In this paper, a novel data-driven fault diagnosis method by combining deep canonical variate analysis and Fisher discriminant analysis (DCVA-FDA) is proposed for complex industrial processes. Inspired by the recently developed deep canonical correlation analysis, a new nonlinear canonical variate analysis called deep canonical variate analysis (DCVA) is first developed by incorporating deep neural networks into canonical variate analysis. Based on DCVA, a residual generator is designed for the fault diagnosis process. Fisher discriminant analysis is applied in the feature space spanned by residual vectors. Then, a Bayesian inference classifier is performed in the reduced dimensional space of FDA to label the class of process data. A continuous stirred-tank reactor and an industrial benchmark of the Tennessee Eastman process are carried out to test the performance of DCVA-FDA fault diagnosis. The experimental results demonstrate that the proposed DCVA-FDA fault diagnosis is able to significantly improve the fault diagnosis performance when compared to other methods also examined in this study.

Incipient fault detection plays a crucial role in preventing the occurrence of serious faults or failures in industrial processes. In most industrial processes, linear, and nonlinear relationships coexist. To improve fault detection performance, both linear and nonlinear features should be considered simultaneously. In this article, a novel hybrid linear-nonlinear statistical modeling approach for data-driven incipient fault detection is proposed by closely integrating recently developed canonical variate dissimilarity analysis and mixed kernel principal component analysis (MKPCA) using a serial model structure. Specifically, canonical variate analysis (CVA) is first applied to estimate the canonical variables (CVs) from the collected process data. Linear features are extracted from the estimated CVs. Then, the canonical variate dissimilarity (CVD) which quantifies model residuals in the CVA state-subspace is calculated using the estimated CVs. To explore the nonlinear features, the nonlinear principal components are extracted as nonlinear features through performing MKPCA on CVD. Fault detection indices are formed based on Hotelling's T ² as well as Q statistics from the extracted linear and nonlinear features. Moreover, kernel density estimation is utilized to determine the control limits. The effectiveness of the proposed method is demonstrated by the comparisons with other relevant methods via simulations based on a closed-loop continuous stirred-tank reactor process.

Bearings are the key and important components of rotating machinery. Effective bearing fault diagnosis can ensure operation safety and reduce maintenance costs. This paper aims to develop a novel bearing fault diagnosis method via an improved multi-scale convolutional neural network (IMSCNN). In traditional convolutional neural network (CNN), a fixed convolutional kernel is often employed in the convolutional layer. Thus, informative features can not be fully extracted for fault diagnosis. In the proposed IMSCNN, a 1D dimensional convolutional layer is used to mitigate the effect of noise contained in vibration signals. Then, four dilated convolutional kernels with different dilation rates are integrated to extract multi-scale features through the inception structure. Experimental results from the popular CWRU and PU datasets show the superiority of the proposed method by comparison with other related methods.

In the blast furnace iron-making process (BFIP), there still has been a significant push to maintain a stable process and ensure maximum efficiency. Although some control systems can compensate for multiple types of disturbances, some significant process faults always require precise human intervention to avoid safety hazards. Therefore, it is crucial to develop an efficient and stable diagnosis system to identify these faults quickly. This article focuses on a novel approach called deep stationary kernel learning support vector machine (DSKL-SVM) for BFIP fault diagnosis. To eliminate the impact of nonstationary property on modeling, analytic stationary subspace analysis (ASSA) is adopted to estimate consistent underlying features. Then, design a multilayer stacked deep kernel network to explore deep nonlinear information further. A DSKL-SVM classifier and the corresponding two-tier model optimization algorithm are constructed to isolate different data types. At last, a series of case studies based on actual BFIP present the effectiveness of DSKL-SVM. The results demonstrate that the proposed method has an outstanding effect on fault diagnosis, and it is verified that the performances of stationary projection (SP), deep stationary nonlinear (DSN) construction, and online computation times are superior to other methods.

For the actual blast furnace ironmaking process (BFIP), sophisticated dynamic, nonlinear and nonstationary characteristics make it hard to be modeled accurately with conventional monitoring methods. In this paper, local dynamic broad kernel stationary subspace analysis (Local-DBKSSA) is developed to improve the monitoring performance. Faced with complex dynamic nonlinear characteristics, a single model is considered to be unable for accurate representation. Thus, dynamic broad nonlinear features established by time shift and multi-kernel projection are adopted from more perspectives. Subsequently, the above features are integrated into stationary subspace analysis (SSA) to estimate stationary projections from time-varying data. In order to reduce the impact of large fluctuations and improve fault detection capability, a local statistic is further proposed. The effects of nonstationary characteristic on monitoring capability and the excellent performance of the local statistic are also theoretically analyzed. Finally, a case study based on actual BFIP data presents that the proposed method can discriminate between normal and sample faults more accurately and timely, and has better robustness to nonstationary perturbations under normal conditions by providing fewer false alarms.

Bearings are indispensable and key components in rotating machinery. To ensure the safe and reliable operation of rotating machinery, bearing fault diagnosis plays a crucial role. To explore the spatial and temporal information in vibration signals, a novel bearing fault diagnosis method is proposed by combining a deep residual shrinkage network (DRSN) and bidirectional long short-term memory (Bi-LSTM) network in this study. Firstly, a DRSN is employed to extract the spatial features from noise-related vibration signals. Then, a Bi-LSTM network is adopted to further address the long-term dependencies problem in vibration signals, where the temporal information is exploited. By integrating DRSN and Bi-LSTM, the spatial and temporal information of vibration signals is fully extracted. Finally, a fully connected layer with Softmax is used to offer the diagnostic results. Experimental results using two case studies demonstrate the effectiveness of the proposed method by comparison with other state-of-the-art methods.