Predicting the Performance of Helico-Axial Multiphase Pump Using Neural Networks
The main geometric structural parameters which affected the performance of the compression cell of the helico axial multiphase pump greatly were selected as the research object. The groups of impeller parameters were determined by the orthogonal experimental design method. Then the pressure rise and efficiency for each group which were obtained through numerical simulation according to CFD method were used as the training samples and testing samples in the artificial neural network forecasting process. Two neural network topology structures were determined based on the Back Propagation Neural Network and Radial Basis Function Neural Network respectively. The structure parameters got from the orthogonal design method were used as the input layer data, and the performance parameters from numerical simulation were used as output layer data. After a training progress, two performance prediction models for the helico axial multiphase pump were established based on the BP and RBF respectively. The testing results showed that the average relative errors for pressure rise and efficiency in the BP network prediction model and were 9.97％ and 7.9％ respectively, while those in the RBF network prediction model were 7.84％ and 5.85％ respectively.
multiphase pump performance prediction neural network BP RBF
Jinya Zhang Hongwu Zhu Huan Wei Zhuowei Li Lei Xiong
Faculty of Mechanical and Electronic Engineering, China University of Petroleum (Beijing), China
国际会议
长沙
英文
2088-2091
2010-05-11（万方平台首次上网日期，不代表论文的发表时间）