A Temperature and Humidity Compensation Method for On-board NOx Sensors with LSTM Network

NOx is a pollutant component of vehicle exhaust. On-board NOx sensors usually operate under electro-chemical reactions. Temperature and humidity can influence these reactions, resulting in the inaccuracy of measurements. Thus, a method based on long short-term memory (LSTM) network for temperature and humidity compensation of on-board NOx sensors is proposed in this paper. The LSTM network is trained with the sensor-measured and the true NOx concentrations alongside with the temperature and relative humidity of tested gas. The testing results of the LSTM network shows that this method has good performance for the temperature and humidity compensation, and can efficiently improve the accuracy of on-board NOx sensors.