Tech Briefs: What inspired you to work on this project about the bridge sensor?
Dr. Fei Lu: I am the co-chair of this project, which was started by Doctor Ivan Bartoli in our civil engineering department. Our vision on this project is to utilize a wireless sensor to monitor the health, the real time status, of bridges. And based on this collected information, we could estimate the health status of the bridge to do any necessary maintenance.
For example, there's an accident on U.S. Route 95, which is really serious. If we could have wireless sensors there to monitor the bridge status, that could be helpful to avoid any future problems.
Tech Briefs: Let me get into a few technical details. First the power system, about how much power does this take?
Lu: The sensor itself is low-power and energy-saving. The power for the sensor board itself is in the range of one watt or even less. We use a combination of solar power and a battery. The problem is we don't always have 24/7 sunshine, so we need energy storage for times when the weather is bad. For example, we might have rainy days for a week. So, we have designed our power system design to always be prepared for the worst case. Our battery can, on its own, support the sensor to work continuously for two weeks without any sun. Once the sun returns, our battery could be fully charged within the next few days. That was a major concern for us when we designed the solar power system for the sensors.
The other thing is the practical implementation. Because everything is installed outside, we had to consider severe weather like cold temperatures, wind, and rain, so we need a waterproof design. Our box and all the connections in the electrical system are designed to be robust enough to operate in severe environments. Last winter we ran it outdoors for two or three weeks and everything worked perfectly. We tested it in December, before Christmas, when the weather was really cold, and it worked well even in snow. The other thing is, when there’s snow in winter, on average sunshine is only around for two hours or so. We considered all these conditions when we designed the power system and the battery.
Tech Briefs: So, the solar panel charges the battery, and the battery runs the system, is that it?
Lu: Yes, that’s right.
Tech Briefs: If the box is sealed against the environment, how is the output data transmitted and how does the sun get to the solar panel?
Lu: The box is plastic — it’s transparent to the gigahertz-range electromagnetic waves we use for our transmitter. As far as the solar panel, it’s mounted outside of the sensor box, and is connected to the rest of the system by wires through a watertight opening.
Last summer, my co-author, Doctor Bartoli, sent his students to California for a real field test for several weeks. During that time, they worked together with a construction company, that was moving a highway bridge. Because of the layout of the highway, they needed to cut the bridge and then move it to a different location. They wanted to make sure there was no damage to the structural integrity of the bridge during the cutting and moving. So, they attached our box to the bottom of the bridge to monitor the strain, forces, and stress displacement. They used a large electric saw to cut the structure, which needed water, which was spraying everywhere, to cool it down. And our sensor had to be close to the cutting point to monitor all the signals. We were very satisfied that it worked well even under these severe conditions.
Tech Briefs: I would think that every application would be unique, so in each case you would have to decide where and how to attach the sensor.
Lu: Yes, but it's very flexible. You are free to choose any location that you think might be particularly sensitive to stress. It can even be used on a tall building.
Tech Briefs: Where would the signal be received and analyzed?
Lu: It could ultimately be a smart phone, but right now we are connected to a computer. We use what we call a communication gateway, which can both transmit and receive signals from the wireless sensor over distances up to the kilometer range.
Tech Briefs: Would you transmit intermittently to save energy, or continuously?
Lu: It depends on the programming. You can program it to receive the signal continuously or you can program it to ask the sensor to sleep for a while to save energy — you can ask to send a signal every few minutes, or every hour, or every other day — it’s very flexible.
Tech Briefs: Could you tell me a little bit about the displacement potentiometer?
Lu: We're using an electric circuit to detect everything. Our sensor is used to convert all the physical or mechanical signals, like vibration displacement, into electric signals. For example, our strain gauge is based on resistance changes. We use a Wheatstone bridge to detect the changes and a custom-designed instrument amplifier to amplify them. And that's all within the sensor package.
Tech Briefs: It seems to me is that data analysis is very important. You know, coming up with a baseline and deciding the significance of changes. I would imagine if it's a bridge, traffic patterns vary, big trucks tend to go on certain days and not others. Do you use AI in your analysis?
Lu: Not yet — our work has been more on the signal side — but we plan to do more work on data analysis. But even for the raw data in the signal, which is my focus, I have some techniques for signal processing. For example, we have to filter the signal because it contains a lot of noise. We do the filtering in both hardware, with an LCL low pass filter, and in software, with digital filtering using MATLAB. One thing we need to pay attention to is the bandwidth of the filter, because we don't want to remove the useful signal, we only want to remove the high frequency non-useful noise. On the software side we export all the data into MATLAB for some digital filter functions. Then we can get a clear signal. We also tried a lot of different frequencies, but right now our sensor can give us a detection frequency of 128 Hertz.
Tech Briefs: So, once you have good signals, how do you decide what's normal for the bridge and what's not normal? And the normal for one day might be non-normal for the next.
Lu: That's a good question. We haven’t gone into that level of data analysis yet.
Tech Briefs: What are your plans for the future?
Lu: I had a meeting with New Jersey Department of Transportation (NJDOT). They have a plan to investigate the safety of their transportation systems, including the highways and bridges. I presented this wireless sensor box monitoring system to them, and they are very interested. So, we're talking about a proposal, which will hopefully turn out to be a project later this year. My plan is to work together with the NJDOT to install sensors to collect ongoing data on real bridges. We will be able to analyze that data to detect patterns, for example on different days.
One thing further thing we’ve talked about was using drones together with our sensor box for better monitoring. If the sensor box is attached to a bridge, you can have 24/7 continuous monitoring. But sometimes, if the signal is really bad, you can send in a drone with a camera to see what’s happening. You could then decide whether to send a person to further inspect the structure.
Tech Briefs: Would you produce the sensor boxes for the New Jersey project at the university or would you use an outside vendor?
Lu: Actually, for the sensor box hardware itself, we could produce it here if there are not too many.