Fault Detection and Diagnosis In Hall–Héroult Cells Based on Individual Anode Current Measurements Using Dynamic Kernel PCA

To solve the problem of low accuracy in traditional fault diagnosis methods, a novel method of combining generalized frequency response function(GFRF) and convolutional neural network(CNN) is proposed. In order to accurately characterize system state information, this paper proposed a variable step size least mean square (VSSLMS) adaptive algorithm to calculate the second-order GFRF spectrum values under normal and fault states; In order to improve the ability of fault feature extraction, a convolution neural network (CNN) with gradient descent learning rate and alternate convolution layer and pooling layer is designed to extract the fault features from GFRF spectrum. In the proposed method, the second-order GFRF spectrum of each state of Permanent Magnet Synchronous Motor (PMSM) is obtained by VSSLMS; Then, the two-dimension GFRF spectrum, which is regarded as the gray value of the image,will be further transformed into image. Finally, the CNN is trained with learning rate by gradient descent way to realize the fault diagnosis of PMSM. Experimental results indicate that the accuracy of proposed method is 98.75%, which verifies the reliability of the proposed method in application of PMSM fault diagnosis.

show abstract

“…[32], KPCA in Ref. [33], t-SNE in Refs. [34]- [35] are the main methods of feature extraction, and SVM in Ref.…”

Section: Comparisons With Traditional Intelligent Diagnosis Methodsmentioning

confidence: 99%

A novel method of combining generalized frequency response function and convolutional neural network for complex system fault diagnosis

2020

View full text Add to dashboard Cite

show abstract

“…To a certain extent, it overcomes the problem of the population falling into the local optimum and improves the convergence speed of the algorithm. The crossover and mutation probability are shown in Formulas (11) and (12).…”

Section: Crossover and Mutationmentioning

confidence: 99%

“…By contrast, developing a soft sensor based on DDMs only requires obtaining the historical operational data without considering the complex mechanism or process knowledge, and this has been successfully applied to many spheres, which has attracted more and more attention from academia and industry. Classical data-driven modeling methods include principle component analysis (PCA) [ 11 , 12 , 13 , 14 , 15 , 16 ], support vector machine (SVM) [ 17 , 18 , 19 , 20 ], partial least squares (PLS) [ 21 , 22 ], Gaussian process regression (GPR) [ 23 , 24 ], Bayesian prediction [ 25 , 26 ], slow feature analysis (SFA) [ 27 , 28 , 29 , 30 ], extreme learning machine (ELM) [ 31 ] and their improved models, artificial neural networks (ANNs) [ 32 ] and two or more hybrid models [ 33 , 34 , 35 ], among others. The advantage of PCA is that it is convenient for simplifying the model and is generally used for the correlation analysis between the same matrix vectors.…”

Section: Introductionmentioning

confidence: 99%

A Soft Sensor Approach Based on an Echo State Network Optimized by Improved Genetic Algorithm

Huang

Cao³

2020

Sensors

View full text Add to dashboard Cite

In the process of fault diagnosis and the health and safety operation evaluation of modern industrial processes, it is crucial to measure important state variables, which cannot be directly detected due to limitations of economy, technology, environment and space. Therefore, this paper proposes a data-driven soft sensor approach based on an echo state network (ESN) optimized by an improved genetic algorithm (IGA). Firstly, with an ESN, a data-driven model (DDM) between secondary variables and dominant variables is established. Secondly, in order to improve the prediction performance, the IGA is utilized to optimize the parameters of the ESN. Then, the immigration strategy is introduced and the crossover and mutation operators are changed adaptively to improve the convergence speed of the algorithm and address the problem that the algorithm falls into the local optimum. Finally, a soft sensor model of an ESN optimized by an IGA is established (IGA-ESN), and the advantages and performance of the proposed method are verified by estimating the alumina concentration in an aluminum reduction cell. The experimental results illustrated that the proposed method is efficient, and the error was significantly reduced compared with the traditional algorithm.

show abstract

“…Investigating correlations between scale growth and other parameters have been a challenging task for several reasons. Process data from aluminium production is known to be temporally correlated [19], and complex codependencies probably exist between the variables, complicating the analysis. Furthermore, the dataset investigated in this study was relatively small, representing a few months of production only.…”

Section: Multivariate Data Analysis To Detect Variables Correlated Tomentioning

confidence: 99%

Alumina Scale Composition and Growth Rate in Distribution Pipes

et al. 2019

View full text Add to dashboard Cite

Scaling in distribution pipes for secondary alumina is a major issue in aluminium smelters. The scale is formed inside the distribution pipes, and gradually reduces the cross section that is available for alumina transport. The scale cannot be removed without dismantling the transport pipes and using chemical and/or mechanical treatments. This leads to interruptions in normal operation, requires additional labour, and a stock of additional pipe sections. To get a better understanding of the scale formation mechanisms, the rate of scale growth was monitored by acoustic measurements in a transport pipe section at an aluminium producing plant over a period of several months. Correlation between growth rate, and recorded parameters from the associated pots, fume treatment system and meteorological data were studied. In addition, samples from the actual scale were examined by SEM, TEM and AFM to investigate the microstructure and chemical composition.

show abstract

Fault Detection and Diagnosis In Hall–Héroult Cells Based on Individual Anode Current Measurements Using Dynamic Kernel PCA

Cited by 22 publications

References 26 publications

A novel method of combining generalized frequency response function and convolutional neural network for complex system fault diagnosis

A novel method of combining generalized frequency response function and convolutional neural network for complex system fault diagnosis

A Soft Sensor Approach Based on an Echo State Network Optimized by Improved Genetic Algorithm

Alumina Scale Composition and Growth Rate in Distribution Pipes

Contact Info

Product

Resources

About