Optimising the steering clutch of a walking tractor using soft computing approaches

Abstract

ABSTRACT Walking tractors are often used for weeding gardens, small fields, vineyards, and greenhouses. The clutch fork arm (CFA), which is used on the rotary axle, is subject to high loads. Therefore, an optimal design must be carried out under all load conditions while minimising weight. The present study aims to reduce the mass and load acting on the CFA and increase its safety factor. To achieve an optimal design, this study used the dimensions of the CFA as independent variables and the mass and maximum load exerted on the CFA as dependent variables. Multi-linear regression (MLR) and Artificial Neural Network (ANN) models were then developed. MLR and ANN models were proposed to predict the mass and load acting on the CFA. The proposed models were trained and validated using a dataset containing dimensions, mass parameters and maximum stresses for eighty-one CFA specimens based on the results of the finite element method (FEM). Based on the optimised structural properties, a FEM model of the CFA was then developed in Ansys R19.2. The results showed that the maximum absolute difference in stresses from the FEM and ANN models was less than 3.7%, while for mass it was less than 2.2%. Furthermore, the maximum reduction in mass and increase in safety factor were achieved by 52.0% (from 1.3 kg to 0.62 kg) and 36.0% (from 1.28 to 1.64), respectively. Therefore, these results were a confirmation of the ANN model to use as an alternative to the FEM method.