Abstract
In this paper we have addressed the problem of observability of power systems from the point of view of topological observability and using genetic algorithms for its determination. The objective is to find a way to determine if a system is observable by establishing if a spanning tree of the system that verifies certain properties with regards to the use of available measurements can be obtained. To this end we have developed a genotype-phenotype transformation scheme for genetic algorithms that permits using very simple genetic operators over integer based chromosomes which after a building process can become very complex trees. The procedure was successfully applied to standard benchmark systems and we present some results for one of them.
Highlights
Electrical power transportation networks are in charge of moving energy from generating points to consumer points, called loads
Different optimization techniques may be employed in order to determine the tree with maximum fitness, but due to the imprecise nature of the fitness function and to the complexity of the search space for large power systems, in this paper we have chosen a different approach by using evolutionary techniques for the determination of the spanning tree of full rank
As evolution progresses the fitness of the best individual increases and, even though the average fitness of the population comes close to that of the best individual in some points it never quite prematurely converges to it, preserving enough variety to allow for continued evolution up to the optimal
Summary
Electrical power transportation networks are in charge of moving energy from generating points to consumer points, called loads. Due to the precision problems of numerical methods when operating over such large systems, some authors have proposed approaches based on graphs of the power systems to determine its observability [4][5][6][7]. To address the problem of finding this spanning tree many heuristic solutions have been proposed such as [5][6][7], they usually differ in the algorithm they use for assigning node measurements to branches when constructing the tree, but their basic procedure is very similar in terms of reading the network sequentially and constructing a tree in a stepwise fashion. Different optimization techniques may be employed in order to determine the tree with maximum fitness, but due to the imprecise nature of the fitness function and to the complexity of the search space for large power systems, in this paper we have chosen a different approach by using evolutionary techniques for the determination of the spanning tree of full rank
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call