Humidity Sensor Accuracy Improvement Based on Two Nested Kalman Filters for Commercial Cultivation of Tropical Orchids

Polymer dielectric-based humidity sensors used in the orchid greenhouse monitoring system usually have a problem concerning the accuracy when used continuously for some time. It is because those sensors are exposed to high humid conditions regularly. In a sense, data read from the humidity sensor is noisier than those from other sensors deployed in the greenhouse. Therefore, this paper proposes a simple data-driven technique based on two nested Kalman filters for sensor accuracy improvement. It aims to minimize the difference between humidity values read from a humidity sensor and those from the more-accurate sensor. The humidity values are estimated by a Kalman filter, of which its prediction is made based on another different Kalman filter. The inner Kalman filter delivers such the prediction by fusing information obtained from surrounded sensors. Experimental results show that this technique can improve measurement accuracy by 32.02%. This paper also discusses the possibility of applying the proposed scheme in the case that the sensor fails to operate normally, in which the Kalman gain will be adjusted so that the Kalman filter relies more on the prediction.