To provide comprehensive and accurate quantitative evaluation of power quality， this paper first quantifies the transient indices of power quality and takes them into account in the synthetic evaluation， and then introduces two similarity theory based intelligent assessment methods into the evaluation， namely， the particle swarm optimization-Shepard similarity interpolation（PSO-SSI） algorithm and particle swarm optimization-ideal interval method（PSO-IIM）. Test results indicate that both methods can accurately identify the differences at the same power quality level. Moreover， the PSO-SSI algorithm is relatively intuitive and convenient， but its performance is dependent on the sample characteristic， while PSO-IIM decreases this dependency， but is relatively complicated.