工艺凝液泵机械密封故障自动检测方法研究及应用

Study and Application of Automatic Detection Method for Mechanical Seal Failure of Process Condensate Pump

针对工艺凝液泵机械密封受环境和自身机械扰动影响易产生失稳故障的问题,构建了基于谱特征分析的故障检测方法:采取大数据传感信息感知方法对故障原始样本信息采样;对采集的的数据进行信号拟合并进行多尺度分解;根据分解的频谱特征分布的差异性进行故障辨识;采用人工神经网络学习算法实现故障自动检测和自适应学习训练。对比分析表明,该方法检测概率提高8%,具有更好的故障辨识和特征分析能力。

The mechanical seal of process condensate pump is prone to occurs instability fault influenced by environment and self-mechanical disturbance,so it established a failure detection method based on spectral characterization analysis;the big data sensing information perception method is adopted for sampling the original failure sample information;the signal fitting of the collected data is done to perform multi-scale decomposition;the difference of the decomposed spectral characteristics distribution is used for fault identification;the failure automatic detection and adaptive learning training is realized by artificial neural network learning algorithms.The comparison and analysis results show,by using the method,the detection probability increases 8%,it has better fault identification and feature analysis capabilities.