The invention discloses a sensitivity size maximization based virtual mass optimization method, and belongs to the field of structural damage identification. A matrix size is used to describe irrelevance among different vectors of a matrix, an optimized target function is constructed into the size of a sensitivity matrix, and a maximized target function is used to optimize the virtual mass. According to the method based on the optimization criterion of maximizing the size of the sensitivity matrix, the matrix size serves as a standard of irrelevancy among the column vectors of measuring pointsin the measuring matrix, the matrix size serves as the optimized target function, the target function is maximized to obtain the optimization sequence of virtual mass arrangement positions, it is ensured that the sensitivity matrix contains information as much as possible, and the damage identification precision is ensured. Operation is convenient and easy to realize, the computing efficiency andpractical value are high, and the method can be widely applied in future.