Estimating the mechanical properties of Heat-Treated woods using Optimization Algorithms-Based ANN

The study investigates the Modulus of Rupture (MOR) and Modulus of Elasticity (MOE) of the heat-treated wood by implementing Artificial Neural Networks (ANNs) on a dataset of 104 Spruce and Larix wood species obtained from literatures. The Feed Forward (FF) networks with various topologies were employed considering heat treatment at varying temperatures, times, and relative humidity (RH) as input parameters. The Grey Wolf Optimizer (GWO) was used to optimize the weight of the networks by reducing the error. The accuracy of the proposed ANN model (GWO-ANN) was obtained by comparing the results of performance indicators obtained by Particle Swarm Optimization (PSO), Multiple Linear Regression (MLR) and Nonlinear Regression (NLR) models. The results concluded higher accuracy of GWO-ANN model with a coefficient of correlation, R2, equals 0.975 and 0.960 and the Average Absolute Error (AAE), equals 0.01 and 0.01 for the MOR and MOE, respectively.

Automated summary

This study used Artificial Neural Networks (ANNs) to investigate the Modulus of Rupture (MOR) and Modulus of Elasticity (MOE) of heat-treated wood. The ANN model (GWO-ANN) with different topologies and parameters was optimized using the Grey Wolf Optimizer (GWO). The accuracy of the GWO-ANN model was compared with other models, and it showed a higher accuracy with a coefficient of correlation (R2) of 0.975 and 0.960 for MOR and MOE respectively, and an Average Absolute Error (AAE) of 0.01 for both.