ElishaKrauss (Talk | contribs) |
ElishaKrauss (Talk | contribs) |
||
Line 19: | Line 19: | ||
<div style="width:70%;margin-left:15%"> | <div style="width:70%;margin-left:15%"> | ||
− | <h2>Thermistor</h2> | + | <h2>Thermistor - Breathing Rate</h2> |
− | + | <p style="width:1200px;margin-left:15%">To measure breathing rate, multiple readings are taken of the NTC thermistor’s resistivity over a set period of | |
− | + | ||
− | + | ||
− | <p style="width:1200px;margin-left: | + | |
time. As the child breathes out, warmer air comes in contact with the thermistor, which is then followed by colder | time. As the child breathes out, warmer air comes in contact with the thermistor, which is then followed by colder | ||
air as the child breathes in. These changes in temperature are reflected in the changes in resistivity of the | air as the child breathes in. These changes in temperature are reflected in the changes in resistivity of the |
Revision as of 17:08, 14 October 2018
Thermistor - Breathing Rate
To measure breathing rate, multiple readings are taken of the NTC thermistor’s resistivity over a set period of time. As the child breathes out, warmer air comes in contact with the thermistor, which is then followed by colder air as the child breathes in. These changes in temperature are reflected in the changes in resistivity of the thermistor, the maximums and minimums of these changes over time can be counted to calculate a rate.
The code written to calculate this rate first takes a reading of the thermistor’s resistivity once every second for 100 seconds and stores the values in an array. This data can be plotted against time to create a sinusoidal wave as shown below. However, to count the peaks corresponding to each individual breath, the data must be preprocessed to remove outlying maximums and minimums.
This is done using an exponential filter, which is a type of low pass filter used to remove high frequency noise. It is also an autoregression filter, which means that it relies on past output and input, so that over time it relies more heavily on past values and is able to detect patterns such as a steady breathing rate. The filtered data can be plotted to produce the sinusoidal wave shown below.
The cleaned data can then be analyzed to detect peaks. This is done by comparing each value to the previous value and the value following it. If the current value is larger than both, it must be a maximum and the number of peaks is increased. The total number of peaks, corresponding to the total number of breathes, is then divided by 100 seconds to determine a rate of breathes per second.