Abstract: By analyzing the factors which affected the main-steam temperature, a multi-variable model was introduced to overcome the bad result of single variable cascade control. An improved differential evolution algorithm was proposed including mutation strategies random selection, crossover ratio and mutation ratio adaptive adjustment, which was used for closed-loop identification of main-steam multi-variable transfer function model. The principle and method how to obtain valid identification data from power plant distributed control system history database was introduced, and the data from some 1 000 MW coal-fired power plant was used to identify and verify the main-steam temperature multi-variable model, the verification result shows the validation of model. By further analyzing the multi-variable model identification result, an optimization method was suggested for the conventional main-steam cascade control.

Key words: main-steam temperature, closed-loop, multi-variable, differential evolution algorithm