Guide to Industrial Analytics: Solving Data Science Problems for Manufacturing and the Internet of Things

Description

This textbook describes the hands-on application of data science techniques to solve problems in manufacturing and the Industrial Internet of Things (IIoT). Monitoring and managing operational performance is a crucial activity for industrial and business organisations. The emergence of low-cost, accessible computing and storage, through Industrial Digital of Technologies (IDT) and Industry 4.0, has generated considerable interest in innovative approaches to doing more with data. Data science, predictive analytics, machine learning, artificial intelligence and general approaches to modelling, simulating and visualising industrial systems have often been considered topics only for research labs and academic departments. This textbook debunks the mystique around applied data science and shows readers, using tutorial-style explanations and real-life case studies, how practitioners can develop their own understanding of performance to achieve tangible business improvements. All exercises can be completed with commonly available tools, many of which are free to install and use. Readers will learn how to use tools to investigate, diagnose, propose and implement analytics solutions that will provide explainable results to deliver digital transformation. Foreword Preface Overview and Goals Target Audiences Organisation and Suggested Use Learning Activities Hands-on Exercises Acknowledgements Contents Contributors Abbreviations blackPart I Introductory Concepts-1pt 1 An Introduction to Industrial Analytics 1.1 What Is Analytics? 1.2 Breaking Boundaries 1.2.1 The Industrial Internet of Things 1.2.2 Disruption Means Change 1.3 Industry 4.0 1.4 Opportunities for Smart Businesses 1.5 What Is Data Science? 1.6 Why Do We Need Data Science? 1.7 A Process for Data Science 1.7.1 Data Preparation 1.7.2 Data Exploration 1.7.3 Model Selection 1.7.4 Evaluation 1.8 Do We Need Machine Learning for Industrial Analytics? 1.9 Learning Activities 2 Data, Analysis and Statistics 2.1 Introduction 2.2 The Need for Analysis and Statistics 2.3 Qualitative and Quantitative Data 2.4 Data Terminology 2.5 Data Quality 2.6 Scales of Measurement 2.6.1 Nominal Data 2.6.2 Ordinal Data 2.6.3 Interval Data 2.6.4 Ratio Data 2.7 Central Tendency 2.7.1 Mean 2.7.2 Median 2.7.3 Mode 2.8 Dispersion 2.8.1 Range 2.8.2 Interquartile Range 2.8.3 Variance 2.8.4 Standard Deviation 2.8.5 Frequency 2.9 Histogram 2.9.1 Cumulative Frequency Graph 2.10 Shape of the Data 2.10.1 Normal Distribution 2.10.2 Uniform Distribution 2.10.3 Bimodal Distribution 2.10.4 Skewed Distributions 2.11 Visualising Data 2.11.1 Pie Charts 2.11.2 Bar Charts 2.11.3 Line Charts 2.11.4 Scatter Plots 2.12 Learning Activities 3 Measuring Operations 3.1 Introduction 3.2 Using Assumptions 3.3 Operations Concepts 3.3.1 Cycle Time 3.3.2 Lead Time 3.3.3 Takt Time 3.4 Using Concepts to Understand Systems 3.5 Resource Utilisation 3.6 Learning Activities 4 Data for Production Planning and Control 4.1 Historical Attitudes Towards the Use of Data 4.2 Need for Data Within the Production Area 4.3 Planning Problems Resulting from Lack of Appropriate Data 4.4 Planning with Appropriate Data 4.5 Need for Optimality in Production Control and Scheduling 4.6 Deriving Generic Models for Planning and Control 4.7 Production Planning in Manufacturing: Small Case Study Results 4.8 Planning and Control in the Case Study Firms 4.9 Manufacturing Production Systems in Case Study Firms 4.10 Summary 4.11 Learning Activities blackPart II Methods-1pt 5 Simulating Industrial Processes 5.1 Understanding Business Operations 5.2 Queues and Queueing 5.3 Modelling an Industrial Process 5.4 Designing a Process Simulation 5.5 Building the Simulation in Ciw 5.6 Confidence 5.7 Conclusion 5.8 Learning Activities 6 From Process to System Simulation 6.1 Simulating Industrial Systems 6.2 Example: Joinery Manufacturer 6.3 Building the Simulation 6.4 Managing Resource Utilisation 6.5 Product Mixes 6.5.1 Sash Windows 6.6 Conclusion 6.7 Learning Activities 7 Constructing Machine Learning Models for Prediction 7.1 Introduction 7.2 Data and Prediction 7.2.1 Example 1: Job Time Prediction Under Varying Demand 7.3 Assessing the Predictive Power of a Model 7.3.1 Root Mean Squared Error (RMSE) 7.3.2 Mean Absolute Error (MAE) 7.3.3 Mean Absolute Percentage Error (MAPE) 7.3.4 Coefficient of Determination (R2) 7.3.5 Underfitting and Overfitting 7.3.6 Cross-Validation 7.3.7 Learning Curves 7.3.8 Validation Curves 7.4 How to Improve Model Accuracy 7.4.1 Feature Selection 7.4.2 Example 2: Improving the Model with Additional Information (Multiple Regression) 7.4.3 More Data 7.4.4 Compare Models 7.4.5 Example 3: Multiple Job Types (Model Comparison) 7.5 Generating Data Via Simulations 7.5.1 Example 4: Simulating Data Under Uncertainty 7.5.2 Kernel Density Estimation and Sampling 7.6 Worked Examples in R 7.6.1 Linear Regression Model 7.6.2 Multiple Regression Model 7.6.3 Cross-Validation 7.6.4 KDE Estimation of Distribution blackPart III Application-1pt 8 Case Study: Confectionery Production 8.1 Introduction 8.1.1 Company Organisation 8.1.2 Production 8.2 Hard Boiled Confectionery Organisation and Planning 8.2.1 Unit Operation 8.2.2 Scheduling 8.2.3 Model to Determine Optimal Long Term, Monthly Production Plans 8.2.4 Implementation of Monthly Planning 8.2.5 Allocating Production Pairs 8.2.6 Implementation of Daily Pair Selection 8.3 Impact of Lack of Information on Company Profits 8.4 Conclusion 8.5 Learning Activities 8.5.1 Machinery and Staffing 8.5.2 Sales Data 9 Minimum Information Set for Effective Control 9.1 Information Flows Within an Organisation 9.2 Deriving Minimum Information Requirements 9.3 Order Book-Based Systems 9.3.1 Deriving an Order Book (OB) Forecasting Model Where Orders for New Jobs Arrive Randomly in Time 9.3.2 Variability in the Final Product Introduced at All Stages 9.3.3 Simple Order Book System 9.3.4 Enhanced Order Book Models 9.3.5 Forecast Accuracy/Validation 9.3.6 Extending the Investigation by Including Data from All Stages 9.3.7 Variability Added (only) at the Final Stage 9.3.8 Conclusion: Order Book (OB)-Based Approaches to Forecasting 9.4 Work Book (WB) Systems 9.5 Evaluating WB and OB When Stage Productions Have Been Balanced 9.5.1 Conclusions and Recommendations MIR 9.6 Data Requirements for Planning and Control 9.7 Minimal Information in Flow Shops with CONWIP Control 9.7.1 Flow Shops with More Than One Like Processor at Each Stage 9.7.2 Job Shops 9.7.3 Effect of Growth on Planning and Control in a Flow Shop 9.7.4 Results from Three-Stage Models 9.7.5 Results from 10-Stage Models 9.8 Conclusion 9.9 Learning Activities 10 Business Adoption of Analytics 10.1 Introduction 10.1.1 Intelligent Manufacturing 10.1.2 Compounded Challenges for SMEs 10.1.3 Regional Challenge 10.2 A Model of Engagement 10.2.1 Proving the Return on Investment 10.2.2 Digital Enablers Network (DEN) 10.3 University Capability 10.3.1 Case Study: DEN in Action 10.4 Discussion 10.4.1 Benefits to SMEs 10.4.2 Benefits to den Members 10.4.3 Benefits to Academia 10.4.4 Human Factors 10.5 Conclusions 10.6 Future Work A Statistics A.1 Basic Descriptive Statistics A.1.1 Measures of Location A.1.2 Measures of Spread A.2 Statistical Distributions A.2.1 Describing Job Arrivals A.2.2 Describing Job Times A.2.3 Adding Distributions A.2.4 Regression Methods A.2.5 Common Failure Cases in Regression A.2.6 Modelling Simulation Data B Simulation Library—Ciw B.1 About Ciw B.2 Installing Ciw C Production Planning Programs in MS Excel C.1 Program 1: Checking Order Delivery Date C.1.1 Set Up Production Stage Data C.1.2 Set Up Stage Characteristics/Capacities C.1.3 Updated Working Sheets C.1.4 Input Date and Know (Feasible Order) C.1.5 Using the Program to Check Capacity C.1.6 Status Plots C.2 Program 2: Allocating Production Time C.2.1 Plan C.2.2 Loading Plan C.3 Assigning Machines to Jobs D Scheduling Jobs Through a Workshop D.1 Scheduling n Jobs Through a Workshop Containing m Machines/Processors/Servers D.1.1 Problem Size D.2 Mathematical Programming Techniques D.2.1 Fixing the Schedule Using an Excel Program D.2.2 Data Input: Defining Job Times D.2.3 Defining Optimality D.2.4 Minimise Lateness D.2.5 Job Shop Scheduling D.3 Rescheduling Jobs Index