论文部分内容阅读
将云模型与BP(back propagation)神经网络以串联方式有机结合,首先利用云变换方法进行网络的结构辨识和云模型的特征提取,同时通过在输入层引入单位延时环节描述发动机工作过程动态特性,研究提出了基于动态云BP网络的液体火箭发动机故障诊断方法.结合实际试车数据的验证结果表明,该方法能够准确识别发动机已有的3种故障模式,通过在试车数据中添加0期望、0.2标准差的随机噪声的方法来模拟环境噪声和测试过程中产生的随机噪声,根据持续性原则,方法仍能够正确进行故障检测与分类.方法单步运行时长为1.124×10-4 s,完全能够满足实时性要求.
The cloud model and BP (back propagation) neural network are organically connected in series. Firstly, the cloud structure is used to identify the structure of the network and extract the features of the cloud model. At the same time, the dynamic characteristics of engine working process , A new fault diagnosis method for liquid propellant rocket engine based on dynamic cloud BP network was proposed.According to the actual test data validation results show that this method can accurately identify the three existing failure modes of the engine by adding 0 expectation, 0.2 Standard deviation of the random noise method to simulate the environmental noise and random noise generated during the test, according to the principle of persistence, the method can still correctly fault detection and classification.Methods Single-step run-time of 1.124 × 10-4 s, fully capable Meet real-time requirements.