In this paper, a decentralized fault diagnosis approach of complex processes is proposed based on multiblock kernel partial least squares (MBKPLS). To solve the problem posed by nonlinear characteristics, (KPLS) approaches have been proposed. MBKPLS algorithm first and applied to monitor large-scale processes. The advantages are: 1) can capture more useful information between within blocks compared (PLS); 2) gives interpretation MBPLS; 3) Fault becomes possible if number sub-blocks equal...

In this paper, a new nonlinear process monitoring method that is based on multiway kernel independent component analysis (MKICA) developed. Its basic idea to use MKICA extract some dominant components capture nonlinearity from normal operating data and combine them with statistical techniques. The proposed applied the fault detection in fermentation compared modified (MICA). Applications of approach indicate effectively captures relationship variables show superior detectability, MICA.

Abstract In this article, first, some drawbacks of original Kernel Principal Component Analysis (KPCA) and Independent (KICA) are analyzed. Then the KPCA KICA for multivariate statistical process monitoring (MSPM) improved. The as follows: data mapped into feature space become redundant; linear introduce errors while kernel trick is used; computation time increases with number samples. To solve above problems, MSPM improved: similarity factors observed in input defined; similar...

In this article, the nonlinear dynamic process monitoring method based on kernel independent component analysis (KICA) is developed. Compared to Support Vector Machine (SVM) method, KICA unsupervised and available for fault detection. Hence, in used detect faults. Because dimension of feature space far less than rank matrix, a basis selected. Specifically, first constructed similarity factor data one group article. A contribution plot impossible, because mapping function from input into...

In the iron flotation production process, stages often undergo updates due to equipment upgrades, changes in raw materials, and other reasons. The operating condition prediction model established based on data from previous may not meet requirements of new stage, resulting a significant waste collected datasets. Data-driven models using small samples during current stage lack accuracy limited sample size. This study proposes method knowledge transfer effectively leverage large amount...

This paper deal with the stabilization networked control system (NCS) quantization and data packed dropout, stability conditions of NCS dropout are established via a packet-loss dependent Lyapunov approach. A new model is proposed to describe both network state system. The corresponding stabilizing controller design techniques also given based upon conditions. For different initial errors, necessary sufficient for be globally exponentially stable or regions presented.

The electro-fused magnesia furnace (EFMF) has complex characteristics, such as strong nonlinearity and multimodes. In this paper, the between-mode process modeling monitoring method of EFMF is proposed. original methods, data are handled in a single mode matrices, influence from one to another tends be ignored. However, hidden effect could useful analysis control. New proposed for part establish an integrated system, which would simplify model structure enhance its robustness. manifold...

Multi‐mode process monitoring is a key issue often raised in industrial control. Most multivariate statistical strategies, such as principal component analysis (PCA) and partial least squares, make an essential assumption that the collected data follow unimodal or Gaussian distribution. However, owing to complexity multi‐mode feature of processes, usually different distributions. This paper proposes novel processing method called weighted k neighbourhood standardisation (WKNS) address...

In this paper, a new regression and reconstruction method for process monitoring is proposed. The main contributions of the proposed approaches are as follows: 1) nonlinear algorithm to extract output-relevant variation, which, compared with conventional algorithm, builds more direct relationship between input output variables; 2) fault direction determined by possible magnitude every principal component; 3) effectively diagnosed kernel partial least-squares (KPLS) method. applied continuous...