Axial force investigation in truss members using modal test and Neural Network

The aim of this research was to have an estimation of the axial loads in the truss member by using modal test. This could be helpful to evaluate the structural member in existing structures.

By using a modal test, the response of the truss being recorded using accelerometers. The recorded responses then being transformed from time-domain to the frequency-domain and considered as input data for the evaluation procedure. As next steps, the main features from the data being selected and were utilized to feed a trained Deep Neural Network. The output of this NN is an estimation of the axial force in the truss element.

For this investigations, finite element models being developed in ABAQUS using python codes and the Neural Network and mathematical methods being done in MATLAB.