A modified clustering procedure for energy consumption monitoring in the steam turbine considering volume effect

A new procedure for energy consumption characteristics of steam turbine is proposed in this paper. The heat rate index evaluates heat energy needed for electricity generation, and fluctuates greatly in real-time operation. The volume effect in the system is thus considered to modify the traditional real-time heat rate calculation. Particle swarm optimization (PSO) algorithm is inserted in the traditional Fuzzy C-means (FCM) to find the optimal initial clustering centers. Three clustering evaluating indicators are further added for the clustering number's adaptive searching. The modified clustering method is then tested to be effective by UCI datasets. The operation data from a 660 MW steam turbine system are then taken for the industrial case study, and the heat rate index values are then calculated by the proposed method with the volume effect taken into consideration. After steady working condition selection, the modified clustering method proposed in the paper is then utilized under the whole working condition. The clustering results are showed in graphs with the power output and temperature as coordinate axes, and coincide well with actual operation of the steam turbine system. The proposed procedure can be taken as a new guidance to obtain the energy consumption indexes' target condition library in steam turbine system.