基于MATLAB的RBF神经网络在离心压缩机性能预测中的应用

The Application of RBF Neural Network Based on MATLAB for the Performance Prediction of Centrifugal Compressors

离心压缩机性能预测在其研究和应用中有着十分重要的作用。利用RBF神经网络,以离心式压缩机性能预测为目标,应用MATLAB7.0分两种情况建立了性能预测仿真模型。首先,建立了转速恒定时的预测模型;其次,建立了转速变化时的预测模型。通过检验样本,结果显示网络的预测效果与实测效果吻合比较好。该方法基于有限的实验数据,可以有效预测离心压缩机的性能。经过对两个典型性能预测进行仿真,结果表明RBF神经网络可以应用在离心压缩机性能预测中。

The Performance prediction of centrifugal compressor takes a very research and application. The performance prediction model has been built network, taking the performance prediction of centrifugal compressor as the target in setting two cases. At first the model is based on an invariable speed, then ba the result has shown that the network result is close to real result. The method important role in its based on RBF neural , applying MATABT. 0 sed on variable speed, is only based on finite experiment data, but it can predict the performance of centrifugal compressor effectively. After two typical performances were emulated, it is proved that RBF neural network can be used in the performance prediction of centrifugal compressor.