Torsional vibrations of pre-twisted blades using artificial neural networktechnology

The free torsional vibrations of a linearly tapered, twisted flexible blade , rotationally constrained at an arbitrary position along the length of bla de, have been investigated using neural networks. The blade has a rectangul ar cross-section with equal taper in th horizontal and vertical planes, in addition to the flexibility at the root portion. The constraint is a rotati onal spring. The constraint on the blade at an optimum location is designed so as to increase the lowest natural frequency of the blades with consider able root flexibility. The optimum location is determined as the position o f the node in the second mode shape of the unconstrained tapered blade with flexible roots. A trained neural network is used to identify the location of the nodal or optimum point for a given blade-taper ratio and root flexib ility parameter. The minimum stiffness of the constraint at an optimum posi tion for a maximum raise in the first eigenfrequency is evaluated. Results are presented in tabular and graphical form.