Data Science for Batch Processes

José M. González-Martínez is Manager of the Department of Chemometrics and Digital Chemistry at Shell Global Solutions International B.V., Amsterdam. José Camacho is a Full Professor at the Department of Signal Theory, Telematics and Communication and leader of the Computational Data Science Laboratory (CoDaS Lab) at the University of Granada, Spain. Joan Borràs-Ferrís is a researcher and specialist in chemical engineering, applied statistics, and process modeling in digitalized industrial environments. Alberto Ferrer is a Full Professor of Statistics at the Universitat Politècnica de València (Spain).

• Foreword Prologue: Challenges for the Third MillenniumAbout the Companion Website1 Introduction1.1 Industrial Batch Processes1.2 Types of Sensors1.3 Batch Process Modeling1.3.1 Knowledge-based Models1.3.2 Data-driven Models1.3.3 Hybrid Models1.4 Bilinear Modeling Cycle for Batch Process Monitoring2 Data-driven Models Based on Latent Variables2.1 Compression2.2 Principal Component Analysis2.2.1 Data Preprocessing2.2.2 Selection of the Number of Principal Components2.2.3 Parameters Stability2.3 Regression2.4 Regression Models based on Latent Variables2.4.1 Principal Component Regression2.4.2 Partial Least Squares2.4.3 Data Preprocessing2.4.4 Selection of the Number of Latent Variables2.4.5 PLS Versus Other Regression Models2.5 Multivariate Exploratory Data Analysis2.6 Missing Data2.6.1 Model Exploitation2.6.2 Model Building2.6.3 Final Reflections about Missing Data Imputation and MSPC3 Batch Data Equalization 3.1 Introduction3.2 Challenges in Batch Equalization3.3 Equalization of Variables within a Batch3.3.1 Discarding Intermediate Values3.3.2 Estimating Missing Values3.3.2.1 Comparison of Equalization Methods Based on Latent Variable Models3.3.3 Rearranging Data3.4 Multirate System4 Batch Synchronization4.1 Introduction4.2 Synchronization Approaches4.2.1 Indicator Variable4.2.2 Time Linear Expanding/Compressing4.2.2.1 Observation (OWU) Level and TLEC Synchronization Approach4.2.3 Dynamic Time Warping4.2.3.1 Warping Function Constraints4.2.3.2 The DTW Algorithm4.2.3.3 Optimization Problem4.2.3.4 End-of-batch DTW Synchronization for Batch Process Monitoring4.2.3.5 On the Use of Warping Information4.2.4 Relaxed Greedy Time Warping4.2.4.1 Enhanced Global Constraints4.2.4.2 Cross-validation for the Estimation of the RGTW Parameters4.2.5 Multisynchro4.2.5.1 Asynchronism Detection4.2.5.2 Specific Batch Synchronization4.2.5.3 Iterative Batch Synchronization and Anomaly Detection Procedure4.3 Effects of Synchronization on the Correlation Structure5 Batch Data Preprocessing 5.1 Batch Preprocessing Operations5.2 Mean Centering5.3 Scaling6 Three-way to Two-way Transformation6.1 Introduction6.2 Single-model Approach6.2.1 Batch-wise Unfolding6.2.2 Variable-wise Unfolding6.2.3 Batch Dynamic Unfolding6.3 K-models Approach6.3.1 Hierarchical-model Approach6.4 Multiphase Approach6.4.1 Phases in Batch-wise Data6.4.2 Phases in Variable-wise Data6.4.3 Phases in Batch Dynamic Data6.5 Conclusion7 Batch Process Data Analysis and Statistical Monitoring 7.1 Introduction7.2 Historical Batch Data Analysis7.3 Batch Multivariate Statistical Process Control7.3.1 Phase I7.3.2 Phase II7.3.2.1 Post-batch Process Monitoring7.3.2.2 Real-time Process Monitoring7.4 Practical IssuesList of AcronymsBibliographyIndex