V. Leonardo Paucar

Artificial neural networks for solving the power flow problem in electric power systems

Abstract In this paper, the use of artificial neural networks (ANN) is proposed for solving the well known power flow (PF) problem of electric power systems (EPS). PF evaluates the steady state of EPS and is a fundamental tool for planning, operation and control of modern power systems. The mathematical model of the PF comprises a set of non-linear algebraic equations conventionally solved with the Newton-Raphson method or its decoupled versions. In order to take advantage of the superior speed of ANN over conventional PF methods, multilayer perceptrons neural networks trained with the second order Levenberg–Marquardt method have been used for computing voltages magnitudes and angles of the PF problem. The proposed ANN methodology has been successfully tested using the IEEE-30 bus system.

A review of power system voltage and angular stability dynamics

This work reviews important concepts related to power system dynamics focusing on voltage stability. Mechanisms inherent to this phenomenon are investigated using an extended transient stability program. Through time domain simulations, it is shown stability scenarios involving under-load tap changers, overexcitation limiters, thermostatically controlled loads, loss of synchronism, partial voltage collapse and total voltage collapse. Remedial actions against voltage collapse are also implemented and tested. The computer simulations are carried out using an equivalent model representing the Brazilian south-southeast power system.

Problem solving in combinatorial optimization based on the algorithmic model of attack behavior of locust swarm

The attack behavior of locust swarm has inspired the recently developed computer algorithmic model of locust swarm optimizer (LS) within the swarm artificial intelligence methods. Some NP-hard problems of combinatorial optimization that LS may solve are the following: the traveling salesman problem, the knapsack problem, routing in telecommunication networks, and cellular robotic systems. General models of these combinatorial optimization problems which can be solved with locust swarm optimizer technique are proposed in the present work.

Optimal power flow solution including the syncronous generator capability curve constraints with a convex relaxation method

In this work is proposed a solution method for the optimal power flow (OPF) problem, including the synchronous generator capability curve constraints (SGCC). OPF is a non-convex, non-linear, and hard to solve optimization problem. When synchronous generator operational limits are not concerned, OPF solution could not be applicable in real scenarios, taking the machine to operate with winding currents above acceptable values. The SGCC used in this research includes current and power limits allowed in the generator, preventing the machine operation against overheating and excessive mechanical efforts. Convex relaxation methods show a great performance, allowing the problem to be solved in a simpler and way. Thus, OPF is formulated over a second order cone programming (SOCP) approach from convex relaxation techniques, aiming the active power generators cost minimization, subject to the electrical transmission network constraints. SGCC used has a convex form, which allows the convex programming method application, with no convexification need for the optimization model. The proposed method was implemented in MATLAB® environment and applied to IEEE 30-bus test system. Simulation results demonstrate the applicability and good performance of the proposed method, so that solutions do not violate the constraints imposed by the actual operational limits of the generators.

Comparative analysis of game theory application to various types of auctions in electric markets

The various changes that have occurred in the electricity markets around the world have introduced competition among the participants in this sector. Thus, the generation agents that participate on electrical energy auctions may prefer to build optimal strategies to maximize their profits. Mathematical models based on game theory have been used in the analysis of electric energy markets and especially in electricity auctions. In this sense, this theory optimizes the decision-making process for setting prices offered by the generators to the system operator in a power auction. This paper deals with a comparative analysis of individual strategies of generating units, using a non-cooperative game theory approach and incomplete information on various types of auctions. A modified version of IEEE 57-bus test system is used to illustrate the main features of the auction models used.

Performance analysis of algorithms based on intelligence of plants

Artificial intelligence (AI) is a branch of computer science that studies the intelligent behavior of living beings, and mimics this intelligence by deploying it in computer programs, machines and systems in order to solve problems related to searching, optimization, planning, control, automation, etc. One of the areas of artificial intelligence is evolutionary computation, which is inspired by the principle of natural evolution of species. Within the evolutionary computation several methods based on the intelligence of plants have been recently proposed. How the plants survive and adapt in harsh environments has aroused interest of researchers in AI. It is remarkable that the life cycle of a plant is extremely intriguing. The way the plants reproduce, propagate, disperse their seeds and select the most resistant is undoubtedly an evidence of intelligence of plants when optimize their existence. In this sense, some computer algorithms have been proposed based on the intelligent lifecycle of plants. These algorithms are in many cases, simple to implement, and efficient in solving complex problems. In this work, the performance of three algorithms, the flower pollination algorithm, strawberry plant algorithm and invasive weed optimization, all of them based on the intelligent behavior of plants, are analyzed when applied to optimization of test functions, and they are also compared with classical genetic algorithms.