粒子滤波在机械密封端面接触状态声发射监测中的应用

Application of Particle Filter on Mechanical Seal Contact Condition Monitoring of Acoustic Emission

机械密封端面运行过程中所产生的声发射信号在传递过程中容易受到环境噪声的干扰,难以有效地从背景噪声中分离出来。研究粒子滤波技术在机械密封端面膜厚及开启状态声发射监测中的应用。将声发射传感器安装在机械密封静环座上,对动静环端面开启状态进行外部间接检测;运用粒子滤波技术处理采集的声发射信号,提取信号时域、频域及小波包能量特征;建立BP神经网络模型,对机械密封端面开启状态及膜厚进行识别。结果表明:粒子滤波技术能够有效地将密封端面产生的信号从背景噪声中分离出来;通过BP神经网络对提取的特征值进行模式识别,实现了密封端面膜厚变化范围的间接测量。该方法分析结果与电涡流传感器直接测量所得到的结果完全一致。

The acoustic emission signal generated by the end face of mechanical seals is easy to be disturbed by the background noise in the transmission process, and it is difficult to separate from the background noise. The application of particle filter technology on monitoring the end face film thickness and open condition of mechanical seals by acoustic emission was studied. Acoustic emission sensor was installed in stationary ring seat of the mechanical seal to detect the contact condition indirectly of the dynamic and static rings. The collected acoustic emission signals were processed by particle filter, and the signal features of time domain, frequency domain and wavelet packet energy was extracted. BP neural net- work model was established to identify the film thickness and open condition of mechanical seals. The result shows that the particle filter technology can effectively separate the signals of seal face from the background noise. By using BP neural network to identify the pattern of the extracted characteristic parameters, the indirect measurement of film thickness varia- tion is achieved. The measurement result by this method is the same as the directly measurement result by eddy current sensor.