Sensor Calibration System for Enhancement of Quality of Information

Abstract

Calibration of different types of industrial instruments is the prerequisite of any of the industrial installations before coming under operation. Modern calibration can be categorized based on the physical quantities they are designed to measure. The categorization may internationally vary but the basic notion of calibration is clear and straight forward whereas the associated standards of calibration vary from parameter to parameter under different physical environment. Here a data acquisition system has been developed using Arduino Uno R3 and infrared emitter- detector pair. An experimental set up has been build which can collect data from the sensor under calibration, namely ultrasonic sensor HC-SR04. The data acquisition system includes the set up called, the ground truth apparatus which specifically has been designed to carry out repeated experiments to collect data. The process of collection of data involves the infrared based obstacle detection technique. A number of errors compensation technique are being applied. These methods of error compensation technique involves feed-forward neural network with backpropagation, conjugate gradient and regularization training algorithm. Finally a parameter based analysis of performance for different error compensation techniques is done. Based on the performance analysis some novel approaches for calibrating the information generated by the sensor has been prescribed. A clear distinction in approach between data calibration and instrument calibration is established.