Search

Traditional multivariate statistical-based process monitoring (MSPM) methods are effective data-driven approaches for monitoring large-scale industrial processes, but have a shortcoming in handling the redundant correlations between process variables. To address this shortcoming, this study proposes a new MSPM method called minimalist module analysis (MMA). MMA divides process data into several different minimalist modules and one more independent module. All variables in the minimalist module are strongly correlated, and no redundant variables exist; therefore, the extracted feature components in one minimalist module will not be disturbed by noise from the other modules. This study also proposes new monitoring indices and a fault localization strategy for MMA, and simulation tests demonstrate that MMA achieves superior performance in fault detection and localization. [ABSTRACT FROM AUTHOR]

To cope with the combined Gaussian and non-Gaussian features in the industrial processes, a novel preliminary-summation-based principal component analysis (PS–PCA) method is proposed in this study. Different from other approaches which improve principal component analysis (PCA) by changing its algorithm structure, PS–PCA just preprocesses the training and monitoring data without modification on PCA. According to the central limit theorem, PS–PCA adds up samples of each variable to make the distribution of the sum approach Gaussian distribution. These sums are then used for state monitoring. It has been proved that preliminary summation can increase the fault detection rate for Gaussian processes. Furthermore, some simulation tests substantiate that PS–PCA can improve the detection capability for non-Gaussian processes and even for nonlinear processes without increasing the computation load. [ABSTRACT FROM AUTHOR]

Partial least squares (PLS) is an algorithm commonly used for key performance indicator (KPI) industrial process monitoring in recent years. However, there are many shortcomings in PLS, such as uncertainty of the optimization solution, an imperfect optimization goal, and information impurity. To overcome these shortcomings, an analytical PLS (APLS) method is proposed in this study. APLS fully analyzes the correlation between process variables and quality variables, and is solved by an analytic solution to avoid the large computational complexity brought by iterative calculations. A computational complexity analysis of PLS and APLS is performed to verify the advantages of APLS in terms of computational complexity compared to PLS. To better further study the information impurity existing in PLS, we present the proof related to this problem. Moreover, in order to verify the effectiveness of APLS, a numerical example and the thermal power plant process are utilized. It can be seen that the proposed method has a better detection performance compared with existing PLS-related methods. • An analytical partial least squares method is proposed for process monitoring. • The analysis of information impurity problem existing in partial least squares (PLS) is performed. • APLS has lower computation complexity than PLS. • Numerical example and thermal power plant process data testing show that APLS has better monitoring performance than PLS. [ABSTRACT FROM AUTHOR]

• A kurtosis function is embedded into NCA to restrict the Gaussianity of PCs. • A novel PC selection mechanism is proposed. • The gradient descent method is rederived for the INCA network. • Testing results show that INCA outperforms NCA, KPCA and O-NLPCA. Neural component analysis (NCA) is one of the latest nonlinear multivariate statistical process control (MSPC) methods, which consists in combining artificial neural networks (ANNs) with principal component analysis (PCA). However, NCA cannot handle the non-Gaussian feature and the extracted principal components (PCs) in NCA may not be the key information in the process data. Herein, we propose an improved NCA (INCA) which introduces a new cost function based on kurtosis to restrict the Gaussianity of PCs. We also propose a novel PC selection mechanism based on the information of PCs in the original data space rather than in the PC data space. INCA achieves almost 100% detection rates in three different types of faults in a simulation model test, and it can detect the fault in the thermal power plant process more than 1 min ahead of orthogonal nonlinear PCA (O-NLPCA) and NCA. [ABSTRACT FROM AUTHOR]

Basal insulin infusion rate which should be adjusted to increase or decrease insulin delivery with the varying blood sugar level plays a key role in type 1 diabetes mellitus (T1DM) patients for maintaining the blood glucose level approximately steady within reference range in order to avoid the complications developed from diabetes. This paper proposes an effective hybrid particle swarm optimization (HPSO) algorithm for solving the basal insulin infusion rate problem. In HPSO, bad experience lesson learning scheme and local search based on chaotic dynamics are proposed to make a good balance between global exploration and local exploitation. Simulation results based on a set of well-known optimization benchmark instances and the basal insulin infusion rate adjustment problem for T1DM demonstrate the effectiveness of the proposed HPSO. In silico tests on standard virtual subject via HPSO show that under nominal condition the blood glucose concentrations could be kept within a range of 80-150 mg/dL within less than 5 days; meanwhile, in case of random variations in meal timings within $$\pm $$ 60 min or meal amounts within $$\pm $$ 75 % deviation from the nominal values, the blood glucose concentrations could be kept within the safe regions. [ABSTRACT FROM AUTHOR]

Partial least squares (PLS) and canonical correlation analysis (CCA) are two most popular key performance indicators (KPI) monitoring algorithms, which have shortcomings in dealing with the KPI-related information leakage problem, the model identification problem, the component selection problem, and the process hypothesis problem. To overcome these shortcomings, this study proposes a new multivariate statistics-based process monitoring (MSPM) method called the orthonormal subspace analysis (OSA) method. OSA divides process data and KPI data into three orthonormal subspaces through an analytic solution. Hence, OSA not only can detect a fault but also can judge whether the fault is KPI-related or KPI-unrelated. OSA is always effective irrespective of the availability of the KPI variables in the online monitoring stage. In OSA, the cumulative percent variance (CPV) method is adopted for principal component selection. In brief, OSA is a more effective method than PLS and CCA, and it achieves superior performance in fault detection and classification. [ABSTRACT FROM AUTHOR]