Identify noise frequency components of each appliances

Thread Starter

rexted

Joined Nov 12, 2017
1
I am a student and for my school group project I will build a power consumption monitoring device that will be connected electrical panel/fuse box.

So I will have a sensor measuring current/amps of the main wire (around 100A max input/ 100 mA input) in the electrical panel. The sensor is connected to a circuit which has microcontroller on it. Then, in the end I will make a mobile app which will display the measurement as waveform for each appliance.

My problem is on how and what circuit should I build that will able to identify the appliance and its corresponding waveform. I arrived at my solution that is to identify the appliance and its current signal by the differentiating its frequencies. Since every appliance has same frequency of 60 Hz, I will use the noise of the current signal instead.

I think I have to build a high pass filter to reject the 60 Hz and so the noise only remains. Then, I will also need several thin bandpass filters in the high frequencies so I will know what frequency components the noise has. For instance, if a dryer has noise frequencies of 2 kHz, 8kHz and 10kHz (values not real!), then I got those values with my bandpass filter, then I will know that it is a dryer. My problem is I do not know how to build this and what do I need? Help! Also, this is my very first circuit that I will build and have little to no knowledge about filters so take it easy on me.
 

ebeowulf17

Joined Aug 12, 2014
2,942
I am a student and for my school group project I will build a power consumption monitoring device that will be connected electrical panel/fuse box.

So I will have a sensor measuring current/amps of the main wire (around 100A max input/ 100 mA input) in the electrical panel. The sensor is connected to a circuit which has microcontroller on it. Then, in the end I will make a mobile app which will display the measurement as waveform for each appliance.

My problem is on how and what circuit should I build that will able to identify the appliance and its corresponding waveform. I arrived at my solution that is to identify the appliance and its current signal by the differentiating its frequencies. Since every appliance has same frequency of 60 Hz, I will use the noise of the current signal instead.

I think I have to build a high pass filter to reject the 60 Hz and so the noise only remains. Then, I will also need several thin bandpass filters in the high frequencies so I will know what frequency components the noise has. For instance, if a dryer has noise frequencies of 2 kHz, 8kHz and 10kHz (values not real!), then I got those values with my bandpass filter, then I will know that it is a dryer. My problem is I do not know how to build this and what do I need? Help! Also, this is my very first circuit that I will build and have little to no knowledge about filters so take it easy on me.
Not practical at all.
While that may be true, it's not very helpful. Any hints as to why you think it's unwise? Too ambitious for the thread starter's experience level, hard to identify unique noise signatures per appliance, wrong circuit plan for implementing this scheme? What do you see as the problem(s)? If you're going to chime in, you could at least try to be constructive and helpful.
 

be80be

Joined Jul 5, 2008
1,930
There are ways to see i don't no the workings of a SDR but it can be used to look at the RF of lights and things that make noise which will be some kind of RF but the stock SDR has to be modified.

I've read about this and it works i can see the output of the light and the monitor of the computer .
the noise they make shows up i played with the ant moving it close and you could see the change.
 

DickCappels

Joined Aug 21, 2008
5,879
One approach is to digitize the current waveform and the perform a Forrier transform to give you a list of amplitudes at various frequencies.
 

ebeowulf17

Joined Aug 12, 2014
2,942
One approach is to digitize the current waveform and the perform a Forrier transform to give you a list of amplitudes at various frequencies.
It's the best approach I can think of. I doubt that different appliances can be recognized accurately based on a few bandpass filters - even if they can, you'd need some way to find what those key frequencies are in the first place, and a Fourier transform seems like the best bet.

Just to elaborate a little, I don't think the bandpass filter idea literally couldn't work - I just think it would get unwieldy if there were more than a few appliances. I imagine that, after you've identified the key signature frequencies for any given appliance, you'd probably need at least 3 bandpass frequencies per appliance in order to recognize it. If you've got 5 appliances, that's 15 filters; 10 appliances is 30 filters, etc. Unless the number of appliances is quite small, I expect FFT to be more practical.
 
Top