Due to the field disturbance of thermal power unit and the complexity of reheat steam temperature

both the conventional identification methods and basic particle swarm optimization(PSO) algorithm have inadequacies of convergence speed and optimization precision in reheat steam temperature model's identification. A particle swarm optimization algorithm based on natural selection and adaptive dynamic weight(D-SPSO) was adopted. The algorithm balanced global searching ability and local improvement ability of the PSO algorithm

and improved the searching performance. Moreover

compared with the basic PSO algorithm

the D-SPSO algorithm didn't introduce new control parameters

avoiding the negative effect of additional parameters on model identification precision. The reheat steam temperature transfer function models of basic PSO algorithm and D-SPSO algorithm were verified by simulation data and an ultra-supercritical 660 MW pressure once-through boiler's on-site operation data. The results show that

the reheat steam temperature obtained by the D-SPSO algorithm was closer to the actual value. Besides

the optimization performance and search capability were significantly improved.