Research On The Optimization Of Virtual Masses For Structural Damage Identification

Structural damage identification,as the core part of structural healt h monitoring system,has become a hot issue for domestic and foreign scholars in the field of civil engineering.The most widely used method is the damage identification method based on vibration characteristics,but the method has many problems: modalities are less sensitive to damage and less experimental modal data are not sufficient to accurately identify damage.This article uses an additional virtual mass method,which can effectively increase the amount of test data,while avoiding the difficulty of adding real physical components to the actual project.The size,number and position of the virtual mass have very important influence on the damage identification.Therefore,the optimization of the additional virtual mass is the main premise of the structural damage identification.This paper mainly studies the problem of the optimization of the additional virtual mass.The optimization of the virtual mass is similar to the optimization of the sensors,so the optimization of the sensor can be used for research.Firstly,we propose two kinds of virtual mass optimization criteria,namely,the optimization criterion based on maximizing the volume of the sensitivity matrix and the optimization criterion based on minimizing the conditions number of sensitivity matrix.The basic principle of the two and the construction of the objective function are studied in detail.The optimization rule based on the largest volume of the sensitivity information matrix uses the directed spatial volume formed by the column vectors of the sensitivity matrix to describe the irrelevance between the vectors.The objective function is constructed as the volume of the sensitivity information matrix.The optimization criterion based on the condition number of sensitivity information matrices is used to target the optimal damage identification.The condition number of the sensitivity matrix is used as the objective function.Both criteria are based on the sensitivity matrix.Therefore,this paper also introduces the calculation formula and construction method of the sensitivity information matrix.Then the paper discusses the virtual mass optimization algorithm used in this paper,including greedy algorithm,standard particle swarm algorithm and discrete particle swarm algorithm.The greedy algorithm obtains the optimal position of the mass position from point to point given a set of discrete points in advance.It has the advantages of easy understanding,simple operation,and high efficiency.The PSO is used to optimize the entire domain.The fitness function is expressed as a continuous function that optimizes the position of the measuring point,and the fitness is used as the basis for evaluating the position of the virtual mass.The basis is gradually approaching the optimal value by continuously adjusting the position and direction of the solution.The DPSO algorithm searches for the optimal combination from the given discrete measurement points.It also uses the fitness function to evaluate the value.The calculation efficiency is high,the search ability is strong,and it is not easy to fall into the local optimum.Finally,the finite element model of simply-supported beam is established to perform numerical simulation.Since the additional virtual mass will increase the sensitivity of the frequency to the local damage,but it will also cause the error of frequency identification,so the two factors should be considered to optimize the size of the virtual mass.Then the above two criteria are respectively used as the optimization objective function,and the greedy algorithm and the particle swarm algorithm are used to optimize position,the number of virtual mass is finally determined.According to the optimization scheme and the sensitivity method for damage identification,the effectiveness of the method are verified.