New method for calibrating optical dissolved oxygen sensors in seawater based on an intelligent learning algorithm

Oxygen sensors based on luminescence quenching are the most commonly used instruments for in situ measurement in seawater due to their accuracy and long-term stability. The calibration method of the sensor is crucial for their accuracy. Conventional methods exhibit some defects, such as strict control of calibration conditions and cumbersome and time-consuming operation. To improve calibration operation and obtain good calibration results, a new calibration method was proposed for the optical dissolved oxygen sensor in seawater based on an intelligent learning algorithm. The sensor to be calibrated and the reference sensor were deployed in the water for synchronous measurements. The calibration system consisted of a temperature-regulated device and a sampling method to improve calibration operation. An intelligent learning algorithm was used to train the calibration data and model the oxygen response of the sensor. Calibration and test results in both laboratory and field showed that the new calibration method is feasible and efficient. It is highly significant for sensor development and in situ measurement in seawater.