A supervised neural computing representation predicts obstacle induced force incitement in off-road vehicles

Abasalt Afranjeh, Aref Mardani


In this paper an attempt has been made to evaluate and predict obstacle induced force incitement in off-road vehicles affected by inflation pressure, wheel load, obstacle type and height, soil texture, tire type, slippage and velocity using artificial neural network (ANN) technique trained by back propagation algorithms from readily available data obtained from experiments conducted in soil bin facility and a single wheel-tester. A total of 6912 samples were available for training, validating and testing the neural networks. Evaluating a three layered architecture with a four layered one, the optimal topology to yield better performance on the criteria of lower root mean squared error (MSE), T value and coefficient of determination (R2) was a four-layered one. Then among 2 hidden layers, each of layers was increased from 0 to 40 to determine the best number of neurons by the best performance. It was divulged that 8-12-10-2 provided the best performance of MSE and T were 0.027, 0.977, and R2 with, respectively. 

Full Text:



  • There are currently no refbacks.