Detecting and identifying cracks in low-speed slender rotating beams through modal parameters and artificial neural networks

Abstract

Rotating beams represent simplified forms of more complex mechanical elements, often found in strategic domains like wind turbine or helicopter blades. The presence of a crack within the beam alters its dynamic behaviour, leading to modifications in modal parameters, such as natural frequencies. This paper presents the theoretical study of an Artificial Neural Network (ANN), using MATLAB commercial software, for the detection and identification of cracks in rotating Euler-Bernoulli type beams at low speeds, typical of wind turbine blades, based on the changes in the natural frequencies of the component. The input data to the Artificial Neural Network have been obtained from an analytical model of the dynamic behaviour of the cracked rotating beam with which the values of the natural frequencies of vibration in a plane perpendicular to the plane of rotation (flapwise) can be calculated as a function of the rotational speed, the slenderness of the beam, and the radius of the hub that separates the end of the beam from the axis of rotation.