Data Science for Wind Energy provides an in-depth discussion on how data science methods can improve decision making for wind energy applications, near-ground wind field analysis and forecast, turbine power curve fitting and performance analysis, turbine reliability assessment, and maintenance optimization for wind turbines and wind farms. A broad set of data science methods covered, including time series models, spatio-temporal analysis, kernel regression, decision trees, kNN, splines, Bayesian inference, and importance sampling. More importantly, the data science methods are described in the context of wind energy applications, with specific wind energy examples and case studies. Features:
- Provides an integral treatment of data science methods and wind energy applications
- Includes specific demonstration of particular data science methods and their use in the context of addressing wind energy needs
- Presents real data, case studies and computer codes from wind energy research and industrial practice
- Covers material based on the author's ten plus years of academic research and insights
Author(s): Yu Ding
Publisher: Chapman And Hall/CRC
Year: 2020
Language: English
Pages: 425
Tags: Data Science, Wind Energy
Cover......Page 1
Half Title......Page 2
Title Page......Page 4
Copyright Page......Page 5
Dedication......Page 6
Table of Contents......Page 8
Foreword......Page 16
Preface......Page 18
Acknowledgments......Page 22
Chapter 1: Introduction......Page 26
1.1 WIND ENERGY BACKGROUND......Page 27
1.2 ORGANIZATION OF THIS BOOK......Page 31
1.2.1 Who Should Use This Book......Page 33
1.2.3 Datasets Used in the Book......Page 34
Part I: Wind Field Analysis......Page 40
Chapter 2: A Single Time Series Model......Page 42
2.1 TIME SCALE IN SHORT-
TERM FORECASTING......Page 43
2.2.2 Weibull Distribution......Page 44
2.2.3 Estimation of Parameters in Weibull Distribution......Page 45
2.2.4 Goodness of Fit......Page 46
2.2.5 Forecasting Based on Weibull Distribution......Page 48
2.3 DATA TRANSFORMATION AND STANDARDIZATION......Page 49
2.4 AUTOREGRESSIVE MOVING AVERAGE MODELS......Page 52
2.4.1 Parameter Estimation......Page 53
2.4.2 Decide Model Order......Page 54
2.4.3 Model Diagnostics......Page 56
2.4.4 Forecasting Based on ARMA Model......Page 59
2.5.1 Kalman Filter......Page 63
2.5.2 Support Vector Machine......Page 65
2.5.3 Artificial Neural Network
......Page 70
2.6 PERFORMANCE METRICS......Page 73
2.7 COMPARING WIND FORECASTING METHODS......Page 75
3.1 COVARIANCE FUNCTIONS AND KRIGING......Page 82
3.1.1 Properties of Covariance Functions......Page 83
3.1.2 Power Exponential Covariance Function......Page 84
3.1.3 Kriging......Page 85
3.2.1 Gaussian Spatio-temporal Autoregressive Model......Page 90
3.2.2 Informative Neighborhood......Page 93
3.2.3 Forecasting and Comparison......Page 94
3.3.1 Definition and Quantification
......Page 98
3.3.2 Asymmetry of Local Wind Field......Page 99
3.3.3 Asymmetry Quantification
......Page 101
3.3.4 Asymmetry and Wake Effect
......Page 103
3.4.1 Asymmetric Non-separable Spatio-temporal Model......Page 104
3.4.3 Forecasting Using Spatio-temporal Model......Page 106
3.5 CASE STUDY......Page 108
4.1 REGIME-SWITCHING AUTOREGRESSIVE MODEL
......Page 118
4.1.1 Physically Motivated Regime Definition
......Page 119
4.1.2 Data-driven Regime Determination......Page 121
4.1.3 Smooth Transition between Regimes......Page 122
4.1.4 Markov Switching between Regimes......Page 123
4.2 REGIME-SWITCHING SPACE-TIME MODEL
......Page 124
4.3 CALIBRATION IN REGIME SWITCHING METHOD......Page 129
4.3.1 Observed Regime Changes......Page 130
4.3.2 Unobserved Regime Changes......Page 131
4.3.3 Framework of Calibrated Regime-switching......Page 132
4.3.4 Implementation Procedure......Page 136
4.4.1 Modeling Choices and Practical Considerations......Page 138
4.4.2 Forecasting Results......Page 140
Part II: Wind Turbine Performance Analysis......Page 148
Chapter 5: Power Curve Modeling and Analysis......Page 150
5.1 IEC BINNING: SINGLE-DIMENSIONAL POWER CURVE
......Page 151
5.2 KERNEL-BASED MULTI-DIMENSIONAL POWER CURVE
......Page 152
5.2.1 Need for Nonparametric Modeling Approach......Page 153
5.2.2 Kernel Regression and Kernel Density Estimation......Page 156
5.2.3 Additive Multiplicative Kernel Model......Page 159
5.2.4 Bandwidth Selection......Page 161
5.3 OTHER DATA SCIENCE METHODS
......Page 162
5.3.1 k-Nearest Neighborhood Regression......Page 163
5.3.2 Tree-based Regression......Page 164
5.3.3 Spline-based Regression......Page 168
5.4.1 Model Parameter Estimation......Page 170
5.4.2 Important Environmental Factors Affecting Power Output
......Page 172
5.4.3 Estimation Accuracy of Different Models
......Page 175
6.1 THREE EFFICIENCY METRICS......Page 184
6.1.2 Power Generation Ratio......Page 185
6.1.3 Power Coefficient
......Page 186
6.2 COMPARISON OF EFFICIENCY METRICS......Page 187
6.2.1 Distributions......Page 189
6.2.2 Pairwise Differences
......Page 191
6.2.3 Correlations and Linear Relationships......Page 193
6.2.4 Overall Insight......Page 195
6.3 A SHAPE-CONSTRAINED POWER CURVE MODEL
......Page 196
6.3.1 Background of Production Economics......Page 197
6.3.2 Average Performance Curve......Page 199
6.3.3 Production Frontier Function and Effi
ciency Metric......Page 202
6.4 CASE STUDY......Page 204
7.1 PASSIVE DEVICE INSTALLATION UPGRADE......Page 212
7.2.1 Hierarchical Subgrouping......Page 214
7.2.2 One-to-One Matching......Page 217
7.2.3 Diagnostics......Page 218
7.2.4 Paired t-tests and Upgrade Quantification
......Page 219
7.2.5 Sensitivity Analysis......Page 221
7.3 POWER CURVE-BASED APPROACH
......Page 222
7.3.1 The Kernel Plus Method......Page 223
7.3.2 Kernel Plus Quantification Procedure
......Page 226
7.3.3 Upgrade Detection......Page 227
7.4 AN ACADEMIA-INDUSTRY CASE STUDY
......Page 229
7.4.1 The Power-vs-Power Method......Page 231
7.4.2 Joint Case Study......Page 233
7.4.3 Discussion......Page 236
7.5 COMPLEXITIES IN UPGRADE QUANTIFICATION......Page 238
8.1 CHARACTERISTICS OF WAKE EFFECT......Page 244
8.2 JENSEN’S MODEL......Page 245
8.3 A DATA BINNING APPROACH
......Page 247
8.4 SPLINE-BASED SINGLE-WAKE MODEL
......Page 248
8.4.2 Power Diff
erence Model for Two Turbines......Page 249
8.4.3 Spline Model with Non-negativity Constraint......Page 251
8.5 GAUSSIAN MARKOV RANDOM FIELD MODEL......Page 255
8.6.1 Performance Comparison of Wake Models......Page 257
8.6.2 Analysis of Turbine Wake Effect
......Page 260
Part III: Wind Turbine Reliability Management......Page 270
Chapter 9: Overview of Wind Turbine Maintenance Opti- mization......Page 272
9.1 COST-
EFFECTIVE MAINTENANCE......Page 273
9.2 UNIQUE CHALLENGES IN TURBINE MAINTENANCE
......Page 274
9.3.1 Failure Statistics-Based Approaches......Page 276
9.4 DYNAMIC TURBINE MAINTENANCE OPTIMIZATION
......Page 277
9.4.1 Partially Observable Markov Decision Process......Page 279
9.4.2 Maintenance Optimization Solutions......Page 281
9.4.3 Integration of Optimization and Simulation......Page 285
9.5 DISCUSSION......Page 288
10.1 FORMULATION FOR EXTREME LOAD ANALYSIS......Page 292
10.2 GENERALIZED EXTREME VALUE DISTRIBUTIONS......Page 295
10.3 BINNING METHOD FOR NONSTATIONARY GEV DISTRIBUTION
......Page 297
10.4.1 Conditional Load Model......Page 302
10.4.2 Posterior Distribution of Parameters......Page 305
10.4.3 Wind Characteristics Model......Page 307
10.4.4 Posterior Predictive Distribution......Page 309
10.6.2 Pointwise Credible Intervals......Page 310
10.6.3 Binning versus Spline Methods......Page 314
10.6.4 Estimation of Extreme Load......Page 318
10.6.5 Simulation of Extreme Load......Page 319
Chapter 11: Computer Simulator-Based Load Analysis......Page 326
11.1.2 Deterministic and Stochastic Simulators......Page 327
11.1.3 Simulator versus Emulator......Page 329
11.2.1 Random Sampling for Reliability Analysis......Page 331
11.2.2 Importance Sampling Using Deterministic Simulator
......Page 332
11.3 IMPORTANCE SAMPLING USING STOCHASTIC SIMULATORS
......Page 336
11.3.1 Stochastic Importance Sampling Method 1......Page 337
11.3.3 Benchmark Importance Sampling Method......Page 339
11.4.1 Modeling the Conditional POE......Page 340
11.4.2 Sampling from Importance Sampling Densities......Page 341
11.4.3 The Algorithm......Page 342
11.5.1 Numerical Analysis......Page 344
11.5.2 NREL Simulator Analysis......Page 348
12.1.1 Types of Anomalies......Page 356
12.1.2 Categories of Anomaly Detection Approaches......Page 358
12.1.3 Performance Metrics and Decision Process......Page 360
12.2 BASICS OF FAULT DIAGNOSIS......Page 361
12.2.1 Tree-Based Diagnosis......Page 362
12.2.2 Signature-Based Diagnosis......Page 363
12.3
SIMILARITY METRICS......Page 365
12.3.1 Norm and Distance Metrics......Page 366
12.3.2 Inner Product and Angle-Based Metrics......Page 367
12.3.3 Statistical Distance......Page 369
12.3.4 Geodesic Distance......Page 370
12.4.2 Local Outlier Factor......Page 372
12.4.3 Connectivity-based Outlier Factor......Page 373
12.4.4 Subspace Outlying Degree......Page 374
12.5.1 Graph Model of Data......Page 376
12.5.2 MST Score......Page 377
12.6.1 Benchmark Cases......Page 380
12.6.2 Hydropower Plant Case......Page 385
Bibliography......Page 392
Index......Page 412
