Back pressure prediction of the direct air cooled power generating unit using the artificial neural network model

In addition to the operating parameters, there were numerous factors, including the meteorological and the geographic conditions, as well as the atmospheric environmental conditions, which could affect the performance of the direct air-cooled power generating unit. In the present study, the artificial neural network (ANN) approach was employed to model the back pressure of the steam turbine, one of the most important parameters of the power generating unit. Based on the actual operating data obtained from the on-site experiments of the direct air-cooled power generating unit in north China, the three-layers back propagation ANN model was trained and tested to predict the back pressures of the steam turbine unit under the different operating conditions. The mean relative error (MRE) of the present ANN model was 9.273%, the root mean square error (RMSE) was 1.83 kpa, and the absolute fraction of variance (R2) was 0.9859, which indicated that the predictions agreed well with the actual values. The present ANN model can also reflect the effects of the weather conditions on the back pressure of the unit, such as the rain or the sandstorm and the air humidity. The influence of the environmental natural wind on the unit performance can be described with robustness and reliability by the present ANN model as well.