量子超球神经网络在振动故障检测中的应用

The application of quantum hypersphere neural network in vibration fault detection

提出了一种基于量子超球神经网络的液体火箭发动机振动故障检测方法,采用可变量子超球代表发动机工作模式,自然地提供了反映故障程度的概率幅;网络的离线学习算法可以从训练样本中自动提取发动机振动知识,监测算法不仅能正确预报故障,还能在线学习新的振动信息。试验数据检验结果表明:量子超球神经网络可以成功用于液体火箭发动机振动故障检测。

The vibration fault detection approach for liquid propellant rocket engine is proposed based on quantum hypersphere neural network. The variable quantum hypersphere is used to mark work mode of engine, which provides probability amplitude that reflects the fault level. The off-line learning algorithm can extract engine vibration knowledge from training samples automatically. The monitor algorithm can forecast fault correctly, also learn new vibration information online. The analysis result of test data shows that quantum hypersphere neural network can be applied to vibration fault detection of liquid propellant rocket engine.