AEM faculty spotlight:
Control theory is used broadly in industry – aerospace and beyond – and is applied to design systems that have predictable behaviors. Applications range from disk drives to pacemakers to space shuttles. AEM Assistant Professor Peter Seiler plans to apply his knowledge of controls – particularly in the area of fault detection – to improve the reliability of unmanned aerial vehicles (UAVs) and other areas such as wind turbines. Below, Professor Seiler discusses some details of his research.
What is your research background?
Broadly, my research is in the area of controls. My background is in mechanical engineering and I spent four years at Honeywell where I worked on flight controls, automotive active safety systems, and sensor fusion for GPS-denied navigation. One of my main projects at Honeywell was the design of the flight control system for the Boeing 787. This project focused on the redundancy management logic for the 787, i.e. the detection of faults and reconfiguration of the system to ensure safe flight.
What is fault detection?
Fault detection is used to discover failures in the system and ensure they don’t lead to catastrophic accidents. For example, an aircraft may have several computers each of which can control the airplane. If one of the computers breaks, the fault detection system should detect the broken computer and reconfigure the system to use one of the remaining operational computers. This ensures that the plane continues to fly safely after the fault. Similarly, commercial aircraft have multiple redundant copies of sensors, e.g. rate sensors, to measure the state of the aircraft as well as multiple redundant actuators. The fault detection system is required to detect failures in any of these redundant components and reconfigure to ensure continued safe operation.
How do you hope to apply your research in fault detection?
Airplanes are very reliable, which is the reason airplane crashes are so shocking. When airplanes go through certification with the Federal Aviation Administration (FAA), manufacturers must demonstrate that their design achieves a reliability of only 1 catastrophic failure for every 109 flight hours (or 1 failure in 1 billion hours of flying). My research objective is to translate some of the design principles used in commercial aircraft to increase the reliability in other domains. Commercial aircraft achieve their reliability by using redundancy - multiple copies of computers, actuators, etc. The drawback of this approach is the additional cost and weight associated with the redundant components. Many other applications cannot afford to use redundancy, e.g. unmanned aerial vehicles, medical devices, or automotive active safety systems. I’d like to increase the reliability in these domains by using algorithms and dynamic models rather than redundant copies of hardware components.
How have you been able to research wind energy at the University?
The University of Minnesota is one of three campuses that won a large DOE (Department of Energy) grant, which allowed for a wind turbine to be constructed in UMore Park (University of Minnesota Outreach, Research, and Education Park). This turbine was installed in 2011 and is currently operational. The turbine has numerous custom sensors installed on the turbine blades and foundation to enable advanced fault detection and control research. This industrial-scale turbine on campus gives us a great opportunity for experimental, hands-on research in the wind energy area.
What do you hope to accomplish through your wind energy research?
I want the turbine to capture more energy and to be more reliable. One of the issues in wind turbines is that there are a lot of vibrations and structural loads on the turbine itself. The blades tend to be 45-50 meters long with some off-shore turbines having even larger dimensions. The varying forces of the winds cause vibrations on the blades and in the drive train. This can lead to failures in the drivetrain and/or broken blades. Replacing either the drive train or a blade is quite expensive. If you’re able to control the turbine to reduce the loads and make the turbine more reliable, the cost of wind energy will be reduced.
What is one of your most memorable research projects?
One project I worked on dealt with how to control groups of cars or groups of aircraft. There are aspects that make this problem very difficult to solve. Imagine that you are at a stoplight and are in the tenth car back. The light turns green and you can see the first car go, and then the second car, but there is a long lag before you get to go. You might think that this can be fixed by putting a radar system in front of each car, and when the light turns green they should all be able to automatically go. But there are fundamental limits on the dynamics which prevent this simple solution. The lag in the situation at the stoplight is just inherent in the mass dynamics of the car.
Last Modified: Tuesday, 28-Feb-2012 12:40:18 CST -- this is in International Standard Date and Time Notation