智能用电大数据分析：用户行为建模、聚合与预测（英文）

　　This book aims to make the best use of fine-grained smart meter data to process and translate them into actual information and incorporated into consumer behavior modeling and distribution system operations. It begins with an overview of recent developments in smart meter data analytics. Since data management is the basis of further smart meter data analytics and its applications， three issues on data management， i.e.， data compression， anomaly detection， and data generation， are subsequently studied.The following works try to model complex consumer behavior. Specific works include load profiling， pattern recognition， personalized price design， socio-demographic information identification， and household behavior coding. On this basis， the book extends consumer behavior in spatial and temporal scale. Works such as consumer aggregation， individual load forecasting， and aggregated load forecasting are introduced. We hope this book can inspire readers to define new problems， apply novel methods， and obtain interesting results with massive smart meter data or even other monitoring data in the power systems.

暂缺《智能用电大数据分析：用户行为建模、聚合与预测（英文）》作者简介

Contents
1 Overview of Smart Meter Data Analytics 1
1.1 Introduction 1
1.2 Load Analysis 4
1.2.1 Bad Data Detection 5
1.2.2 Energy Theft Detection 6
1.2.3 Load Profiling 8
1.2.4 Remarks 9
1.3 Load Forecasting 11
1.3.1 Forecasting Without Smart Meter Data 11
1.3.2 Forecasting with Smart Meter Data 14
1.3.3 Probabilistic Forecasting 16
1.3.4 Remarks 18
1.4 Load Management 19
1.4.1 Consumer Characterization 19
1.4.2 Demand Response Program Marketing 21
1.4.3 Demand Response Implementation 22
1.4.4 Remarks 23
1.5 Miscellanies 25
1.5.1 Connection Verification 25
1.5.2 Outage Management 26
1.5.3 Data Compression 26
1.5.4 Data Privacy 27
1.6 Conclusions 28
References 28
2 Electricity Consumer Behavior Model 37
2.1 Introduction 37
2.2 Basic Concept of ECBM 39
2.2.1 Definition 39
2.2.2 Connotation 41
2.2.3 Denotation 42
2.2.4 Relationship with Other Models 43
2.3 Basic Characteristics of Electricity Consumer Behavior 45
2.4 Mathematical Expression of ECBM 47
2.5 Research Paradigm of ECBM 50
2.6 Research Framework of ECBM 51
2.7 Conclusions 57
References 57
3 Smart Meter Data Compression 59
3.1 Introduction 59
3.2 Household Load Profile Characteristics 61
3.2.1 Small Consecutive Value Difference 61
3.2.2 Generalized Extreme Value Distribution 62
3.2.3 Effects on Load Data Compression 64
3.3 Feature-Based Load Data Compression 66
3.3.1 Distribution Fit 66
3.3.2 Load State Identification 67
3.3.3 Base State Discretization 67
3.3.4 Event Detection 68
3.3.5 Event Clustering 69
3.3.6 Load Data Compression and Reconstruction 69
3.4 Data Compression Performance Evaluation 71
3.4.1 Related Data Formats 71
3.4.2 Evaluation Index 72
3.4.3 Dataset 72
3.4.4 Compression Efficiency Evaluation Results 73
3.4.5 Reconstruction Precision Evaluation Results 74
3.4.6 Performance Map 74
3.5 Conclusions 77
References 77
4 Electricity Theft Detection 79
4.1 Introduction 79
4.2 Problem Statement 81
4.2.1 Observer Meters 81
4.2.2 False Data Injection 81
4.2.3 A State-Based Method of Correlation 83
4.3 Methodology and Detection Framework 83
4.3.1 Maximum Information Coefficient 84
4.3.2 CFSFDP-Based Unsupervised Detection 85
4.3.3 Combined Detecting Framework 86
4.4 Numerical Experiments 88
4.4.1 Dataset 88
4.4.2 Comparisons and Evaluation Criteria 89
4.4.3 Numerical Results 90
4.4.4 Sensitivity Analysis 93
4.5 Conclusions 97
References 97
5 Residential Load Data Generation 99
5.1 Introduction 99
5.2 Model 101
5.2.1 Basic Framework 101
5.2.2 General Network Architecture 102
5.2.3 Unclassified Generative Models 106
5.2.4 Classified Generative Models 110
5.3 Methodology 113
5.3.1 Data Preprocessing 114
5.3.2 Model Training 115
5.3.3 Metrics 118
5.4 Case Studies 122
5.4.1 Data Description 122
5.4.2 Unclassified Generation 123
5.4.3 Classified Generation 125
5.5 Conclusion 134
References 134
6 Partial Usage Pattern Extraction 137
6.1 Introduction 137
6.2 Non-negative K-SVD-Based Sparse Coding 139
6.2.1 The Idea of Sparse Representation 139
6.2.2 The Non-negative K-SVD Algorithm 140
6.3 Load Profile Classification 141
6.3.1 The Linear SVM 141
6.3.2 Parameter Selection 142
6.4 Evaluation Criteria and Comparisons 143
6.4.1 Data Compression-Based Criteria 143
6.4.2 Classification-Based Criteria 144
6.4.3 Comparisons 145
6.5 Numerical Experiments 146
6.5.1 Description of the Dataset 146
6.5.2 Experimental Results 147
6.5.3 Comparative Analysis 152
6.6 Further Multi-dimensional Analysis 154
6.6.1 Characteristics of Residential & SME Users 154
6.6.2 Seasonal and Weekly Behaviors Analysis 156
6.6.3 Working Day and Off Day Patterns Analysis 158
6.6.4 Entropy Analysis 159
6.6.5 Distribution Analysis 160
6.7 Conclusions 161
References 161
7 Personalized Retail Price Design 163
7.1 Introduction 163
7.2 Problem Formulation 165
7.2.1 Problem Statement 165
7.2.2 Consumer Problem 166
7.2.3 Compatible Incentive Design 166
7.2.4 Retailer Problem 167
7.2.5 Data-Driven Clustering and Preference
Discovering 168
7.2.6 Integrated Model 171
7.3 Solution Methods 172
7.3.1 Framework 172
7.3.2 Piece-Wise Linear Approximation 172
7.3.3 Eliminating Binary Variable Product 173
7.3.4 CVaR 173
7.3.5 Eliminating Absolute Values 174
7.4 Case Study 174
7.4.1 Data Description and Experiment Setup 174
7.4.2 Basic Results 175
7.4.3 Sensitivity Analysis 178
7.5 Conclusions and Future Works 183
Appendix I 183
Appendix II 184
References 185
8 Socio-demographic Information Identification 187
8.1 Introduction 187
8.2 Problem Definition 189
8.3 Method 190
8.3.1 Why Use a CNN? 190
8.3.2 Proposed Network Structure 191
8.3.3 Description of the Layers 19