DATA ANALYTICS AND DESIGN OF INDUSTRIAL EXPERIMENTS

OBJECTIVES

Part 1 – DATA ANALYSIS
1.Extracting information from process data: multivariate statistical techniques and data driven methodologies for exploratory analysis
Identification of the major driving forces of a process
Understanding the process parameters that are critical for the product quality
Case studies: monitoring the dimension of mechanical parts; process understanding in the production of paracetamol tablets; monitoring of an infection process for the production of vaccines; prediction of the product quality in the production of resins and coatings
2. Quality improvement
Data-based modeling of both process and product quality
Statistical process control, control charts, and process capability
Estimation of product quality
3. Identification of the main characteristic of a process from the data routinely collected in process historians and classification of product quality
Linear and nonlinear, supervised and unsupervised classification techniques;
Case studies: management of the historical knowledge in secondary manufacturing by mining of process database; classification of semiconductors from image analysis of the product surface characteristics
4. Process monitoring and fault detection and diagnosis
Case study: monitoring of the production of pharmaceutical products; monitoring of the manufacturing of fine chemicals

Part 2 – DESIGN OF EXPERIMENTS
1. Why carrying out experiments in an industrial/laboratory environment? How can experiments be planned in a smart and optimal manner? How to maximize the information obtained by experiments?
How samples should be collected? How should be organized an experimental campaign?
Comparison between experiments
Case studies: planning experimentation in the manufacturing sector and in the pharmaceutical industry
2. Conducting experiments with single process parameters and multiple process parameters.
Analyzing the variability of the experiments
Carrying out experiments
Case studies: planning experimentation in the manufacturing sector and in the pharmaceutical industry
3. Selecting the optimal conditions for running an experimental campaign and obtaining optimal combination of raw material properties and process settings to obtain a product of desired quality

METHODS

Part 1 – DATA ANALYSIS
1. Quality improvement and Statistical Process Control:
The “dimension” of quality
Management aspects of quality improvement (Deming cycle, Shewhart cycle, Quality systems and standards, DMAIC, Six-sigma, Lean manufacturing)
Quality, productivity and quality costs
Legal aspects of quality
2. Introduction to probability theory and statistical inference:
Sampling, sampling distributions and sample size
3. Control charts and process capability
SPC and control charts (for variables and attributes)
Process capability
Control limits, correlation and adaptive charts
Sampling and acceptance sampling
4. Multivariate statistical techniques and data driven methodologies for exploratory analysis
Principal Component Analysis PCA
Partial Least Squares PLS
5. Pattern recognition and classification techniques
Linear Discriminant Analysis and Quadratic Discriminant Analysis
PCA and PLS-DA
kNN and hierarchical custering
6. introduction to machine leatning and deep learning

Part 2 – DESIGN OF EXPERIMENTS
1. Experiments on single factors:
Comparison between experiments
2. Factorial design: full factorial and fractional factorial
2-factor factorial design and 2k factorial design
Fractional factorial design
Central composit design
3. Regression models and response curves
Model parameter estimation
Experimental design for fitting response surfaces