Bi-directional prediction of wood fiber production using the combination of improved particle swarm optimization and support vector machine

Abstract

In order to investigate the relationship between production parameters and evaluation indexes for wood fiber production, a bi-directional prediction model was established to predict the fiber quality, energy consumption, and production parameters based on the improved particle swarm optimization and support vector machine (IPSO-SVM). SVM was applied to build the bi-directional prediction model, and IPSO was used to optimize the SVM parameters that affect the performance of prediction greatly. In the case of the forward prediction, the model can predict the fiber quality and energy consumption using the production parameters as input information; in the case of the backward prediction, the model can estimate production parameters using the fiber quality and energy consumption as evaluation indexes for input information. The results showed that the IPSO-SVM model had high accuracy and good generalization ability in the prediction for the wood fiber production. Additionally, based on the effectiveness of the proposed model and preset evaluation indexes, the corresponding production parameters were determined, which was able to save the wooden resources as well as reduce energy consumption in the fiberboard production.