks at archived data, saved data from your system, and analyzes that to build up an idea, or a model, of how we expect the system to behave. It characterizes what normal behavior looks like based on what it sees in that data.
After that you can feed IMS real-time data coming from the system or look at other archived data and compare it to the model or characterization that you've built. IMS will give you an indication – hey, this looks like it has before, the system behavior hasn't changed, or it might deviate somewhat from what you've seen before. You can think of the output from the IMS as a sort of a distance from what we expect to see from a normally operating system. A small distance means your system is looking like what we expect from the training data. If that distance number increases you're moving farther and farther away from what you expect, from what you've seen before, and that could be an indication that something is not working quite right in your system and you should probably check things out.
NTB: How were you that job before your team developed this new software?
Iverson: Well, there are a number of different techniques that have been applied. A lot of them look at individual parameters from whatever system you are watching, like temperatures, or electrical current, and let you know if that parameter is running high or low. You can set limits that you expect that parameter to stay within. Those techniques are still in common use; they're still a good basic approach to take.
IMS looks at multiple parameters at once and figures out how they relate to each other, how each part of the system behaves relative to the other parameters. That can give you a better picture of what you can expect to see in the system and you might be able to catch some unusual activity or anomalies before any individual parameter reaches a limit.
NTB: I understand it took five years for you to develop this software. What were some of the technical obstacles your team had to overcome?
Iverson: Well, it's been, on and off, about five years that the software's been evolving over time. We've applied it to several different projects and platforms. Part of the trick with a system like this is, you need to find good data to work with to build your model or system characterization, and it helps to have a person with system experience to point you in the right direction and explain what the data means. Accessible data and knowledgeable people can be scarce resources.
But we've come across a few different opportunities that had nice clean data and helpful folks to keep us oriented. I guess that was probably the major challenge that we ran across – getting access to good, clean data.
NTB: How rigorous was the beta testing program before you felt comfortable enough to install it in the Space Station?
Iverson: Well, of course, the Johnson Space Center has a verification process for each piece of software that's used in mission control. And we are in the middle of that right now. What's involved is making sure the code does what you expect it to do with different data inputs, making sure that it runs properly on the mission control computers, and a lot of the evaluation has been done by the controllers themselves. We've been running on the ADCO (Attitude Determination and Control Officer) console in the mission control room for the Space Station, in an uncertified mode, so the software is there, up and running, and they can watch it and see how it works. But since it hasn't been fully certified yet, they don't use it for their actual control decisions.
We're hoping for certification in the next couple of months. A lot of the process depends on the availability of some very busy people at JSC (Johnson Space Center) to help get the whole package put together. If all goes well, we may be able to get it certified before the next shuttle launch in October. That's our hope.
NTB: You said it's running on the ADCO console?
Iverson: Right. That stands for Attitude Determination and Control Officer. They're the folks who are in charge of keeping the Space Station pointed in the right direction.
NTB: This software has been used in F-18 fighter jets; it's been used by the space shuttle's leading edge impact detection system; and it's been used for electric power plant and water quality monitoring. What other commercial applications do you envision?
Iverson: Well, the nice thing about the software is it doesn't care what it's monitoring, so any type of application where you have a number of data parameters that you need to keep track of is a potential target for the software. There have been discussions about power grid monitoring, production lines, petroleum refineries, things like that. Any type of situation where you have a control room with people or computers to watch over the system and make sure that everything that is going on is what you expect to be going on.
NTB: Has anyone expressed an interest in licensing it yet, or is that still off in the future?
Iverson: We actually already have two licenses issued. One to a company from Half Moon Bay, California, iSagacity is their name. They specialize in remote data analysis and monitoring products. Another company, in Chicago, called InStep Software, has licensed it to integrate it with their data analysis and management products.
NTB: Now that the software is approaching certification and the project is just about at an end, what new projects are you and your team working on?
Iverson: Well, we're thinking this isn't the end; this is really sort of the beginning of the curve. Our first mission control deployment was a proof of concept task to see how well it worked in a mission control environment, and how helpful it was. That's turned out pretty well. We're looking next year to deploy the IMS system in other mission control positions. There is quite a bit of interest from other areas there.
We're also looking at some aircraft data analysis, watching for unusual occurrences in behavior or performance of commercial aircraft flights. This could help aircraft operators and controllers refine their procedures to help things run more smoothly and efficiently. And we're looking for potential applications in pre-launch ground checkout and processing of new launch vehicles supporting NASA's Constellation program that will take us to the Moon in the next decade.