A data-driven force-thermal coupling load identification method considering multi-source uncertainties of structural characteristics and measuring noises

Abstract

The problem of load identification denotes identifying loads based on the measurement of structural responses, which is the inverse problem in structural dynamics. With advancements in aeronautical technology, the working environment of aircraft becomes even more complex. To accurately monitor and forecast the load environment where the structure works can provide guidance for the structural design of aircraft and predict potential structural damage. However, with the development of hypersonic aircraft, the influence of thermal field also cannot be ignored besides external forces that are applied to the structure, thus, there is an urgent need for effective methods to identify both force and thermal loads. In this paper, firstly, A data-driven load identification method and various load identification strategies for the identification of force-thermal load based on Artificial Neural Networks are proposed, and in different loading cases the relative error of identification is less than 5 %, with the training process converging efficiently within a short time; additionally, multi-source uncertainties, including Gaussian white noises and structural uncertainties, are simulated, and their influence on load identification is also evaluated; moreover, sensor placement optimization based on particle swarm optimization is carried out to enhance accuracy of load identification, and the number of sensors and the corresponding optimal placement of sensors are determined, it can be proved via numerical examples that the optimized sensor placement can reduce the error of load identification by more than 90 % of that of a random sensor placement.