With the successful application of swarm intelligence algorithms in the back analysis of pavement structural parameters, the problems of complex multivariate nonlinear optimization have been solved, while how to choose an appropriate algorithm is always the urgent problem in the back analysis of pavement structural parameters. In view of the characteristics of the back analysis of pavement structural parameters，such as complex models, numerous inversion parameters, and quite time-consuming forward calculation procedures, eight common swarm intelligence algorithms are selected in this paper. Related researches on the performance of the algorithms under the limited number of forward calculation calls are carried out. In this paper, the swarm intelligence algorithm is further tested by taking the inversion problem of the pavement structure parameters considering the material transverse isotropy and the contact state between layers as an example. The research results show that different algorithms have their own characteristics. Among them, particle swarm optimization (PSO), genetic algorithm (GA), brain storm optimization (BSO), artificial bee colony (ABC) and fireworks algorithm (FWA) work better in multi-peak problems. The firefly algorithm (FA) has a faster rate of convergence when solving the problem of a flat area near the optimal solution. For genetic algorithms , the later rate of convergence of the real number coding method is higher than that of the binary coding method, but the search ability for multi-peak problems is weaker. Artificial fish-school algorithm (AFA) and shuffled frog leaping algorithm (SLA) have better optimization ability only under a larger number of forward calculation calls. For inversion of pavement structure parameters, PSO, GA, BSO and FA have good inversion results in deflection curve matching, while BSO can get the best inversion result in the view of correlation coefficient.