Prediction Model of Centrifugal Fan Performance Based on BP Neural Network

In view of the G4-73 type backward centrifugal fan widely used in power plant, training samples for orthogonal test is conducted. A Centrifugal fan performance parameters prediction model based on BP neural network is built. Through the contraction between the predictive value and samples, the accuracy of prediction model is verified. The maximum relative errors of total pressure and efficiency are respectively 0.974% and 0.3327% with the trained neural network to predict on each single test sample, which reaches the predetermined training accuracy, and can be used for fan performance prediction.