KPCA-Based Visual Fault Diagnosis for Nonlinear Industrial Process

Fault diagnosis plays an important role in complicated industrial process. It is a challenging task to detect, identify and locate faults quickly and accurately for large-scale process system. To solve the problem, a novel MultiBoost-based integrated ENN (extension neural network) fault diagnosis method is proposed. Fault data of complicated chemical process have some difficult-to-handle characteristics, such as high-dimension, non-linear and non-Gaussian distribution, so we use margin discriminant projection(MDP) algorithm to reduce dimensions and extract main features. Then, the affinity propagation (AP) clustering method is used to select core data and boundary data as training samples to reduce memory consumption and shorten learning time. Afterwards, an integrated ENN classifier based on MultiBoost strategy is constructed to identify fault types. The artificial data sets are tested to verify the effectiveness of the proposed method and make a detailed sensitivity analysis for the key parameters. Finally, a real industrial system—Tennessee Eastman (TE) process is employed to evaluate the performance of the proposed method. And the results show that the proposed method is efficient and capable to diagnose various types of faults in complicated chemical process.

This work concerns the issue of quality-related fault detection and diagnosis (QrFDD) for nonlinear process monitoring. A kernel principal component analysis (KPCA) based canonical correlation analysis (CCA) model is proposed in this paper. First, KPCA is utilized to extract the kernel principal components (KPCs) of original variables data to eliminate nonlinear coupling among the variables. Then, the KPCs and output are used for CCA modeling, which not only avoids the complex decomposition of kernel CCA but also maintains high interpretability. Afterwards, under the premise of Gaussian kernel, a proportional relationship between process variables sample and kernel sample is introduced, on the basis of which, the linear regression model between process and quality variables is established. Based on the coefficient matrix of the regression model, a nonlinear QrFDD method is finally implemented which has both the data processing capability of nonlinear methods and the form of linear methods. Therefore, it significantly outperforms existing kernel-based CCA methods in terms of algorithmic complexity and interpretability, which demonstrated by the simulation results of the Tennessee-Eastman (TE) chemical process.

In order to improve the accuracy and stability of industrial fault detection and diagnosis, this paper introduces the deep learning theory and proposes an improved Deep Belief Networks (DBNs). In the first, this paper introduces the “centering trick” in the pre-training process of network. This method is done by subtracting offset values from visible and hidden variables. Then, in the process of network fine-tuning, the weights of network are added the White Gaussian Noise. This improved Deep Belief Networks combines with Kernel Principal Component Analysis (KPCA) into KPCA-CDBNs model. This paper applies this model into fault detection research of Tennessee Eastman (TE) and shows that the method of this paper has higher prediction and stability of fault detection and diagnosis.

Dynamic artificial immune system with variable selection based on causal inference

Fault detection and diagnosis for non-linear processes empowered by dynamic neural networks

With the coming of the big data era, the data collected in the process industry shows features of high volume, high-dimensional and non-linear. Meanwhile, these process data present imbalanced feature, leading to a lack of fault information. These problems mentioned above have brought difficulties to fault diagnosis. To solve the above difficulties, a new synthetic minority over-sampling technique (SMOTE) considering the correlation of sample integrated with locality sensitive discriminant analysis (LSDA) and LightGBM fault diagnosis methodology (CSMOTE-LSDA-LightGBM) is proposed in this article. In our proposed methodology, firstly, the SMOTE fully considering correlation (CSMOTE) which uses both Euclidean and Mahalanobis distance to calculate the nearest neighbor relationship is used to resample the imbalanced samples and expand the number of small classification fault samples; secondly, the LSDA is used to dimensionality reduction (DR) to extract the fault-related critical features; finally, the LightGBM classifier is used for fault classification. The Tennessee Eastman (TE) process case is selected for simulation to verify the effectiveness of the proposed CSMOTE-LSDA-LightGBM in fault diagnosis. The simulation results of TE process case show that the proposed method has improved the accuracy of fault diagnosis compared with imbalanced data and traditional DR methods indicating the CSMOTE-LSDA-LightGBM methodology is applicable to fault diagnosis for imbalanced samples.

A large amount of noise is mixed in the actual process industrial data, which leads to a decline in data quality, which in turn affects the diagnosis results. In reality, the wavelet threshold method is generally used to remove noise. However, the hard threshold method which is in the wavelet threshold method has the disadvantages of poor signal continuity and signal oscillating. The soft threshold method has the disadvantages of constant deviation and large error of reconstructed signal. To this end, we propose a random soft-hard threshold algorithm which can combine hard and soft thresholds. Then the new algorithm is combined with the principal component analysis for fault diagnosis of the TE (Tennessee Eastman) process. The MATLAB simulation results show that the above method can effectively denoise, identify and detect faults more accurately, and can better find the variables that have the greatest impact on the fault.

Nowadays, chemical processes are becoming more complex, and so the safety requirement is getting higher and higher. Intelligent and effective fault detection and diagnosis are becoming more and more important. However, complex industrial systems generate high-dimensional data, which has bad influence on fault detection and diagnosis. Thus, it is necessary to handle industrial data with high dimensionality. Discriminant Locality Preserving Projections (DLPP) has attracted much interest as a dimensionality reduction method. However, there is a small size sample problem when the data dimension is higher than the number of data classes. Under this condition, DLPP cannot achieve acceptable performance in reducing data dimension. In order to solve this problem, this paper proposes a novel DLPP based on improved Synthetic Minority Oversampling Technique (SMOTE-DLPP). The improved SMOTE is used to sample the original data set to generate new data sets so that the number of data classes is basically the same as the number of data dimension. Simulation results on the Tennessee Eastman process (TEP) show that the proposed SMOTE-DLPP can achieve acceptable performance in fault diagnosis.

Kernel principal component analysis (KPCA) has become a popular technique for process monitoring in recent years. However, the performance largely depends on kernel function, yet methods to choose an appropriate kernel function among infinite ones have only been sporadically touched in the research literatures. In this paper, a novel methodology to learn a data-dependent kernel function automatically from specific input data is proposed and the improved kernel principal component analysis is obtained through using the data-dependent kernel function in traditional KPCA. The learning procedure includes two parts: learning a kernel matrix and approximating a kernel function. The kernel matrix is learned via a manifold learning method named maximum variance unfolding (MVU) which considers underlying manifold structure to ensure that principal components are linear in kernel space. Then, a kernel function is approximated via generalized Nystrom formula. The effectiveness of the improved KPCA model is confirmed by a numerical simulation and the Tennessee Eastman (TE) process benchmark.

The complexity and dynamism of business management system pose higher requirements for the accuracy and timeliness of fault diagnosis. This paper proposes a compensation distance evaluation technique-kernel principal component analysis-convolutional neural network-bidirectional long short-term memory network (CDETKPCA-CNN-BiLSTM) that integrates attention mechanism to address the limitations of traditional diagnostic methods in nonlinear and high-dimensional data scenarios. The bidirectional long short-term memory (BiLSTM) layer and attention mechanism layer further improve the accuracy and reliability of fault diagnosis. Feature extraction is performed in business management system data from both time and frequency domains, effectively utilizing temporal information to form an initial feature set. To address the issue of data redundancy in business management system, a compensation distance evaluation technique and kernel principal component analysis (CDETKPCA) feature fusion method is proposed. Through CDET, the initial feature set is screened and weighted to guide KPCA feature fusion processing, generating a fused feature set for subsequent fault diagnosis research. The experimental results show that CDETKPCA-CNN-BiLSTM can extract effective information more efficiently and significantly improve analysis accuracy. And this provides a new technical method for fault diagnosis in business management system.

Fault detection and diagnosis based on transfer learning for multimode chemical processes

Most chemical processes, such as distillation, absorption, extraction, and catalytic reactions, are extremely complex processes affected by multiple factors. As a result, the relationships between their input and output variables are non-linear, and it is not easy to optimize or control them using traditional methods. Artificial neural network is a systematic structure composed of multiple neuron models. By simulating many basic functions of the nervous system of living organisms, nonlinear control can be realized without relying on mathematical models, and it is especially suitable for more complex control objects. This article will introduce artificial neural networks' basic principles and development history, and review its application research progress in chemical process control, fault diagnosis, and process optimization.

Fault identification and diagnosis based on KPCA and similarity clustering for nuclear power plants

Fault detection and diagnosis (FDD) has received considerable attention with the advent of big data. Many data-driven FDD procedures have been proposed, but most of them may not be accurate when data missing occurs. Therefore, this paper proposes an improved random forest (RF) based on decision paths, named DPRF, utilizing correction coefficients to compensate for the influence of incomplete data. In this DPRF model, intact training samples are firstly used to grow all the decision trees in the RF. Then, for each test sample that possibly contains missing values, the decision paths and the corresponding nodes importance scores are obtained, so that for each tree in the RF, the reliability score for the sample can be inferred. Thus, the prediction results of each decision tree for the sample will be assigned to certain reliability scores. The final prediction result is obtained according to the majority voting law, combining both the predicting results and the corresponding reliability scores. To prove the feasibility and effectiveness of the proposed method, the Tennessee Eastman (TE) process is tested. Compared with other FDD methods, the proposed DPRF model shows better performance on incomplete data.

Taking the common continuous chemical reaction process in industry as the research object, this paper analyzes the function of instruments and equipment, system process flow and sequence control requirements of each part of chemical reaction process. Combined with the object control circuit and operation experience, taking safety, stability and energy saving as the production goal, the start-up and stop sequence control system of chemical reaction process is designed, and the flow chart is given. Based on SIMATIC PCS7 control platform and SMPT-1000 as an object, the experiment was carried out. The experimental results show that the operation of the system is stable and the performance indexes are controlled within the requirements of safe and efficient operation of the system, i.e. the flow of final product, product concentration, reactor temperature, reactor liquid level, flash tank pressure, flash tank liquid level, mixing tank liquid level and preheater temperature are all kept within the safe production range, which can meet the requirements of sequence control of chemical reaction process.