On-line Monitoring Of Transformer Windings Based On Parameter Estimation

Oil-gas and part-discharging signals are most frequently studied in on-line monitoring of transformers. Oil-gas can reflect the operation situation and is used for diagnosis of fault type. However, oil-gas cannot meet the needs of fault location and state maintenance. Part-discharging signals can realize fault location in theory, it subjects deeply to the influence of circumstances and cannot be carried out easily. This paper proposes real-time monitoring partly via the important component - windings. It is possible to judge the side or phase of fault windings, which satisfies the meet of state maintenance.After analyzing the phenomenon of magnetizing inrush current and theory limitations of differential protection. A winding model is developed. It is proposed to use resistance and leakage inductance of windings for on-line monitoring. It is found that the two parameters aren't influenced by external fault and magnetizing inrush current, which ensures good selectivity.A recursive least-square technique is used for the on-line estimation of the twro parameters. The results of test simulations indicate that it is possible to judge whether occurring fault on windings and even which side or phase by parameter changing curves.Finally, this paper also provides the realization of software and hardware based on the above principium.