论文部分内容阅读
为解决液体火箭发动机故障预测这一难题,提出一种基于误差预测修正的故障预测方法。在历史数据的基础上建立小波过程神经网络故障预测模型,同步计算学习样本的预测误差,根据上述误差建立双并联离散过程神经网络预测模型。预测时,将预测误差值实时补偿到小波过程神经网络预测模型以提高预测精度。通过液体火箭发动机地面试验中的涡轮泵数据对该方法进了验证。结果表明,该方法在预测精度和适应能力上较单一的过程神经网络预测模型有显著提高,进行10步预测时,预测值的标准化均方根误差为0.392,预测平均耗时为76ms,能够用于解决液体火箭发动机故障预测问题。
In order to solve the problem of liquid propellant rocket engine fault prediction, a fault prediction method based on error prediction and correction is proposed. Based on the historical data, a wavelet neural network fault prediction model is established, and the prediction errors of learning samples are calculated simultaneously. Based on the above error, a BP neural network prediction model is established. When predicting, the prediction error value is compensated to the wavelet neural network prediction model in real time to improve the prediction accuracy. The method was validated by turbo pump data from a liquid rocket engine ground test. The results show that the proposed method has significantly improved prediction accuracy and adaptability compared with a single neural network prediction model. When the 10-step prediction is carried out, the standardized root mean square error of prediction value is 0.392 and the average prediction time is 76ms. To solve the problem of liquid rocket engine failure prediction.