The optimization of the start-up scheduling for a 320 MW steam turbine

It is the aim of thermal power plants start-up scheduling to minimize the start-up time and limit the turbine rotor stresses, which is a highly nonlinear problem and has multiple local optima. A novel mathematical model of start-up schedule was proposed in this study. Based on the operation regulations of 320 MW subcritical condensing steam turbine at a power plant, different cold start-up optimization schedules were made according to the mathematical model of start-up schedule. The finite element method (FEM) was applied to calculate the temperature field and stress field of the steam turbine rotor under different schedules. Taking the calculation results as the sample data, support vector machines (SVM) was employed to establish the regression model of start-up schedules and the dangerous point stress under the start-up schedule. Combining the regression model, the optimal start-up parameters could be obtained by the particle swarm optimization (PSO) algorithm. The results show that limiting the maximum Von Mises stress, the start-up time of the optimized schedule was reduced by almost 150 min compared with the original schedule. At the same time the optimal start-up schedule was verified by FEM, and the maximum Von Mises stresses at the dangerous points satisfy the stress requirements.