基于改进证据理论的中速磨煤机故障诊断

Fault Diagnosis of Medium Speed Coal Mill Based on Improved Evidence Theory

作为锅炉的重要辅机设备,磨煤机现场工作环境复杂,各种参数的变化起伏不同,故障时其运行信号容易出现冲突、噪声和延迟等问题。针对此问题,提出了一种小波去噪重构与改进后的证据理论相结合的目标模式识别方法。首先通过故障状态时主要参数的变化确定原始证据信号并利用典型样本法来构造信度函数。在此基础上,利用峰效比和相关函数选择最佳小波基来对原始证据信号的数据进行去噪重构处理;同时,利用信度函数获得处理后证据信号数据的信度值。最后引进证据权重改进证据理论融合公式,运用改进后的证据理论进行融合得出识别结果。磨煤机实际运行结果表明,该方法能提前80 s判断磨煤机处于故障状态,有效改善信号延迟和冲突的问题,且对目标模式的支持度增加,提高了故障状态识别的准确性。

As an important auxiliary equipment of the boiler,the working environment of the coal mill is complex,and the fluctuation of various parameters is different.When the fault occurs,the running signal is prone to conflict,noise and delay.To solve this problem,a target pattern recognition method based on wavelet de-noising reconstruction and improved evidence theory was proposed.Firstly,we determined the original evidence signal according to the change of main parameters in fault state,and constructed the reliability function with typical sample method.On this basis,the best wavelet basis was selected by peak to efficiency ratio and correlation function to denoise and reconstruct the original evidence signal.At the same time,the reliability value of the processed evidence signal data was obtained by using the reliability function.Finally,the weight of evidence was introduced to improve the fusion formula of evidence theory,and the improved evidence theory was used for fusion to get the recognition results.The actual operation results of coal mill show that the method can judge the fault state of coal mill by 80 s in advance and effectively improve the problem of signal delay and conflict.It can also increase the support of target mode and improve the accuracy of fault state recognition.