Wood pellet bulk density determination by machine vision deep learning technique

Abstract

Bulk density is one of the physical parameters required by ISO 17225–2 to evaluate the quality of wood pellets. A change in this pellet characteristic leads to considerable variations in combustion efficiency. Pellet bulk density calculation is a time-consuming operation which can be carried out since the pellet production phase. Our research aims to develop an alternative method potentially applicable also on-board heating systems. This work has the task of testing and verifying the efficiency of a system that uses a deep neural network, to determine the pellet bulk density. Our implemented system detects, segments, and determines the volume of wood pellets in a bunch. This problem is not trivial, due to the irregular lighting conditions that affect the quality of the images and the overlapping of the wood pellets. However, the differences between estimated and measured bulk density appear to be non-negligible but this approach provides promising results, especially because it is one of the first approaches in the energy sector.