IDENTIFICATION ACCURACY EVALUATION OF OPTIMIZED RANDOM SUBSPACE METHOD BASED ON DATABASE

The evaluation of modal identification accuracy runs through the whole process of modal recognition research, but the evaluation method of modal identification accuracy based on a single or a few examples is accidental. Therefore, a method for evaluating modal identification accuracy based on structural response database is proposed. The key step of database construction is to calculate the structural response based on modal parameters. A database construction method is proposed to construct the physical model of structures (mass, stiffness and damping matrix) based on modal parameters, and then calculate the structural response, and the feasibility is verified with an example. Random subspace legal order is difficult, and false mode elimination is difficult; An optimized random subspace method based on singular entropy incremental order determination and two-stage spurious mode elimination in steady state diagram is proposed, and the identification accuracy of the optimized random subspace method is evaluated by database method. The results show that the method based on database is feasible. The optimized random subspace method has high identification accuracy of frequency and mode shape, but low identification accuracy of damping ratio.