This book provides a comprehensive coverage of Machine Learning (ML) methods that have proven useful in process industry for dynamic process modeling. Step-by-step instructions, supported with industry-relevant case studies, show (using Python) how to develop solutions for process modeling, process monitoring, etc., using classical and modern methods.
This book is designed to help readers gain a working-level knowledge of machine learning-based dynamic process modeling techniques that have proven useful in process industry. Readers can leverage the concepts learned to build advanced solutions for process monitoring, soft sensing, inferential modeling, predictive maintenance, and process control for dynamic systems. The application-focused approach of the book is reader friendly and easily digestible to the practicing and aspiring process engineers, and data scientists. The authors of this book have drawn from their years of experience in developing data-driven industrial solutions to provide a guided tour along the wide range of available ML methods and declutter the world of machine learning for dynamic process modeling. Applications on time series analysis, process disturbance modeling, system identification, and process fault detection are illustrated with examples. Upon completion, readers will be able to confidently navigate the system identification literature and make judicious selection of modeling approaches suitable for their problems.
This book has been divided into three parts. Part 1 of the book provides perspectives on the importance of ML for dynamic process modeling and lays down the basic foundations of ML-DPM (machine learning for dynamic process modeling). Part 2 provides in-detail presentation of classical ML techniques (such as ARX, FIR, OE, ARMAX, ARIMAX, CVA, NARX, etc.) and has been written keeping in mind the different modeling requirements and process characteristics that determine a model’s suitability for a problem at hand. These include, amongst others, presence of multiple correlated outputs, process nonlinearity, need for low model bias, need to model disturbance signal accurately, etc. Part 3 is focused on artificial neural networks and deep learning.
The following topics are broadly covered:
· Exploratory analysis of dynamic dataset
· Best practices for dynamic modeling
· Linear and discrete-time classical parametric and non-parametric models
· State-space models for MIMO systems
· Nonlinear system identification and closed-loop identification
· Neural networks-based dynamic process modeling
Who should read this book:
The application-oriented approach in this book is meant to give a quick and comprehensive coverage of dynamic modeling methodologies in a coherent, reader-friendly, and easy-to-understand manner. The following categories of readers will find the book useful:
1) Data scientists new to the field of system identification
2) Regular users of commercial process modeling software looking to obtain a deeper understanding of the underlying concepts
3) Practicing process data scientists looking for guidance for developing process modeling and monitoring solutions for dynamic systems
4) Process engineers or process engineering students making their entry into the world of Data Science
Pre-requisites:
No prior experience with Machine Learning or Python is needed. Undergraduate-level knowledge of basic linear algebra and calculus is assumed.
Author(s): Ankur Kumar, Jesus Flores-Cerrillo
Series: ML for Process Industry Series
Publisher: Leanpub
Year: 2023
Language: English
Pages: 208
Part 1 Introduction and Fundamentals
• Chapter 1: Machine Learning and Dynamic Process Modeling 1
1.1. Process Systems Engineering, Dynamic Process Modeling, and Machine Learning
-- components of a dynamic process model
1.2. ML-DPM Workflow
1.3. Taxonomy of ML-based Dynamic Models
1.4. Applications of DPM in Process Industry
• Chapter 2: The Scripting Environment 15
2.1. Introduction to Python
2.2. Introduction to Spyder and Jupyter
2.3. Python Language: Basics
2.4. Scientific Computing Packages: Basics
-- Numpy
-- Pandas
2.5. SysID-relevant Python Libraries
2.6. Typical SysID Script
• Chapter 3: Exploratory Analysis and Visualization of Dynamic Dataset: Graphical Tools 35
3.1. Visual Plots: Simple Yet Powerful Tools
3.2. Autocorrelation Function (ACF)
-- Examples on inferences drawn using ACF
3.3. Partial Autocorrelation Function (PACF)
3.4. Cross-correlation Function (CCF)
-- Examples on inferences drawn using CCF
3.5. Power Spectral Density (PSD) and Periodogram
• Chapter 4: Machine Learning-Based Dynamic Modeling: Workflow and Best Practices 48
4.1. System Identification Workflow
4.2. Identification Test/Input Signal Design
-- Pseudo Random Binary Sequence (PRBS)
-- Generalized Binary Noise (GBN)
`
4.3. Data Pre-processing
-- (Measurement) Noise removal
-- Centering and scaling
-- Trend and drift removal
4.4. Model Structure Selection
4.5. Model ID and Performance Assessment
-- Model order selection
4.6. Model Quality Check and Diagnostics
-- Residual analysis
-- Transient response checks
-- Simulation response checks
-- Parameter error checks
• Part 2 Classical Machine Learning Methods for Dynamic Modeling
• Chapter 5: Time Series Analysis: Concepts and Applications 76
5.1. Time Series Analysis: An Introduction
5.2. Autoregressive (AR) Models: An Introduction
-- Model order selection
5.3. Moving Average (MA) Models: An Introduction
-- Model order selection
5.4. Autoregressive Moving Average (ARMA) Models: An Introduction
-- Model order selection
5.5. Monitoring Controlled Variables in a CSTR using ARMA Models
5.6. Autoregressive Integrated Moving Average (ARIMA) Models: An Introduction
-- Model order selection
5.7. Forecasting Signals using ARIMA
• Chapter 6: Input-Output Modeling – Part 1: Simple Yet Popular Classical Linear Models 100
6.1. FIR Models: An Introduction
6.2. FIR Modeling of Industrial Furnaces
6.3. ARX Models: An Introduction
6.4. FIR Modeling of Industrial Furnaces
6.5. FIR and ARX MIMO Models
• Chapter 7: Input-Output Modeling – Part 2: Handling Process Noise the Right Way 120
7.1. PEM Models
7.2. ARMAX Models: An Introduction
7.3. ARMAX Modeling of Distillation Columns
`
7.4. OE Models: An Introduction
7.5. Box-Jenkins Models: An Introduction
7.6. ARIMAX Models: An Introduction
• Chapter 8: State-Space Models: Efficient Modeling of MIMO Systems 138
8.1. State-Space Models: An Introduction
8.2. State-Space Modeling via CVA
-- Mathematical background
-- Hyperparameter selection
8.3. Modeling Glass Furnaces via CVA
8.4. Monitoring Industrial Chemical Plants using CVA
-- Process monitoring/fault detection indices
• Chapter 9: Nonlinear System Identification: Going Beyond Linear Models 158
9.1. Nonlinear System Identification: An Introduction
9.2. NARX Models: An Introduction
9.3. Nonlinear Identification of a Heat Exchanger Process Using NARX
9.4. Introduction to Block-Structured Nonlinear Models
• Addendum A1: Closed-Loop Identification: Modeling Processes Under Feedback 170
• Part 3 Artificial Neural Networks & Deep Learning
• Chapter 10: Artificial Neural Networks: Handling Complex Nonlinear Systems 174
10.1. ANN: An Introduction
10.2. Modeling Heat Exchangers using FFNN-NARX
10.3. RNN: An Introduction
-- RNN outputs
-- RNN-based NARX topology
-- LSTM Networks
10.4. Modeling Heat Exchangers using LSTM
