No abstract

A method of prioritizing sensor locations on a¯exible structure for the purpose of determining damaged structural elements from measured modal data is presented. This method is useful in applications where practicality dictates only a small subset of the total structural degrees of freedom can be instrumented. In such cases, it is desirable to place sensors in locations yielding the most information about the damaged structure. No a priori knowledge of the damage location is assumed. The prioritization is based on an eigenvector sensitivity analysis of a ® nite element model of the structure. In addition, the dual problem is presented and solved, which determines the observability of change in the measured eigenstructure from the instrumented degrees of freedom. This analysis is used to determine the extent to which damage can be localized. An analytical example is presented that illustrates the relationship between the number of measured modes, the number of instrumented degrees of freedom, and the extent to which damage can be localized. Additionally, an analysis of an experimental cantilevered eight-bay truss assembly consisting of 104 elements instrumented with eight single-axis accelerometers is presented. The extent to which structural damage can be localized from the measurement data is limited by the number of measured modes.

A method of identifying damaged structural elements from measured modal data using an incomplete measurement set is presented. The method uses a mathematical optimization strategy to minimize deviations between measured and analytical modal frequencies and partial mode shapes. Damage is identi® ed by determining the stiffness change to a ® nite element model required to match the measured data of the damaged structure. Damaged elements are obtained directly from the results of the cost-function minimization because the allowed structural changes are consistent with the original ® nite element formulation. The cost-function minimization is based on an assigned partial eigenstructure algorithm in which the physical properties of the structural elements are treated as control variables, which are chosen to achieve the measured partial eigenstructure. An iterative solution is used to solve the nonlinear optimization problem. Experimental results are reported for a cantilevered eight-bay truss assembly consisting of 104 elements.