Sensor correction and Calibration advice

Question

Dharmesh Joshi 2022-5-16

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1720705-sensor-correction-and-calibration-advice

Hi

I am working on a project that uses using electrochemical N02 sensor in a system.

For your reference, the link to the datasheet is below:

https://www.alphasense.com/wp-content/uploads/2019/09/NO2-B43F.pdf

The sensor outputs two signals, which are the WE and AE.

The sensor is kept at a reference site, so we have reference data to compare the results in regard to calibrating the sensor. This is the primary goal, but I feel the data needs to be filtered and ideally corrected a little.

In my Matlab program, no2() generates a plot with the following

1. Raw Voltge of WE and AE

2. Temperature

3. Humdity.

4. N02 sensor values(full data per minute) and Reference data(hour averaged)

5. N02 sensor values(data average to hour) and Reference data(hour averaged)

From plots 4 and 5, you will see that when we compare our sensor to the reference sensor data, it generally follows the trend, but then there is a sharp drop with the data on the negative y-axis.

The reason for this negative issue is due to oversampling, which can be detected by observing that when WE voltage drops below AE voltage or gets very close to it. This issue occurs when the gas concentration is very low (less than 25 ug, for example. Ideally, the sensor should not be below 0, but this is not compensated for due to limits in the original algorithm or calculation

Before i start pasting any code, can someone advice me on what tools i should be using to resolve this issue?

Can we use the Raw voltage to detect when this issue arises or is likely to arise and perform some type of compensation so that output values are close to the reference data? This would be only when the concentration is getting low.

An additional issue which is shown in plot 4 is that there might need some filtering as some samples are peaking and then going back to the normal trend. These peaks will affect the average value.

Can it be possible to perform some type of filtering?

In general, instant peaks are not expected as gas concentration increase and drop gradually.