Information Fusion Approach for Transformer Protection: Concept, Application, and Field Data Verification

The ultimate goal of power transformer protection system is to quickly and reliably detect internal faults while remaining stable for external faults and transients. Since in the conventional protection system, each relay decides about the transformer condition independently, it can be considered that the output signals of relays are logically “OR”ed with each other. In this case, confirming internal fault condition by more than one relay does not increase the confidence level of decision-making and consequently the operation speed, while maloperation of one function causes the failure of overall protection system. To cover this drawback, this article introduces a new protection paradigm based on fusing the decision of independent protection functions. The fundamental concepts along with an application of this approach are proposed in the article. Besides, to improve the fusion performance, a probabilistic approach for the protection functions is suggested. Results of the comprehensive evaluations and field data verifications demonstrate that the proposed approach not only increases the operation speed and dependability of protection system, but also improves its security by accurately detecting the transformer status even in cases where a protection function maloperates for a transient condition.