Predicting the effects of tool geometries on friction stirred aluminium welds using artificial neural networks and fuzzy logic techniques

摘要

Effect of friction stir welding (FSW) tool geometries on aluminium welds were investigated using different tool shoulder and pin probe geometry profiles. A combination of 27 tool shoulder and pin profile geometries were used for the experimental purpose using a design matrix. The effect of these tool geometries on the friction stir welds like the weld strength, weld cross-section area and grain sizes were investigated. The effects of the tool geometries were predicted using artificial intelligence techniques such as artificial neural networks (ANN) and fuzzy logic modelling. It was observed that, for a combination of FSW tool geometries, the ANN model was not so effective in predicting the FSW weldment characteristics, while the fuzzy logic model was able to predict the same with much lower percentage of error for the test cases.