I 've decided to make a WSN for Forest Fires as my thesis project.
Soo, after reading relevant posts on this and other forums and researching some papers, I wanted to ask for help about the detection part:
If I rely solely on temperature and RH readings, the fire will likely be detected during the spreading phase, rather than the ideal smoldering phase.
That's the reason I chose BME688, but How should I record my data? Spark fires?
I 've read enough papers, about fire products and stuff. The main suspects are CO, CO2, C2H6, C2H2 and some VOCs.
Some info about the thesis: The WSN will use LoRa (using RFM95W). The forest is divided into "territories", each territory will have 1 concentrator and 8 nodes around it, every concentrators then sends the data to a gateway. Soo my task is to build just one of them.
My professor mentioned that the Forest environment poses challenges for propagation, as do the GPS modules I plan to include in the final circuit. However, I will handle this, one way or another.
Perhaps bit of study of topography would add to your understanding of forests and fires. Fires always burn up the hill, so your sensor should always be at the top! The gasses you detect will always be moved as part of the heated air. Your communications will never work with mountains in the way, so at the top of hills, again.
Add a lightening detector as most natural fires are lightening caused.
Heated air, I didn't think of that.
Shouldn't the nodes be positioned a few meters above the ground, depending on the foliage, so they can be close to the fire?
I don't plan on having a very sparse network, so mountains are not a problem (I think).
I've never heard of a lightning detector, but I will look into it.
Got any ideas on how I will record data from those gases?
Thank you!
My English may lack proficiency.
How will I isolate CO to sample the gas readings, for training a NN?
Should I sample gas mixtures from a forest fire or spark a fire in a controlled environment?
I think you are right it doesn't have the sensitivity of CO/CO2 sensors in its price range, I wanted to use the BME688 and rely on its gas readings for detecting mixtures that signify potential fire.
Idk.. Should I ditch the BME688 idea?
Tbh I wanted to avoid simplicity, so I considered this my go-to option.
A single sensor to measure CO/CO2 may not be available. A sensor such as the Sensirion SCD30 could be used for CO2. I would investigate what sensor(s) are use in home fire alarms as well.
My impression is that it won't give you the detection of gases common to forest fires that you are looking for. Maybe not futile but certainly suboptimal.
I did a little preliminary work on a much shorter range ‘local’ forest fire detection system, and the thing that stayed with me was the initial detection of ‘smoke’ and other outgassing, long before radiant heat became an issue,
I’d be thinking a low-velocity fan forced chamber to measure the characteristics s of the ambient air - being somewhat down wind of the fire,.. suggests multiple multiple monitoring locations around the fire ground,
The final consideration would be ensuring RF propagation once the fire is near enough. A wireless mesh at lower frequencies may help with that,
Thanks for the response!
Certainly, a chamber would be the perfect solution!
The problem is that I will have to find one first
Nevertheless, your solution is welcome!
I will have to deal with propagation problems later, whether it's 868ΜHz or 433ΜHz band, something will work out (you right that lower frequencies are less prone to obstacles).
If the simple comms you suggested wont work, then you need much more complex comms, mobile phone or Internet over satellite etc. Ground based comms do have significant limitations.
No, I'd say no.
Although it is intended as a ( low cost ) indoor air quality measurement and its measure are calibrated in that direction I think it can be used also in different ways.
If I recall correctly somenone was using this sensor to 'analyze' ( I know the term is not correct ) gasses during home composting.
I think its sensors are not 'narrow band' and are sensible to a wide range of gasses ( and imagine smoke in forest fires should be able to target it ).
Regarding the rf I think lora + mesh, of some sort, could be a good solution ( as already noted )
That might actually be an advantage in this application. I wondered if its apparent intended application might make it susceptible to false triggers from animal poop, decaying organic material, etc.
