Forced outages information treatment system and cause identification based on mobile computing and neural networks

Abstract

The use of outage management systems are becoming vital for distribution utilities in the new paradigm of power systems competition and deregulation. These efforts include fuzzy logic, expert systems, heuristic applications and neural networks. These techniques allow to process and validate real-time data in order to feed a datahase for statistic analysis. This paper presents a computational system for forced outage event acquisition, validation and cause identification based on mobile computing and neural networks. The system is divided into two modules: one to collect and validate, based on mobile computing, forced-outage event data information (conditions in the surrounding environments) at local; and an other one to identify and classify the forced-outage caused using artificial neural networks (ANN).

Utilização de computação móvel para qualificação de rotinas de operação e manutenção de redes de distribuição

“…Esse sistema de coleta de dados pode ser considerado o primeiro módulo de um sistema de tratamento das informações de falta de energia. O segundo módulo consiste em um aplicativo que irá analisar as informações colhidas em campo pelo computador de mão e, através de um algoritmo computacional, identificar as possíveis causas do evento (Pretto at al, 2003a). …”

“…O usuário podia entrar com os dados com o uso da caneta touch screen ou através das teclas de acesso. Na figura 6 pode ser vista uma tela do sistema desenvolvido em Java para Palm OS R (Pretto et al,2003a).…”

Forced Outage Cause Identification Based on Bayesian Networks

“…An accurate diagnostic is directly related to the amount and the quality of the collected information at the event site, the conditions of the surrounding environment and the correct determination of the facts and activities that could indicate the cause of the forced outage. These facts had motivated the creation of a system to collect and process all the relevant information about forced outages based on mobile computing [11] [12], removing from the maintenance technician the function of identifying the outage cause at the moment of energy supply restoration. In this system, the electrician has to choose the most appropriate answer in a questionnaire, attempting to describe the network conditions and surrounding variables at the time of the event.…”

“…Techniques such as Artificial Neural Networks [11] and Fuzzy Logic [12] have been previously used in an attempt to create a system capable to identify, based on evidences collected in the field, the most probable causes that led to an outage. However, due to the lack of appropriate information about the forced outage events, was difficult to develop a computational system based on historical data.…”

“…The first module of the system is the mobile application used to collect the outage event data, developed in [11] and [13]. In the second module, a Bayesian Network is trained with the collected data to identify possible causes for a forced outage event.…”