Drift compensation of commercial water quality sensors using machine learning to extend the calibration lifetime

There are specific issues in the multi-sensor systems used for water quality monitoring, which prevents these systems for routine measurement of water samples. An important issue is drift; related to sensor readings, which may refute the calibration of sensors leads to the necessity of frequent recalibration of the sensors that required effort as well as shut down the system. An alternative approach for drift correction is based on the mathematical correction method. The paper proposed a regression calibration method and implemented by the machine learning approach. In this paper, we have used a feed-forward artificial neural network based regression model to extend the calibration lifetime of sensors. The evaluation of the model was performed based on the root mean square error and the root mean square error for cross-validation. The proposed model is also compared with the traditional statistical method and proved to be superior to the traditional one. The experimental results demonstrate the best performance with a negligible error rate. Based on the results of the current study, ANN appears to be more adaptive for data analysis in environmental monitoring applications.