V.H. Quintana

Knowledge-base reduction based on rough set techniques

Knowledge acquisition is one of the most difficult tasks during the construction of an expert system. Usually, the experts have difficulty in explaining to the knowledge engineers how they solve a given problem. This fact may result in superfluous information about some specific points, while in other points, only an incomplete set of information is available to prepare the knowledge base. This paper presents a contribution to help knowledge engineers to manipulate and reduce knowledge bases for power system operation problems using a systematic approach. The approach is based on rough set theory. An illustrative example is presented in this paper.

Pattern analysis in power system state estimation

In recent years, interest in the application of artificial intelligence technologies to power system operation, planning and design has grown rapidly. The application of non-symbolic techniques, particularly Artificial Neural Networks (ANNs), is a new area of research in this field. In this paper, intelligent systems for solving power system state estimation problems are investigated. A new framework for the solution of the topology determination, observability analysis and bad data processing tasks is proposed. Pattern analysis techniques have been developed to deal with noisy environments. An ANN for topology determination and a supervised learning algorithm for very large training sets, the Optimal Estimate Training 2 (OET2), are introduced. OET2 overcomes the major shortcomings of the back-propagation learning rule and can also be very useful for other problems. Power system network decomposition techniques are used to decrease the computational burden of the topology classfier training session. Tests using the IEEE 24- and 118-bus systems illustrate situations in which the existent tools for data processing fail.

Real-time optimal reactive power control for distribution networks

Abstract A computer algorithm for optimal reactive power and voltage control, suitable f OR large distribution networks, is presented in this paper. The proposed algorithm starts with implementing a partitioning and Krons reduction technique in order to reduce the size of the problem. The algorithm decouples the capacitor problems from the regulator problems and it uses a dynamic programming technique to maximize the reduction in line losses in a real-time environment. The net outcomes of the algorithm are the capacitor bank locations, rating and switching scheme and location of voltage regulators, and tap settings that maximizes the loss reduction. The algorithm ensures that the voltage levels for all buses in the system are within permissible limits at all times. The algorithm is efficient, fast and reliable and can be easily implemented by any utility to optimize line loss reduction on distribution feeders.

Classification of power system operation point using rough set techniques

During the operation of a power system, the system operator is supplied with many data. These data come from measurements into the system or from computational processes. The size of the current database in power control center has increased so much in the last years due to the use of telecommunication process. The system operator needs, to take a decision about an operation in the system (switching, changing taps and voltage levels, and so on), to know the current state of the system and same forecasted position, such as load forecasting, maintenance schedule, and so on. This paper presents an approach to reduce the size of database system, keeping only the essential information to the process. This approach is based on rough set theory.

Dynamical analysis of voltage collapse in longitudinal systems

Abstract In this paper an analysis of the voltage collapse phenomenon in longitudinal systems is presented. The analysis is performed through a small signal perturbation approach where the nonlinear multimachine dynamic model is linearized around a steady-state operating point, and the stability of the system is evaluated through the eigenvalues of the Jacobian matrix of the dynamic state equations. The analysis clarifies the difference between the maximum loadability point and the voltage stability limit. In addition, a voltage collapse proximity indicator, taking into account the dynamic of the system, is proposed and tested in three different networks.

A pattern analysis approach for topology determination, bad data correction and missing measurement estimation in power systems

The authors propose a novel methodology for the combined solution of the topological identification, observability analysis and bad data processing problems in power systems. The idea is to provide, in a very fast way, a reliable input database for the state estimator, without interacting with it. The solution is based on a pattern analysis approach. An efficient framework for solving data acquisition and processing problems, joining pattern analysis and analytical procedures, is suggested. Two different techniques of pattern analysis are combined to produce a classifier and an estimator with unique characteristics to deal with noisy environments. The patterns required for the training process can be acquired from the SCADA (supervisory control and data acquisition) system and/or from load-flow simulations. Test results have been obtained for the IEEE 24-bus reliability test system.<<ETX>>

Associative memory models for data processing

Abstract This paper is an extension of work on a new framework for solving data acquisition and processing problems in power systems. The proposed methodology applies pattern analysis techniques to solve the network configuration, observability analysis and bad data processing problems. Other associative memory models are investigated for the solution of the observability analysis and bad data processing tasks. Special emphasis has been given to pattern analysis tools suitable for massively parallel implementations, such as artificial neural network models. Test results have been obtained for the IEEE 24- and 118-bus test systems.

A probabilistic associative memory and its application to signal processing in electrical power systems

Abstract This paper presents a new associative memory model. Its development is a consequence of work on a new framework (using pattern-analysis techniques) for solving data-acquisition and processing problems in power systems. The proposed probabilistic associative memory is compared with other associative memory models (particularly the ones suitable for massively parallel implementations, such as artificial neural networks) for the solution of the observability analysis and bad data processing tasks in power systems.

Neural networks for topology determination of power systems

The authors describe a parallel distributed topology classifier. The idea is to determine the system configuration in a very fast way, even in the presence of incorrect or unavailable switch/breaker status and analog measurements. A new supervised learning algorithm suitable for very large training sets is introduced.<<ETX>>

A pattern recognition approach for data estimation and debugging in power systems

The authors extend their previous work (1990, 1991) on solving data acquisition and processing problems in power systems. The proposed methodology applies pattern analysis techniques to solve the network configuration, observability analysis and bad-data processing problems. Further associative memory models are investigated for the solution of the observability analysis and bad-data processing tasks. Special emphasis has been put on pattern analysis tools suitable for massively parallel implementations, such as artificial neural network models. Test results have been obtained for the IEEE 24- and 118-bus test systems.