Abstract
Process monitoring and fault diagnosis (PM-FD) of coal mills are essential to the security and reliability of the coal-fired power plant. However, traditional methods have difficulties in addressing the strong nonlinearity and multi-modality of coal mills. In this paper, a novel multi-mode Bayesian PM-FD method is proposed. Gaussian mixture model (GMM) is first applied to identify the operating modes of the coal mill. Subsequently, combined with multi-output relevance vector regression (MRVR), Bayesian inference is introduced to reconstruct and monitor the newly observed samples from different running modes. Additionally, the squared prediction error and the contribution plot method are employed for fault detection and isolation. The performance of the proposed PM-FD method is verified through its application in a self-defined nonlinear system and two actual fault cases of a medium-speed coal mill. Compared with the traditional methods, the experimental results demonstrate the effectiveness of the proposed method.
Highlights
Coal mills are crucial equipment of the coal-fired power plant’s pulverizing system, in which the raw coal is crushed and ground into coal powder
Effective Process monitoring and fault diagnosis (PM-FD) methods of coal mills are urgently needed for the operational security and reliability of the power plant
The highlight of Bayesian multi-output relevance vector reconstruction (BMRVR) is that a probabilistic mapping between inputs and targets is established, and the requirement for faulty training data is eliminated
Summary
Coal mills are crucial equipment of the coal-fired power plant’s pulverizing system, in which the raw coal is crushed and ground into coal powder. Ge and Song [24] discussed and confirmed the effectiveness of applying distributed PCA to plant-wide process monitoring These methods do help improve the PM-FD performance, but lack of robustness and with inadequate monitoring accuracy for complex nonlinear systems [25]. The highlight of BMRVR is that a probabilistic mapping between inputs and targets is established, and the requirement for faulty training data is eliminated Another important aspect is that coal mills are always running at various different operating modes due to unit load fluctuations, resulting in the invalid assumption of a multivariate Gaussian distribution [43].
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
