The inverse matrix of frequency response function is a common method for load identification in practice. The key point of the method is to solve the ill-conditioned problem, which was dealt with Tikhonov regularization and singular value decomposition (SVD) in the paper. Simulation of a plate was carried out to compare the load identification accuracy of the two methods under different ill-conditions, which were evaluated by the condition number of the normal matrix of frequency response function. Conclusions show that Tikhonov regularization is more suitable when the condition number is greater than 1000 and SVD method is better for other situations. What’s more, a novel method is proposed to identify loads with the combination of the two methods based on the condition number. Comparing with SVD and Tikhonov regularization methods through results of simulation and experiment, the novel method can obviously improve the load identification accuracy, and provides a certain guide for practical engineering.