Edge disaggregation
The energy disaggregation described in the previous sections is performed in the cloud with a relatively large set of historical data. In case users have connected NET2GRID IoT hardware to their meter, some events detection and disaggregation can also be performed in middleware.
NET2GRID is pioneering real-time analytical and ML models on distributed hardware. The accuracy of these models and the set of appliances that can be recognized in this way are offcourse much less than the more advanced models that are used in the cloud. However, edge analytics has some advantages and gets better with higher data granularity, compute power and RAM on the device. One advantage is cost reduction. All disaggregation steps that can be taken in embedded software on IoT devices do not have to be performed in the cloud.
Another advantage is that disaggregation on the edge is real-time available directly at the moment the user requests it from the app.
The platform CE-API call instantaneous/appliances proxies the real-time disaggregation performed on the edge:
An example return from Ynni would be:
{
"status": "ok",
"data": {
"timestamp": 1671714048,
"electricity": {
"power": 1804,
"L1": 25,
"L2": 1857,
"L3": -72
},
"appliances": [
{
"power": -313,
"appliance_instance_id": "Solar"
},
{
"power": 223,
"appliance_instance_id": "AON"
},
{
"power": 29.75,
"appliance_instance_id": "REF"
},
{
"power": 2,
"appliance_instance_id": "LIG"
},
{
"power": 8,
"appliance_instance_id": "ENT"
},
{
"power": 0,
"appliance_instance_id": "WM"
},
{
"power": 0,
"appliance_instance_id": "IH"
},
{
"power": 1854.25,
"appliance_instance_id": "EV"
},
{
"power": 0,
"appliance_instance_id": "APP"
},
{
"power": 0,
"appliance_instance_id": "OTH"
}
]
},
"data_type": "current_power"
}
A creative visualization of this data is performed in this Ynni view:
Updated almost 2 years ago