Predicción por Series Temporales Aplicadas a la Estimación de Cargabilidad de Transformadores de Distribución para la Toma de Decisiones.

Abstract

Transformers are electrical machines of major importance for distributors. Their operation is subject to conditions that degrade their insulation system, such as: excessive temperature, presence of humidity combined with electrical and mechanical stress, among others. The technical life expectancy of a transformer is determined by several factors: equipment design, historical and future events, present and future working conditions. The objective proposed was to design and develop an algorithm with the definition of tools associated with the problem of the life of the distribution transformer, applying the use of normal load guides in the framework of the IEC 60076-7 standard, to manage their life history in scenaries of uncertainty (based on weather forecasts). Using an annual database of a typical SET of the EPEC Distributor of the Province of Córdoba, Argentina, a possible tuple scenary of values defined as K1, K2 and TK2 was analyzed and developed, which describes the daily load state of the transformers. For this, a function of correspondence between the historical values of these variables and the present value was defined. Then, through the prediction algorithm, the future values of the tuples were generated with their respective variances. Based on the results obtained in the case study developed, a technique has been defined and used to obtain precise indicators that helps, in advance, to program and manage the loadability of the distribution transformers accurately.