A Modified Gaussian Process on Temperature Compensation of Zirconia Sensor

Aviation is a critical area of development, making the issue of oxygen concentration at high altitudes particularly important. Currently, the most common and widely used sensor for measuring oxygen concentration in the air is the zirconia sensor, which must be calibrated before use. Currently, the calibration of zirconia sensors is limited to a working temperature of 700°C. However, wear and aging during use can cause the actual working temperature to deviate from 700°C, leading to inaccurate calibration results. To address this issue, we calibrated the zirconia sensor at intervals of 50°C between 600°C and 900°C, calculating the final results using different methods. In this paper, we propose an improved gradient descent method that is more suitable for detecting oxygen in zirconia sensors across different temperatures, significantly improving the accuracy of the results through the use of Gaussian Process regression. Since the Gaussian Process is used to predict oxygen concentration and yields the most accurate results, it was adopted and further improved to reduce both the mean squared error (MSE) of the output results and the computational time required for Gaussian Process regression.