首页|A novel fuzzy inspired machine learning framework for relative humidity estimation using time-of-flight of ultrasonic sensor
A novel fuzzy inspired machine learning framework for relative humidity estimation using time-of-flight of ultrasonic sensor
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NSTL
Elsevier
This paper proposes an ultrasonic time-of-flight (ToF) based technique to estimate the relative humidity of the environment accurately. The framework of the proposed model is to let a fuzzy controller classify the input data to different segments based on the fuzzy output, and for each segment of data fed into a specific pre-trained neural network to predict the relative humidity. Neural networks are first trained, and the trained networks are used to estimate the relative humidity. The result shows that the proposed fuzzy-artificial neural network model gives better performance with an average root means square error (RMSE) of 1.269, mean absolute error (MAE) of 1.0415, mean absolute percentage error (MAPE) 1.962 and the coefficient of correlation (R-values) of 0.9468 compared to other methods. Experimental results indicate that the variation in relative humidity estimation is bounded by +/- 3% which is as good as commercially available off-the-shelf relative humidity sensors.
Ultrasonic sensorTime of flightRelative humidityTemperatureFuzzy logicArtificial neural networkMEASUREMENT SYSTEMNEURAL-NETWORKSTEMPERATUREAIRSPEEDLOGICSOUNDPREDICTION
NSTL
Elsevier
