An Application of Neurogenetic Algorithm System to Individual Tree Growth Model

Abstract

A neural network system with genetic algorithms (Neurogenetic Algorithm System, or NGAS) was employed to develop individual coniferous tree growth models. A multivariable regression model was applied to compare the performance of NGAS. An IBM personal computer with the BioComp System's software program of NGO was used to execute this comparison. The results indicate that NGAS is more accurate and effective than the conventional regression method in modeling individual tree growth based on the criteria of Sum of Squared Error (SSE), Average Absolute Error (AAE), Mean Squared Error (MSE) and Final Predicted Error (FPE). This study also suggests that individual tree growth may indeed be a non-linear process. Using this flexible neural network system to model individual tree growth can yield satisfactory prediction results.

Key words:

• Artificial Neural Network (ANN)
• Genetic Algorithms (GAs)
• individual tree growth
• Neurogenetic Algorithm System (NGAS)