Inference of the Tensile Force in Cables with Insulators Using an Artificial Neural Network

Abstract

The stability of guyed structures is highly dependent on the cables tensile force. Therefore, a versatile and reliable method for the verification of such force starting from the construction and during its service lifespan, is desirable. In the present work, an inverse method for the determination of the tensile force in cables with insulators is proposed, through the implementation of an Artificial Neural Network (ANN). Firstly, a Finite Element model is used to obtain the first natural frequency of different cable configurations. In this way it is possible to vary the tensile force, length, tilt angle, and number of insulators. The data resulting from the computational simulations is used to train the ANN. During this stage, corresponding input and output samples are introduced to the network. Once the training is completed, the ANN is capable of representing the relation between the input parameters (length, tilt angle, number of insulators and first natural frequency) and the cable tensile force. Additionally, a physical model of the cable is developed in the laboratory. A dynamic study of several configurations is performed in order to obtain the corresponding experimental natural frequencies. Finally, in order to validate the method, the parameters of the physical model configurations are introduced as the inputs of the ANN and the tensile force values are inferred. The results are compared to the actual force on the laboratory model. The resulting error is acceptable in all the cases.