Objectives  High fidelity finite element models and effective load identification technology are very important for ship structure health monitoring and evaluation. Therefore, a model updating and load identification method based on improved particle swarm optimization (PSO) is proposed.

  Methods  The Rastrigin function is used to compare the improved PSO algorithm with the classical PSO algorithm. An I-beam structure is adopted with pressure applied at the middle of the beam. A limited number of strain sensors are pasted on the surface and divided into two sets: the measured set for model correction and load identification, and the monitoring set for verification. The elastic modulus of the block division of the I-beam is modified and verified, and the load identification based on the modified numerical high fidelity model is verified.

  Results   In the test of the improved PSO algorithm by Rastrigin function, it shows better global optimal solution searching ability under different particle numbers. In the I-beam experiment, the elastic modulus of the two parts of the partition converges to the optimal solution after 23 iterations. By comparing the strain data of the test monitoring points with the data of the numerical calculation results after model correction, the relative error of the strain is within 2%, which verifies the correctness of the model updating method. In addition, the external load pressure of the structure is identified by the load identification method, and the error between the identified calculated value and the load applied in the test is within 2%. The maximum error between the strain value of the monitoring points calculated by the combination of the identified load and modified model and test data is 3.74%, which verifies the effectiveness of the load identification.

  Conclusions  The proposed method has a good precision performance in the inversion of the global state of the structure, and can provide technical support for hull structure health monitoring, residual life prediction and predictive maintenance.