Abstract:
In the past two decades, sensor technology has achieved the manufacturing of low-cost and portable sensors that can be used for different environmental applications such as water quality monitoring, air quality monitoring, and soil quality monitoring. The sensors used for environmental monitoring face the problem of drift sooner or later after installation. The drift may occur due to sensor aging, temperature and humidity variation, poisoning among the sensor array, or due to a combination of all. This analysis will lead us to a different track. This sensor drift will demolish the calibration model of any instrument. This issue can be solved by the calibration of the sensors, which is also a challenge for field-deployable instruments. In this chapter, an alternate solution is provided for the drift compensation based on artificial neural network (ANN). A low-cost pH sensor is used for the research work and explanation as well. The pH sensor readings were observed 66 times during the measurement session in the reference solution. The drift was observed in the pH sensor readings and compensated using a feed-forward neural network. The simulation was performed on the Python platform. The drift compensation was successfully achieved using the ANN model as the RMSE was reduced to as minimum of 0.0001%.