It can be challenging, keeping track of multiple electrical devices to see which ones are currently running, and which may be about to fail. MIT's NILM (non-intrusive load monitoring) system is designed to help, using sensors and a computer dashboard to track the status of devices in settings such as factories, high-rises or even ships.
At the heart of the system is a sensor that is mounted on the outside of a single wire within an electrical circuit – that wire does not have to be cut or spliced. The sensor proceeds to monitor the current passing through the wire, and tracks the activity of different devices that are running on that circuit. It's able to do so based on telltale fluctuations in the current, that occur whenever each device switches on or off.
Based on these readings, the system can determine not only when and if each device is running (perhaps when it doesn't need to be), but also if it's drawing more current than is normal. The latter could indicate that the device has become defective.
Data is relayed to a central computer, where an onscreen dashboard features dials for each device. If the needle on any one device's dial is in the green zone, then things are going fine, but attention may be required if the needle swings over to the yellow or red zones.
The system was tested last year on the Coast Guard cutter ship Spencer, wherein two of the NILM sensors were used to monitor approximately 20 different devices. When it was discovered that a "jacket water heater" on one of the diesel engines was drawing a suspicious amount of power, the researchers investigated. Upon removing that heater's cover, smoke poured out, with severe corrosion and broken insulation being revealed. If not discovered, that compromised heater could have caused an electrical fire.
The sensor that relayed the warning did so through a hard-wired connection, although the system can also operate wirelessly. No internet connection is required, so the communications network is reportedly unlikely to be hacked by third parties.
A paper on the research, which is being led by Prof. Steven Leeb, was recently published in the journal IEEE Transactions on Industrial Informatics.
Source: MIT
