Measurement of the moisture content in woodchips through capacitive sensing and data driven modelling

In this paper, a helical capacitive sensor is developed to measure the moisture content (MC) in woodchips. Firstly, based on the orthogonal test method, the structure of the capacitive sensor is optimized to obtain the best possible uniform sensitivity. Then, the effect of the type and random distribution of woodchips on the capacitive MC measurement is investigated. Finally, three different algorithms, including support vector machine, random forest and deep neural network, are employed to establish the data driven models. Experimental results demonstrate that the proposed system is capable of measuring the MC in woodchips with absolute error within ±5%. The generalization capability is verified using the cedarwood with three size ranges, with R2, RMSE and MAE of 0.95, 1.69% and 1.28%, respectively. The absolute error of the predicted MC in cedarwood over the range 24.3% and 25.2% is found to be within ±2% for a range of packing densities.