ANN-based data fusion for lumber moisture content sensors

Abstract

Lumber moisture content (MC) is one of the main feedback quantities of the wood drying process. Its precision directly affects the drying quality, the cost and the drying time of the wood. The measurement precision of the MC sensor is mainly affected by ambient temperature. In this paper, an artificial neural network (ANN) data fusion method is put forward to eliminate the temperature disturbance. The output of the lumber MC sensor and environmental temperature are considered inputs of the ANN. In the data fusion model, the non-linear response characteristics between the lumber MC and ambient temperature are modelled by an ANN with a back-propagation algorithm. The experimental data show that the data fusion method based on an ANN converges quickly and effectively eliminates measurement errors. The ANN automatically compensates for the changes in ambient temperature based on the network information stored in its weights. The lumber MC online measuring system realizes steady, real-time, high-accuracy measurement. The output stability was increased by 75 times compared with traditional methods, and the measuring error is within 0.8% (full scale) over a range of temperature variations from 308 to 808C.