Prediction of tractor power take-off performance using artificial neural network

Prediction of tractor PTO performance can lead to simulation and optimization of tractor performance, allowing optimum setting of different parameters as well as enhancing decision-making of manufacturer in the design of new tractor. Twenty different parameters were selected as input for PTO performance prediction. The data used as input to train the network were collected from 141 tractor test reports tested between 1997–2013 at Central Farm Machinery Training and Testing Institute, Budni. A Back propagation artificial neural network (ANN) was developed using Neural Network Toolbox in Matlab software. A Matrix of 1704^20 and 1704^1 was used as input for PTO prediction in ANN. The optimum structure of neural network was determined by a trial and error method and 30 different structures were tried. For prediction ANN model with 2 hidden layers having 40 and 35 neurons in first and second layer, respectively gave highest performance. Regression coefficient and MSE for this model was 0.996 and 1.080.