Metaheuristics for Intelligent Electrical Networks

Frédéric Héliodore is Director of Data & Analytics at General Electric Grid Solutions.Amir Nakib is Associate Professor in Computer Science at University Paris-Est, France and member of the optimization team of the LISSI laboratory.Boussaad Ismail is Research Engineer at General Electric Grid Solutions.Salma Ouchraa is a PhD student in Computer Science at the Laboratory of Conception and Systems in the Faculty of Science at University Mohammed V in Rabat, Morocco.Laurent Schmitt is the Secretary General of the European Network of Transmission System Operators for Electricity (ENTSO-E).

• Introduction xiChapter 1 Single Solution Based Metaheuristics 11.1 Introduction 11.2 The descent method 21.3 Simulated annealing 31.4 Microcanonical annealing 41.5 Tabu search 61.6 Pattern search algorithms 61.6.1 The GRASP method 71.6.2 Variable neighborhood search 81.6.3 Guided local search 101.6.4 Iterated local search 111.7 Other methods 121.7.1 The Nelder–Mead simplex method 131.7.2 The noising method 141.7.3 Smoothing methods 151.8 Conclusion 16Chapter 2 Population-based Methods 172.1 Introduction 172.2 Evolutionary algorithms 182.2.1 Genetic algorithms 182.2.2 Evolution strategies 202.2.3 Coevolutionary algorithms 212.2.4 Cultural algorithms 212.2.5 Differential evolution 232.2.6 Biogeography-based optimization 252.2.7 Hybrid metaheuristic based on Bayesian estimation 272.3 Swarm intelligence 292.3.1 Particle Swarm Optimization 292.3.2 Ant colony optimization 322.3.3 Cuckoo search 352.3.4 The firefly algorithm 362.3.5 The fireworks algorithm 382.4 Conclusion 42Chapter 3 Performance Evaluation of Metaheuristics 433.1 Introduction 433.2 Performance measures 443.2.1 Quality of solutions 443.2.2 Computational effort 453.2.3 Robustness 463.3 Statistical analysis 463.3.1 Data description 473.3.2 Statistical tests 483.4 Literature benchmarks 493.4.1 Characteristics of a test function 493.4.2 Test functions 503.5 Conclusion 58Chapter 4 Metaheuristics for FACTS Placement and Sizing 594.1 Introduction 594.2 FACTS devices 614.2.1 The SVC 624.2.2 The STATCOM 634.2.3 The TCSC 634.2.4 The UPFC 634.3 The PF model and its solution 644.3.1 The PF model 644.3.2 Solution of the network equations 664.3.3 FACTS implementation and network modification 694.3.4 Formulation of FACTS placement problem as an optimization issue 694.4 PSO for FACTS placement 724.4.1 Solutions coding 734.4.2 Binary particle swarm optimization 754.4.3 Proposed Lévy-based hybrid PSO algorithm 824.4.4 “Hybridization” of continuous and discrete PSO algorithms for application to the positioning and sizing of FACTS 994.5 Application to the placement and sizing of two FACTS 1004.5.1 Application to the 30-node IEEE network 1034.5.2 Application to the IEEE 57-node network 1044.5.3. Significance of the modified velocity likelihoods method 1094.5.4 Influence of the upper and lower bounds on the velocity Vciof particles ci 1114.5.5 Optimization of the placement of several FACTS of different types (general case) 1154.6 Conclusion 118Chapter 5 Genetic Algorithm-based Wind Farm Topology Optimization 1215.1 Introduction 1215.2 Problem statement 1225.2.1 Context 1225.2.2 Calculation of power flow in wind turbine connection cables 1255.3 Genetic algorithms and adaptation to our problem 1295.3.1 Solution encoding 1295.3.2 Selection operator 1315.3.3 Crossover 1325.3.4 Mutation 1355.4 Application 1375.4.1 Application to farms of 15–20 wind turbines 1405.4.2 Application to a farm of 30 wind turbines 1405.4.3 Solution of a farm of 30 turbines proposed by human expertise 1445.4.4 Validation 1455.5 Conclusion 145Chapter 6 Topological Study of Electrical Networks 1496.1 Introduction 1496.2 Topological study of networks 1506.2.1 Random graphs 1516.2.2 Generalized random graphs 1516.2.3 Small-world networks 1526.2.4 Scale-free networks 1526.2.5 Some results inspired by the theory of percolation 1536.2.6 Network dynamic robustness 1606.3 Topological analysis of the Colombian electrical network 1616.3.1 Phenomenological characteristics 1616.3.2 Fractal dimension 1696.3.3 Network robustness 1796.4 Conclusion 182Chapter 7. Parameter Estimation of α-Stable Distributions 1837.1 Introduction 1837.2 Lévy probability distribution 1847.2.1 Definitions 1847.2.2 McCulloch α-stable distribution generator 1897.3 Elaboration of our non-parametric α-stable distribution estimator 1917.3.1 Statistical tests 1927.3.2 Identification of the optimization problem and design of the non-parametric estimator 1957.4 Results and comparison with benchmarks 1977.4.1 Validation with benchmarks 1977.4.2 Parallelization of the process on a GP/GPU card 2117.5 Conclusion 220Chapter 8 SmartGrid and MicroGrid Perspectives 2218.1 New SmartGrid concepts 2218.2 Key elements for SmartGrid deployment 2248.2.1 Improvement of network resilience in the face of catastrophic climate events 2258.2.2 Increasing electrical network efficiency 2278.2.3 Integration of the variability of renewable energy sources 2298.3 SmartGrids and components technology architecture 2318.3.1 Global SmartGrid architecture 2318.3.2 Basic technological elements for SmartGrids 2328.3.3 Integration of new MicroGrid layers: definition 235Appendix 1 241Appendix 2 245Bibliography 251Index 265