Research On Optimal Sensor Placement For Structural Health Monitoring

Structural health monitoring system is now widely used in various industries such asaerospace, civil engineering, offshore platforms and other large and complex structures. Methodof model test analysis based on vibration is more and more useful, but the optimal sensorplacement that includes the number and place of sensors has important influence to the test result,because it is a key technology to collect structure damage information. The main research subjectof this thesis is to select the optimal numbers and locations of the limited sensors so that theresulting measured data are most informative.In order to achieve the object, both genetic algorithms (GA) and simulated annealingalgorithms (SA) are improved; a new method named adaptive simulated annealing geneticalgorithms is proposed. Using strain modal assurance criterion as the optimization principle,taking the model of a composite wing-box as the research object, the number and place ofsensors are optimized in this thesis.The main works are showed as follows:Firstly, this thesis discusses the main principle of model test analysis based on vibration,analyzes the displacement mode and strain mode; the analysis results indicate that strain mode ismore sensitive to the small changes of structure damage information, so strain modal assurancecriterion (SMAC) is proposed.Secondly, both GA and SA are introduced in detail, in order to overcome the disadvantagesof two algorithms, a new method named adaptive simulated annealing genetic algorithms(ASGA) is proposed. A dual-structure coding genetic algorithms is developed to solve theconstraint problem that the number of sensor doesn’t change; and improved SA is used as anoperator of GA. Therefore, adaptive crossover and mutation probabilities are adopted to preventpremature convergence. The principles and procedures of algorithms are described in detail.Thirdly, the dynamics model of a composite wing-box is established by finite elementanalysis software ANSYS, both strain modal matrix and displacement modal matrix are gainedby modal analysis. Finally, ASGA is taken as optimization method, both strain modal assurance criterion andmodal assurance criterion are used as optimization criterion, the results of two methods meet thedesign requirements. In addition, in order to improve the visualization and easy use of theproposed algorithms, optimization software is developed by using GUI of MATLAB.It’s clearly indicate that the optimization result based on strain mode is better than thatbased on displacement mode, therefore, this method is feasible in optimal sensor placement ofcomposite laminate aimed at damage diagnosis.