Neural modelling of cavitation erosion process of 34CrNiMo6 steel

Abstract

Artificial neural networks (ANN) are commonly used to solve many industrial problems. However, their application for cavitation erosion evaluation is a quite new attempt. Thus, the aim of this work was to elaborate the neural model of the cavitation erosion process of 34CrNiMo6 steel. Cavitation erosion tests were conducted with a usage of the ultrasonic vibratory method with stationary specimen that relies on the ASMT G32 standard. The proceeding damage of marked steel surface area was observed by means of a scanning electron microscope. Wear was evaluated with profiler measurements, image analysis of cavitation worn surface areas and weighing done in stated time intervals. The cavitation erosion results were analysed with Matlab software by Neural Network Toolbox. The developed neural model of cavitation erosion process that combines exposure time, roughness, area fraction of worn surfaces, and mass loss gives promising results.