Blind modal identification for truss structures based on reduced measurements

The out-put only modal identification of aerospace structural systems has attracted much attention during recent years, as input excitations can be hardly measurable for in-flight environments. This paper presents a developed blind modal identification method considering incomplete measurements. Firstly, the sparse time-frequency features of modal responses are represented as the independent sources hidden in those of the system responses which have been mapped into a time-frequency blind source separation framework through Short-time Fourier transform; then these representations can be effectively extracted by the Independent-Component-Analysis algorithm. Hence the time-domain modal responses are recovered such that the modal parameters are readily obtained. Furthermore, a new genetic algorithm called NSGA-II is introduced and an optimized sensor placement technique is established, which promises to reduce sensor-installation cost in large-scale structures. The simulation results of a fivestorey truss Finite Element Model show that the proposed optimization method is efficient for finding better sensor layouts by selecting solutions from the pareto front. The laboratory experiments of the real five-storey truss also demonstrate the availability of the method by collecting dynamic responses from optimized accelerometers and conducting blind modal identification on the complex space truss structure.