基于灰色理论、径向基神经网络的大型压缩机故障预测方法的研究

Research on Fault Forecast of Great Compressor Based on Gray Theory and RBF Neural Net

为解决大型空气压缩机工作状态监测与故障预测这一复杂问题,从模型设计基础出发,创立了基于灰预测理论的径向基神经网络预测模型,成功突破了传统神经网络故障预测模型的局限。该预测模型充分提取了设备在过去时刻的运行信息,利用灰预测模型,对未来的特征参量进行长期预测,避免了由未来网络输入的不确定性导致的预测不确定性问题。对压缩机的故障预测实验表明,该模型具有明显的预测可靠性。

In order to monitoring and forecasting the machine′s working state fault,we establish a forecasting model based on gray theory and RBF neural net.This model successes to break through the limitation of forward ANN and feedback ANN.It predicts machine′s future featured parameters considering the past and current machine′s information and avoids the prediction′s indeterminacy caused by unsure future net input.Experiment on compressor has proved that this model is feasible and has a high forecasting precision.