Abstract: Due to the shortcomings of strong subjectivity for selecting the parameters of ordered subsets-simultaneous algebraic reconstruction technique(OSSART), in this paper, a random particle swarm optimization algorithm with reconstruction area minimal error as fitness function is proposed to obtain the best reconstruction parameters. The quality of reconstruction from sparse angle of the asymmetric arc emission coefficient is evaluated. The results show that, compared with the maximum likelihood expectation maximum (MLEM) algorithm, the OSSART algorithm based on particle swarm optimization can not only reduce the reconstruction error significantly under the condition of large projection angle interval, but also have a stronger edge retention ability, which can effectively improve the reconstruction quality of the central area of the arc. In order to ensure the reconstructed quality of VPPA-MIG hybrid welding arc emission coefficient using OSSART algorithm, six feature line projections with equal spacing should be collected at least in the range of 180 degrees. The results provide a theoretical basis for the reliable study of asymmetric arc spectroscopic diagnostics.