Machine Learning-Based Calibration of Low-Cost Air Temperature Sensors Using Environmental Data

Abstract

The measurement of air temperature is strongly influenced by environmental factors such as solar radiation, humidity, wind speed and rainfall. This is problematic in low-cost air temperature sensors, which lack a radiation shield or a forced aspiration system, exposing them to direct sunlight and condensation. In this study, we developed a machine learning-based calibration method for air temperature measurement by a low-cost sensor. An artificial neural network (ANN) was used to balance the effect of multiple environmental factors on the measurements. Data were collected over 305 days, at three different locations in Japan, and used to evaluate the performance of the approach. Data collected at the same location and at different locations were used for training and testing, and the former was also used for k-fold cross-validation, demonstrating an average improvement in mean absolute error (MAE) from 1.62 to 0.67 by applying our method. Some calibration failures were noted, due to abrupt changes in environmental conditions such as solar radiation or rainfall. The MAE was shown to decrease even when the data collected in different nearby locations were used for training and testing. However, the results also showed that negative effects arose when data obtained from widely-separated locations were used, because of the significant environmental differences between them.

Prediction of Climatic Parameters from Physicochemical Parameters using Artificial Neural Networks

“…In this study, two statistical approaches (MLR and MLP-ANN) were developed based on twelve independent vari-ables (year, month, water temperature, oxygen saturation, chemical oxygen demand, pH, electrical conductivity, ammonium ion, nitrate ion, turbidity, organic matter, and dry residue) to predict successively the maximum and minimum air temperature, the relative humidity, and sunshine duration in the area of Ain Defla of Algeria. The models were trained, and tested using a sample of 156 data of climatic and physicochemical parameters of Ghrib dam water (Ain Defla), measured monthly over a period of 13 were used to test the predictive power of the MLP-ANN model. The variation inflation factor (VIF) and correlation analysis showed the rightness of the choice of the twelve variables.…”

Self-calibration methods for uncontrolled environments in sensor networks: A reference survey

“…Conducting a multiple linear regression with respect to ground-truth O 3 concentrations in a place with different environmental conditions than those of the place in which the sensor node will be deployed can produce large errors in the predicted O 3 concentrations [16]. Yamamoto et al [23] have recently observed a similar behavior in temperature sensors. Temperature sensors calibrated in a place behave differently when placed in another location because of the difference in environmental conditions (such as solar radiation, humidity, wind speed, rainfall, and azimuth) between the two locations.…”

“…Learning which sensor technologies are more stable, what differences exist between sensors from different manufacturers for the same application, and under what conditions the sensors have to be calibrated will trigger the deployment of real applications. Studying temperature and relative humidity sensors, Yamamoto et al [23] reported a mismatch of the calibration error produced in a calibration place and the error obtained in the prediction in another place with different environmental conditions. The same problem has been stated in air pollution low-cost sensors.…”

A Comparative Study of Calibration Methods for Low-Cost Ozone Sensors in IoT Platforms

“…In recent years, there has been greater interest in learning how low-cost sensors behave in terms of quality of information (QoI) metrics such as the root mean square error (RMSE), mean bias error (MBE), or the short-term or large-term capacity prediction of the sensors. Many of the low-cost sensors in IoT platforms are not calibrated by the manufacturers or if they are calibrated by them, the calibration has been done in laboratory chambers and not in the environmental conditions of the place where the nodes are deployed [3], [4]. In this case, the sensors of the IoT platform is calibrated during network deployment in an uncontrolled environment without laboratory instruments [5], [6].…”

“…For this reason, much research has focused on the interaction of environmental conditions such as temperature and relative humidity [3], [4], [7], [8], [9] or on the interactions of other pollutants [10], [11] with respect to one pollutant sensor. In addition, there is recently a greater interest in comparing and studying [11], [12], [13], [14], [15] how signal processing techniques behave for calibrating different air pollution lowcost sensors in IoT platforms.…”