APPLICATION OF ARTIFICIAL NEURAL NETWORKS AND PARTICLE SWARM OPTIMIZATION FOR TIMBER EXTRACTION WITH CABLE CRANE

The fact that forest areas in Turkey and the world are located at high and steep mountainous areas makes it more difficult to extraction the wood raw material. Therefore, the retreat of forest areas to mountainous areas has brought cable crane to the forefront. Thus, it has become possible to solve complex problems by way of artificial intelligence techniques. The purpose of this study was to determine the impact of factors related with timber extraction via URUSMIII cable crane on total time via Artificial Neural Network (ANN), Particle Swarm Optimization (PSO) and Multiple Regression Analysis (MRA). The data were obtained from oriental spruce timbers which were acquired from spruce stands in the Artvin Forest Directorate, located at NE Turkey. The factors with impact on total time (ground slope, line slope, lateral pull, number of logs, diameter of logs, length of logs, log volume, yarding distance) were measured along with the total time. Determination coefficient (R) and the expressions that indicate error variance (MSE, RMSE and MAE) were taken into consideration for determining the model with the best results. PSO model was determined as the best structure (R = 0.85 MSE = 0.0143, RMSE = 0.1194, MAE = 0.0839): in this study according to the obtained results. The results indicate that PSO had the best performance in the study followed by ANN and finally MRA with the lowest performance. The PSO model can be used for similar conditions on the planning of forest operation, the control of applications and the determination of unit of price for forest workers.