Damage Detection of Structures Using Reference-Based Stochastic Subspace Method

Damage to the structure during operation has always been possible, and therefore the issue of monitoring the health of important structures in order to control and manage safe operation has received attention in recent years. The process of extracting the parameters involved in identification the inherent characteristics of the target structural systems in order to monitor and detect damages is possible with different methods which have been introduced and investigated under the title of environmental modal analysis, which have been developed both in the time domain and in the frequency domain. Among the time domain methods, we can mention the stochastic subspace identification (SSI) method. The stochastic subspace identification method is one of the Output Only Modal Analysis (OOMA) methods that has been taken into consideration assuming the existence of uncertainty in modeling as well as noise in data observation and measurement, and system parameters are identification by applying statistical relationships to the output data. Due to the high accuracy of the covariance-drived stochastic subspace (SSI-cov) method which performs data monitoring by constructing the covariance function related to the recorded output data, therefore this sub-method has been used. Due to the limitation in the number of sensors that can be installed in real structures, the use of the Reference-based covariance-drived stochastic subspace identification (SSI-cov-ref) method will make it possible to identification the structure under investigation with a limited number of sensors and if there is damage, Its location can be recognized. The efficiency of the reference-based covariance-drived stochastic subspace identification method with a more limited number of sensors can also be presented. In order to damage detecting in the structure, the method of measuring changes in the modal strain energy of the members in healthy and damaged states has been used, and an index called modal strain energy has been defined and presented But in the case of damage in the structure, the undamaged members will also have strain energy due to the strain effect of the damaged members. Therefore, by using the probabilities approach, it will be possible to provide an index of the probability of structural member damage under a specific failure scenario using different information sources based on the mode shapes. In the following, using the genetic algorithm in the form of an optimization problem, the installation location for the available sensors for the target structure is optimized and presented. During the current research, after finite element modeling in MATLAB software and dynamic analysis of several samples of structure with the assumption of installing a limited number of sensors ,and the results obtained from the output of the reference-based covariance-drived stochastic subspace identification method using the changes in the modal strain energy of the members And the Bayesian strategy has been used in the damage detection in the affected members, and the efficiency of the proposed method has been discussed in the framework of the mentioned process. Based on the results obtained from the output of the proposed process, the affected members will be identified and distinguishable.
.