Lidia Auret

Unsupervised Process Monitoring and Fault Diagnosis with Machine Learning Methods

This unique text/reference describes in detail the latest advances in unsupervised process monitoring and fault diagnosis with machine learning methods. Abundant case studies throughout the text demonstrate the efficacy of each method in real-world settings. The broad coverage examines such cutting-edge topics as the use of information theory to enhance unsupervised learning in tree-based methods, the extension of kernel methods to multiple kernel learning for feature extraction from data, and the incremental training of multilayer perceptrons to construct deep architectures for enhanced data projections. Topics and features: discusses machine learning frameworks based on artificial neural networks, statistical learning theory and kernel-based methods, and tree-based methods; examines the application of machine learning to steady state and dynamic operations, with a focus on unsupervised learning; describes the use of spectral methods in process fault diagnosis.

Data-driven fault detection with process topology for fault identification

Abstract In this paper a fault diagnosis framework based on detection with feature extraction methods and identification based on data-driven process topology methods was investigated. A simulation of a simple system consisting of two tanks with heat exchangers was used to generate data for normal operating conditions and a number of faults. Fault detection methods included principal component analysis and kernel principal component analysis feature extraction with Shewhart, cumulative sum and exponentially weighted moving average monitoring charts. Process topology information was extracted with linear cross-correlation, partial cross-correlation and transfer entropy. Connectivity maps were constructed to identify possible fault propagation paths to aid root cause analysis and changes in connectivity structure due to faults were exploited for fault identification. Kernel principal component analysis with a CUSUM chart gave the best detection performance, while connectivity graphs based on partial correlation gave an accurate representation of the system and assisted fault identification.

Early Detection of risk of autism spectrum disorder based on recurrence quantification analysis of electroencephalographic signals

Early detection of autism spectrum disorder (ASD) in infants is vital in maximizing the impact and potential long-term outcomes of early delivery of rehabilitative therapies. To date no definitive diagnostic test for ASD exists. Electroencephalography is a noninvasive method used to capture underlying electrical changes in brain activity. This proof-of-concept study suggests that recurrence quantification analysis features computed from resting state spontaneous eyes-closed electroencephalographic (EEG) signals may be useful biomarkers for early detection of risk of ASD.

Fault detection and diagnosis with random forest feature extraction and variable importance methods

The ever-present drive to safer, more cost-effective and cleaner processes motivates the exploration of a variety of process monitoring methods. In the domain of data-driven approaches, random forest models present a nonlinear framework. Random forest models consist of ensembles of classification and regression trees in which the model response is determined by voting committees of independent binary decision trees. Data-driven approaches to fault diagnosis often involve summarizing potentially large numbers of process variables in lower dimensional diagnostic sequences. Random forest feature extraction allows for the monitoring of process in feature and residual spaces, while random forest variable importance measures can potentially be used to identify process variables contribution to fault conditions. In this study, a framework for diagnosing steady state faults with random forests is proposed and demonstrated with a simple nonlinear system and the benchmark Tennessee Eastman process.

Statistical Learning Theory and Kernel-Based Methods

The basics of kernel methods and their position in the generalized data-driven fault diagnostic framework are reviewed. The review starts out with statistical learning theory, covering concepts such as loss functions, overfitting and structural and empirical risk minimization. This is followed by linear margin classifiers, kernels and support vector machines. Transductive support vector machines are discussed and illustrated by way of an example related to multivariate image analysis of coal particles on conveyor belts. Finally, unsupervised kernel methods, such as kernel principal component analysis, are considered in detail, analogous to the application of linear principal component analysis in multivariate statistical process control. Fault diagnosis in a simulated nonlinear system by the use of kernel principal component analysis is included as an example to illustrate the concepts.

Monitoring of Mineral Processing Operations based on Multivariate Similarity Indices

Abstract Multivariate process monitoring through covariance control charts considers changes in the relationships among process variables, but is limited by linearity assumptions. In this paper two nonlinear indicators of multivariate structure are considered, viz. mutual information and random forest proximity measures. Similarity matrices are constructed from data encapsulated by sliding windows of different sizes across the time series data associated with process operations. Diagnostic metrics reflect the differences between stationary base windows representative of normal operating conditions and test windows containing new process data. A case study in mineral processing shows that better results can be obtained with these nonlinear methods.

Artificial Neural Networks

The history, development and basic methodology of the most important artificial neural networks are reviewed, focusing on unsupervised learning, in keeping with the generalized framework for process fault diagnosis proposed previously. This starts with a brief summary of the construction of multilayer perceptrons, with some examples of their strengths and limitations, as well as their application in the analysis of models. Various autoassociative neural networks are discussed as an extension of multilayer perceptrons to nonlinear principal component analysis, including simple, hierarchical, inverse and circular nonlinear principal component analysis. This is followed by an introduction to radial basis function neural networks and self-organizing feature maps with basic examples of their applications. In the last parts of the chapter, restricted Boltzmann machines and deep neural network architectures are considered, as well as the basic operation of extreme learning machines. These are relatively new additions to the family of neural networks that have not become established in process fault diagnosis as yet but have the potential to make a significant contribution in the near future.

A comparison of control techniques for dairy falling film evaporators

Abstract Falling film evaporators (FFE) are widely used in the dairy industry to pre-concentrate milk for powder production. FFE control is, however, not performed well, with many plants still under operator or proportional and integral (PI) control. Several authors have created fundamental models to use for controller development, yet these models have various differences in structure and span feed flow rates ranging from laboratory scale (2 500kg/h) to industrial scale (27 000kg/h). This paper used a single semi-empirical model developed by Haasbroek (2013) to offer a sensible comparison of the most often seen dairy FFE controllers. Disturbance rejection was tested by introducing a feed dry mass fraction (W F ) step and then comparing the product dry mass fraction (W P ) increase as a percentage (Δ W P / Δ W F x100). It was found, as shown in figure 7, that linear quadratic (LQR) control (Haasbroek et al., 2013) and fuzzy predictive controllers showed the best performance (70% and 69% respectively), followed by cascade control (77%) and lastly PI control (123%). The fuzzy controller does, however, struggle with disturbances it has not been tuned for, while cascade and LQR controllers still perform well, as seen in figure 8. Taking into account the involved design required for LQR control, cascade control offers a well balanced approach to FFE disturbance rejection.

Performance Audit of a Semi-autogenous Grinding Mill Circuit

Abstract The paper proposes a novel performance audit report for a SAG Mill Circuit. The audit report is demonstrated on a validated run-of-mine ore grinding circuit model, which the authors have captured in a simulator, using Simulink. The elements of the report combine established statistics and views with specific mill control context. A representative case study demonstrates the construction and interpretation of the audit report, so as to gain insight into the circuit performance over a historical data episode.

The Application of Classification Methods to the Gross Error Detection Problem

Abstract All process measurements are corrupted by the presence of measurement error to some degree. The attenuation of the measurement error, especially large gross errors, can increase the value of the process measurements. Gross error detection has typically been performed through rigorous statistical hypothesis testing. The assumptions required to derive the necessary statistical properties are restrictive, which lead to investigation of alternative approaches, such as artificial neural networks. This paper reports the results of an investigation into the utility of classification trees and linear and quadratic classification functions for resolving the gross error detection and identification problems.