Detecting Running Appliances by measuring ONLY Mains I or V

Go To Last Post
32 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi, I am thinking about ideas for final project of my BE Degree. I want to know if this is possible somehow or if someone has done something similar. I have googled it and found this (https://drv.googlecode.com/files/genetic-dynamic-algos-for-nilm.pdf) though I haven't completely read this YET. Anyways, this is what I have in mind.

I want to make a system which monitors and Logs the MAINs voltage and current, using AVR, CT for current measurement and some voltage divider and ADC setup for voltage - it will Log data on an SD Card (every 5 minutes or more frequently) and when connected to PC the data can be analyzed in-depth and kept as history record.

Now, the main part is to find out what devices are running by JUST analyzing the data. Is it possible to do this...? I don't want VERY accurate, but should be able to find the bigger appliances(fridge, TV, etc) accurately. this has to be done on the PC side, so I don't think processing power will be a problem.

If possible, please guide me in the right direction, maybe give out proper Google Search terms so I can search more myself. and as always, THANKS in advance for the help.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MORE INFO:
>> This is a BE / Under-Grad Project.
>> The system has to be NON-intrusive to the HOME and ideally, there should be NO wiring etc connection to the MAINS
>> However, the system may be trained initially by manually turning ON a certain appliance and letting the system record the variation in Load and startup pattern etc
>> I think that by measuring at higher frequencies more info can be obtained about the Loads, but that will increase the Data Size that the AVR has to store at each iteration on the SD Card

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You measure current with a current transformer. Look at a regular electrician's current meter. The one with the lobster claws.

Mains voltage tends to be either present or not. Well, it is either 230V or 0V in Wormshill. We don't get low voltages.

If your power is liable to be low voltage sometimes, log the voltage value too.

Be sensible about your sample times. An SD Card gives you limitless storage. How frequently do you turn a fridge on or off ? 5 minutes seems fine for sampling.

Good Luck.

David.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

THANKS for the reply.

We have much MAINS voltage problems and frequent power outages, so the voltage needs to be measured and logged too and due to power outages the system needs to be battery-powered with auto-charge when electricity is available.

I think a once a minute sampling will be good enough, I mentioned higher sampling rate as when a fridge turns on, the spike it generates has a sort of signature to it (I have read this in a similar project blog, but can't find the link at the moment) - so i was hoping to catch that signature and detect the fridge.

But my primary question is, is this POSSIBLE, to actually detect the devices running in a home by only measuring the MAINs... ??

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Of course you can. Just put a clamp-ammeter over the mains supply to your house. Observe the reading when a fridge turns on and off. When lights are turned on. When a kettle boils / oven cooks.

You need to log an averaged RMS value. A 2kW kettle or a 6kW hob or a 3kW immersion heater is easily detected. After all, for a typical 1kW average consumption, a high-power appliance makes a big difference.

One 20W lamp altering 1000W -> 1020W is detectable. A 1W phone charger is not worth bothering with.

If you want to identify different appliances, you will be monitoring every 20ms or so. Even a lamp will have different characteristics. You don't log every 20ms though. Unless you intend the PC to do all the analysis.

Oh, if you do want to do intimate monitoring, log the time+delta rather than a value for every sample. Most of the samples will be identical for long periods. Let's face it. How often do you switch lamps on/off? Even the thermostat on a fridge will only do 100 on-off cycles a day.

David.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If I have both a light bulb and an LCD TV that consume 100W how are you going to tell them apart? They are both 230V and they both consume 0.43A so how do you differentiate them?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

david.prentice wrote:
Oh, if you do want to do intimate monitoring, log the time+delta rather than a value for every sample. Most of the samples will be identical for long periods. Let's face it. How often do you switch lamps on/off? Even the thermostat on a fridge will only do 100 on-off cycles a day.

David.

Good point, am making notes. THANKS.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I bet that the start-up of a lamp and a TV is very different. Once running, they will be difficult to tell apart.

Your "100W lamp" is likely to be incandescent with a high inrush current. The "energy saving" replacements will also take more current at start up. Pity they don't actually supply many lumens though !

If you are really technical, you can measure the power-factor of each load.

David.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How about this: Attach an AVR to each appliance to be measured. The AVR bd has a dipswitch to select an address. When the appliance turns on, the AVR starts injecting a spike to the line shortly after the zero crossing. A human with a scope can tell which appliance is on by counting spikes. 1,2,3,4 etc. A smarter human can make another AVR count em and log whats on and home much juice its eating.

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

An idea I would like to throw into the pool is a 'teachable' system. Put the measurement unit in 'learn' mode. have it measure a spike when the device turns on. Tell the unit that is a "refridgerator", put it back into learn mode and turn on the tv, tell the system that spike level was a "TV" etc. The only downside is when two devices turn on at the same time for example.

I would rather attempt something great and fail, than attempt nothing and succeed - Fortune Cookie

 

"The critical shortage here is not stuff, but time." - Johan Ekdahl

 

"Step N is required before you can do step N+1!" - ka7ehk

 

"If you want a career with a known path - become an undertaker. Dead people don't sue!" - Kartman

"Why is there a "Highway to Hell" and only a "Stairway to Heaven"? A prediction of the expected traffic load?"  - Lee "theusch"

 

Speak sweetly. It makes your words easier to digest when at a later date you have to eat them ;-)  - Source Unknown

Please Read: Code-of-Conduct

Atmel Studio6.2/AS7, DipTrace, Quartus, MPLAB, RSLogix user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

david.prentice wrote:
If you are really technical, you can measure the power-factor of each load.

David.


NOTED, another good point to consider. THANKS. and Yes I am also hoping for all devices to have some sort of signature to their ON behaviour.

bobgardner wrote:
Attach an AVR to each appliance to be measured

The idea is to be as less INTRUSIVE to the home as possible, putting an AVR with each appliance, even the major ones is not gonna look good in my project proposal.
Though I was thinking of making SMALL AVR Modules to hide inside the wall-sockets which can check status of ALL switches and even possibly then control the switches - but that's for another project, this project is for analysis, control comes later.

jgmdesign wrote:
An idea I would like to throw into the pool is a 'teachable' system. Put the measurement unit in 'learn' mode. have it measure a spike when the device turns on. Tell the unit that is a "refridgerator",

This is ONE of the methods I intend to use for identification of the loads, the other is to keep track of similar loads and show them to the user during analysis on PC and ask the user what the particular load is.

THANK YOU ALL... keep those ideas coming..

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I found this while Googling a BIT
http://www.rotwang.co.uk/projects/kettle.html

There was another link, which I had seen a few months back, still finding it.

P.S. my post count just went over 50 so am now officially a wannabe :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Consider this: you can't measure voltage changes directly unless the wire to your premises is long and the devices take huge currents, so current measurement is your only option. However, the current taken by individual items is not constant, both from the operating characteristics of the device and from variations in the supply voltage.

Even a nominally resistive device like an incandescent bulb will have a high-current startup phase while it heats up; inductive or capacitive loads will show a current which is not in phase with the voltage. Switch mode devices - computer power supplies, phone chargers etc - take their current close to the peak of the mains rather than continuously...

While you may be able to identify a *class* of device being turned on or off, I think you're going to have difficulty identifying a particular item.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Running appliances should be quite easy to detect, their legs moving fast would give them away. :mrgreen:

John Samperi

Ampertronics Pty. Ltd.

https://www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The project sounds fun. The analysis will be a bit hairy. All the same, it should impress a lay person.

Anouncing (correctly) that a fridge or immersion has switched on/off will look good.
Announcing that a bedside lamp has switched off when it was a living room lamp would require intrusion on your house wiring. Any false guess would look bad.

David.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thank you guys so much, I have been discussing this here and with a few friends and seniors as well, looks like I will need to add in the CONTROL part as well in the project, the analysis will be only half of the work.

Now the talks have gone to things like having either one AVR for each appliance or having an AVR inside the wall-switchboards, those AVRs would control the electrical devices and would report back to a central AVR via Power Line Communication - is node-based power-line communication possible - the central controller could then report everything to a PC or make the controls available ONLINE, or on bluetooth etc and maybe even a mobile App...

looks like that escalated quickly..

BTW, again THANKS..

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It's definitely possible to communicate from node to node through the power lines, with certain caveats. X10 does half of what you're talking about--you plug any appliance or lamp that you want to control into one of their plug-in automation modules, which in turn plugs into the wall. Then a central control station is able to send on/off or even dimming commands to the modules from elsewhere in the house. I'm not sure if any of their remote modules can talk back to the base station, but with the right protocol it should clearly be possible. See also the various ethernet over powerline systems that are available.

The downsides to powerline communication, aside from the obvious ones of getting a signal though at all, are that communication across phases or poles of the mains supply can be an issue--whether that's the two poles of a North American-style center-tapped supply or three phases in a large apartment building (or I believe in some European countries 3P is common to households as well?)--getting a signal from one to the other can require special consideration. And depending on how the neighborhood is wired, you may have issues with signals propagating through adjacent houses as well, especially in for instance apartment buildings, where multiple homes may be supplied from a single transformer--maybe not an issue for your prototypes, but certainly would be an issue for a commercial implementation!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You can know more if you will measure also phase shift between voltage and current, and harmonic distortion of the current.
Different devices have different phase shift between voltage and current - motor is inductive, bulb is resistive.
Also different devices have different harmonic distortions of the current.

Run ADC on both voltage and current every 1 ms for example. Log all measured for analisys.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I know Intel have been working on the problem you describe, and according to the article below, seem to have had some success.

http://www.itworld.com/business/...

Given theirs can be plugged into any old outlet and see the whole house, it's clearly not measuring the real current to appliances. I figure it must just be recognising device signatures via noise on the V curve. (presumably after a training session like you described above)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You could check Echelon with their LonWorks nodes.

No RSTDISBL, no fun!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I don't believe that within the $$$'s budget OR your time frame you will achieve what you want. Ie. the detection of individual devices amongst the possible number of devices that you might have in the average home.
It would be possible using an MCU, a CT & an X10 interface, but your $'s & time budget will be blown.

There are many power monitors that will measure the total mains consumption. You could easily implement power & light separately if they are fused separately. There are many examples of how to do that.

Your BE degree is in what field? Electrical engineering, Electronics Engineering, Instrumentation?
Do you actually have to build it & get it working (as in the rubber hitting the road) or is it just a paper exercise (as in pie in the sky)?

Charles Darwin, Lord Kelvin & Murphy are always lurking about!
Lee -.-
Riddle me this...How did the serpent move around before the fall?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This sort of 'diagnostics by current measurement' also applies to vehicles and the military has a Vehicle Test Meter that can measure the compression of each cylinder of an engibe while cranking by measuring current during each compression stroke, and you can measure the RPM by counting the AC peak spacing on the altenator waveform floating up around 12V. Just some alternate ideas for you.

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Does it give relative compression or absolute? How does it know which cylinder is which?

Charles Darwin, Lord Kelvin & Murphy are always lurking about!
Lee -.-
Riddle me this...How did the serpent move around before the fall?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I guess it syncs to the ignition system. If used on a diesel, a clip-on pickup element on the feed line to one of the injectors works too.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

well this has gone from home power metering and analysis to military vehicles and then the engines inside those vehicles... carry on, I am learning stuff either way..

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

LDEVRIES wrote:
Your BE degree is in what field? Electrical engineering, Electronics Engineering, Instrumentation?
Do you actually have to build it & get it working (as in the rubber hitting the road) or is it just a paper exercise (as in pie in the sky)?

My degree is actually in Industrial Electronics. but our projects may wander a bit away from the industrial side but the electronics remain (the project should have aspects of one or more of these: control, power, embedded systems, communications, PLCs etc)

At the moment this idea is one of many things that I am considering and I have ample time to decide on one, so I put it here for discussion. Though this seems interesting enough to do as a side-project and not for the degree specifically, usually my side projects don't get out of the paper designs... :/

I wish to do the measurement and analysis part working and have some sort of prototype setup ready which shows the capability of controlling the appliances.

I hope I have answered quite much in detail.

BTW, Many THANKS to all who have answered and will be answering in future...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Loosely related, I can tell when my pool skimmer basket is clogged with leaves, because the power consumption on the pump goes down the more it clogs. I've got a current transformer and monitor on that feed.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Brutte wrote:
You could check Echelon with their LonWorks nodes.

http://www.echelon.com/

GREAT STUFF... (Y)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

zaidpirwani wrote:
GREAT STUFF... (Y)

They have a starter kit consisting of one USB network interface (something like a pendrive), one PL3120 based node and one PL3150 based node. With some demo software. Mind these chips already have 24kB OS in ROM so 12kB application limit is not a problem, typically 2kB will do:
http://store.echelon.com/item.as...

The chips, "tripple-core" 8-bitters:
http://www.echelon.com/products/...

Programming language is "almost C":
http://www.echelon.com/products/...

EDIT: Seems this kit is a modern version of what I have seen, it has PL3170 node instead of PL3120 that was shipped before.

No RSTDISBL, no fun!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have heard they already have such devices. These are being used for SmartGrid applications. The utility companies thus can give you a graph at the end of the day showing you how much your bill goes in what appliances etc.

Yes it would be possible to use some signal processing techniques on the mains voltage and current levels to determine what appliances are being turned on. Ofcourse you will have to train the system by turning on each appliances one by one and letting the system remember the disturbances caused in the electrical system as each appliance is turned on one by one.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ok, so am now thinking of making a small prototype, at first I was of the opinion that I will be needing to make the complete solution. i.e. all the work regarding sending data over the Power Lines.

Just found out that there are Power Line Communication Modem ICs available which do all the heavy lifting and give controller a simple Rx/Tx interface.

Any suggestions regarding those ICs, your experiences maybe.. ??

These are the ones I found

I have BOLD 2 of these, or should I start a NEW post concerning Power Line Communication/Modem ICs

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I am thinking about ideas for final project of my BE Degree.
Any suggestions regarding those ICs, your experiences maybe.. ??

Perhaps you should just do some reading & re-search. After all it is a BE degree, not a degree in team working!
Quote:
should I start a NEW post concerning Power Line Communication/Modem ICs

No! Just write an article when you are finished!

Charles Darwin, Lord Kelvin & Murphy are always lurking about!
Lee -.-
Riddle me this...How did the serpent move around before the fall?