Measurement Of The Moisture Content In Biomass Based On The Capacitive Sensor

Woodchips are widely applied in industrial processes and daily lives.The processing and combustion efficiency of the woodchips is significantly affected by the moisture content(MC)and an appropriate method is needed for measuring the MC in woodchips.The capacitive method is widely used in various industrial processes because of its simple structure and low cost.However,the measurement accuracy is easily affected by the factors such as the types of woodchips,temperature and the volume density.A helical capacitive sensor is developed in this paper to measure the MC in woodchips.Compared with the parallel capacitive sensor,it has more uniform sensitivity distribution,which can improve the measurement accuracy of the MC in woodchips.Firstly,based on the finite element model,the structure of the capacitive sensor is optimized.Taking the sensitivity uniformity as the optimization criteria,the influence of four structural parameters on the sensitivity uniformity is analyzed by the orthogonal test method and the final structure of the sensor is determined.Then the measuring system is built,which is mainly composed of the precision LCR meter,the helical capacitive sensor and the computer.The LCR meter sweeps the frequency range from 20Hz to 1MHz and the influence of the type,the packing density of woodchips and moisture distribution on the capacitive MC measurement is investigated.The results show that:(1)With the increase of the frequency,the capacitance value decreases monotonically under lower MC.In addition,a crest appears when the MC increases to about 20%.The results show an excellent correlation between the frequency at the peak and the MC.(2)The particle size of woodchips and the void age of packing will affect the packing density.The larger the packing density,the greater the capacitance value,so the mass of woodchips is introduced to represent the packing density.(3)Different kinds of woodchips have different dielectric properties due to their different densities.The larger the density,the stronger the polarization effect and the larger the capacitance value.Therefore,it is necessary to overcome the influence of the kind of woodchips on the MC measurement.(4)When the measurement system is used to measure the samples with uneven moisture distribution,the prediction accuracy is greatly affected by the water distribution.Finally,the prediction models based on the random forest(RF),deep neural network(DNN)and support vector machine(SVM)algorithm are established.The mass of the woodchips,the frequency at the peak of the capacitance spectrum,and the capacitance values at 300kHz and 1MHz are selected as the input parameters.The models are trained with the data from applewood,and the prediction accuracy of the models are verified with the data from other two biomass of cedarwood and oakwood,DNN and SVM models show good generalization performance.In addition,the measurement system can predict the MC of other woodchips without calibration and the mean value of DNN and SVM model is used as the final moisture prediction value with R~2>0.950,RMSE≤2.362%,MAE≤1.978%.The repeatability of the measurement results is affected by the particle size of the sample and the absolute error is within 1.58%.