SO WHAT IS NILM?
First of all, NILM is a Non-Intrusive Load Monitoring Techniques. Within the last years these techniques become more and more popular because solutions that use them don't require multiple sensors at the household. Also it's cheaper to have them for both consumer and provider.
With NILM techniques you don't need a plug-in meter, but just one sensor per home.
NILM was invented in 1980s by George W. Hart, Ed Kern and Fred Schweppe from MIT but only within the past years due to the massive improvement in software and hardware development it became possible to implement the techniques.
When using NILM only one sensor attached to the phases is needed to receive disaggregated information. Basically the sensor is analyzing the changes in voltage and current in the house or apartment and detects what devices are working, using different approaches.
There are many approaches that can be used. To detect devices, different products can use steady-state signatures or transient signatures. There are also hybrid approaches where detection is based on analysis of same-class appliances or such indicators as active and reactive power, harmonics, current waveform etc.
Majority of the products that use NILM classify devices into such categories:
- Permanent (something like a stationary phone that is always on)
- On/Off (Like a toaster)
- Multistate (like a coffee machine)
- Variable (like a TV)
BUT HOW IT DETECTS HOME APPLIANCES?
By using machine learning algorithms, the system can classify a number of appliances used at home. There are supervised and unsupervised techniques for sensors to remember home appliances. The majority of sensors use supervised learning, where users have to add appliances manually into the application so it will remembers the pattern of energy usage. -
SO WHAT IT ALL MEANS?
It means that modern software as well as hardware which is in combination called NILM is the technological answer to the old era with "not-so-smart" grids and regular meters. Modern energy monitoring systems that use NILM can disaggregate and gather specific information about home appliances which were used, how people use them and how energy is being consumed or wasted. Modern energy monitors can help utilities to create dynamic pricing schemes that will be cost efficient for their clients, eliminate grid overloads and waste of resources, which is very important for creating a real smart grid.