Assessment of Artificial Neural Networks ability in Winching Time Study of Timber Jack 450C

Estimating of forest equipment productivity is an important aspect of managing cost in forestry, which leads to reduction of operations expenses. In other words, high capital cost in forest harvesting, is a good reason to argue forest engineering research and time modeling. This paper applied one of the Artificial intelligence subsets, which are called Artificial Neural Networks (ANNs), to predict winching time of wheeled skidder Timber jack 450C in Neka Choob forests. Continuous time study was performed to collect winching data during skidding operations. Winching time and related effective factors including slope, distance, number and volume at every winching cycle was measured simultaneously. Two neural networks type- Radial Basis Function and Multi Layer Perceptron- were used to develop winching time model. Moreover, in order to compare accuracy of ANNs and mathematical model, the regression analysis method was developed. Results showed that RBF network provided more accurate results in winching net time estimation compare to MLP neural network. The most effective variable in both networks was determined distance to the center of the skid trail. The results showed that the model developed by neural networks has more precision than the linear regression method.