Dynamic Security Assessment and Enhancement of Large Scale Electric Power Systems via Machine Learning Techniques and Population based Optimization Methods

 

Project Details

Funding institution: The Scientific and Technological Research Council of Turkey (TÜBİTAK)

Funding program: 1001 – Scientific and Technological Research Projects Funding Program

Project no: 114E157

December 2014 – December 2016

Principal Investigator: Prof. Dr. V. M. Istemihan Genc

Researchers:  Prof. Dr. Zehra Cataltepe

Advisor: Assoc. Prof. Dr. O. Kaan Erol

 

Project Summary

The main objective of this project is to develop new methods in the field of transient stability related dynamic security assessment and in the field of optimization of preventive and corrective control for enhancing the dynamic security of large scale electric power systems, whereas its main contribution is the general online control methodology proposed.

Conventional methods for dynamic security assessment are far from being a real-time application due to their extensive computations. In the project, artificial neural network based methods are preferred for real-time dynamic security assessment. In this regard, methods using multilayer perceptrons and radial basis function neural networks, as well as a new feature selection method, has been developed.

A proper generation and load rescheduling of some selected generators and loads can be used to drag the operating point of the power system into a secure region from an dynamically insecure point for some contingencies. In this project, generation and load rescheduling has been applied through optimizing a cost subject to many static and dynamic constraints. When preventive control is not sufficient or uneconomical, there may be a need for corrective control in case of the occurrence of contingencies. In this project, corrective control has been applied by load shedding and subsequently by defensive islanding, if necessary.

Both the preventive control and load shedding designs necessitate the solution of some optimization problems under dynamic and static constraints. These problems have been solved by using various population based optimization techniques. Among these methods, genetic algorithms, particle swarm optimization, differential evolution, big bang big crunch, artificial bee colony, and mean-variance mapping optimization methods are selected and their performances are compared. In addition, a new method based on a dynamic search space, which improves the solutions and provides a faster convergence, is proposed.  As a corrective control, in the field of defensive islanding, model and measurement based methods and new methodologies have been developed using machine learning techniques. The methods developed in this project have been applied to different test systems and a model of Turkish electric power system and their performances have been demonstrated.

Publications

  1. Mahdi, M., & Genc, V. M. I. (2019). A real-time self-healing methodology using model- And measurement-based islanding algorithms.
    IEEE Transactions on Smart Grid, 10(2), 1195–1204.
    doi:10.1109/TSG.2017.2760698
    full text
  2. Kucuktezcan, C. F., Genc, V. M. I., & Erol, O. K. (2019). Preventive and Corrective Control Actions on Power Systems via Heuristic Optimization Methods with Consecutive Search Space Reduction.
    Electric Power Components and Systems, 47(1–2), 90–100.
    doi:10.1080/15325008.2019.1575933
    full text
  3. Jafarzadeh, S., Genc, V. M. I., & Cataltepe, Z. (2018). An Online Dynamic Security Assessment in Power Systems Using RBF-R Neural Networks.
    IETE Journal of Research, 2063.
    doi:10.1080/03772063.2018.1527258
    full text
  4. Beyranvand, P., Genc, V. M. I., & Çataltepe, Z. (2018). Multilabel learning for the online transient stability assessment of electric power systems.
    Turkish Journal of Electrical Engineering and Computer Sciences, 26(5), 2661–2675.
    doi:10.3906/elk-1805-151
    full text
  5. Küçüktezcan, C. F., & Genc, V. M. I. (2017). Dynamic security enhancement of power systems using mean-variance mapping optimization.
    Turkish Journal of Electrical Engineering and Computer Sciences, 25(4), 3188–3200.
    doi:10.3906/elk-1608-147
    full text
  6. Beyranvand, P., Kucuktezcan, C. F., Cataltepe, Z., & Genc, V. M. I. (2018). A Novel Feature Selection Method for the Dynamic Security Assessment of Power Systems Based on Multi-Layer Perceptrons. International Journal of Intelligent Systems and Applications in Engineering, 6(1), 53-58.
    full text
  7. Mahdi, M., & Genc, V. M. I. (2017). Adapting the defensive islanding for arming the power system under changing operating conditions. In 2017 IEEE Manchester PowerTech (pp. 1-6). IEEE.
    full text
  8. Kucuktezcan, C. F., Genc, V. M. I., & Erol, O. K. (2016). An optimization method for preventive control using differential evolution with consecutive search space reduction.
    In IEEE PES Innovative Smart Grid Technologies Conference Europe.
    doi:10.1109/ISGTEurope.2016.7856215
    full text
  9. Mahdi, M., & Genc, V. M. I. (2016). K-means and fuzzy relational eigenvector centrality-based clustering algorithms for defensive islanding.
    In IEEE PES Innovative Smart Grid Technologies Conference Europe.
    doi:10.1109/ISGTEurope.2016.7856210
    full text
  10. Mahdi, M., & Genc, V. M. I. (2016). A coherency based generation rescheduling against multiple contingencies. In 2016 4th International Istanbul Smart Grid Congress and Fair (ICSG)(pp. 1-5). IEEE.
    full text
  11. Mahdi, M., & Genc, V. M. I. (2015). A security assessment for slow coherency based defensive islanding. In 2015 9th International Conference on Electrical and Electronics Engineering (ELECO) (pp. 536-540). IEEE.
    full text
  12. Mahdi, M., & Genc, I. (2015). Defensive islanding using self-organizing maps neural networks and hierarchical clustering. In 2015 IEEE Eindhoven PowerTech (pp. 1-5). IEEE.
    full text

Ph.D. Theses

    1. Mahdi, M. (2018). Wide-area measurement-based early prediction and corrective control for transient stability in power systems.
      Thesis no: 506374
      search for full text
    2. Kucuktezcan, C. F. (2015). Dynamic security enhancement of power systems via population based optimization methods integrated with artificial neural networks.
      Thesis no: 397782
      search for full text