Artificial Neural Network For Automated Gas Sensor Calibration
One of the major problem in the calibration of solid state gas sensor is the accuracy of the calculated
concentration values from the voltages measured. The customary method employed till date is through the regression
analysis of the dataset. The best fit equation obtained after plotting the voltage vs concentration values can be programmed
in a microcontroller to attain the concentration values. These results cannot be relied upon as they were less accurate.
Instead, an Artificial Neural Network is created, which will learn the dataset and provide accurate results for any unknown
input fed to it within the range of the dataset. The results that were obtained show that artificial neural network provides
maximum accuracy in determining the analyte concentrations. Hence, this method ensures lower calibration cost and very
Keywords— Artificial Neural Network, Backward Propagation, Forward Propagation, Gas Calibration, Gas Sensors,
Gradient Descent, Layers, Weights.