A feedforward neural network (FFNN) has been trained to discriminate between power transformer magnetizing inrush and fault currents The training algorithm used was backpropagation, assuming initially a sigmoid transfer function for the network's processing units ("neurons") Once the network was trained the units' transfer function was changed to hard limiters with thresholds equal to the biases obtained for the sigmoids during training The off-line experimental results presented in this paper show that a FFNN may be considered as an alternative method to make the discrimination between inrush and fault currents in a digital relay implementation >

Training an artificial neural network to discriminate between magnetizing inrush and internal faults

In this work, an artificial neural network based algorithm for a three phase power transformer protection scheme is developed and implemented in real time using the DS-1102 digital signal processor. Distinguishing between the magnetizing inrush and internal fault currents is always a consideration for power transformer protection. Existing methods, mostly based on harmonic restraint, are not very reliable for modern transformer protection. The reason for loss of reliability is that the use of low-loss amorphous material in modern transformer cores causes reduced second harmonic content of the magnetizing inrush current. Other methods based on the transformer equivalent circuit model are susceptible to parameter variations and hence are not suitable under all operating conditions. The work presented here shows the usefulness of the artificial neural network which is able to distinguish between the magnetizing inrush and the internal fault currents without harmonic decomposition or using the transformer equivalent circuit model. The inherent advantages of the generalization and the pattern recognition characteristics make the artificial neural network based method quite suitable for distinguishing between magnetizing inrush and the internal fault currents. In this work, a two-layer artificial neural network with sixteen inputs and one output is designed. The data to train and test the artificial neural network are experimentally obtained. The artificial neural network is trained with an input data set and subsequently tested with a different data set. The off-line test results show that the artificial neural network is quite capable of distinguishing between the magnetizing inrush and internal fault currents. Finally, the on-line implementation successfully establishes the efficacy of the ANN based algorithm for power transformer protection.

Artificial neural network based protection of power transformers

This paper presents a digital relay scheme for the primary protection of power transformers. A multi-criteria algorithm is developed based on fuzzy sets for the decision making part of the scheme. The cost of wrong decision-making and the amount of information inflow are used along with several standard criteria to improve the reliability of the protection. It is shown that the proposed scheme, if properly tuned, can enhance the sensitivity and selectivity of the digital relay and mitigate problems associated with conventional relay schemes. >

A multi-criteria differential transformer relay based on fuzzy logic

Power transformer protective relay should block the tripping during magnetizing inrush and rapidly operate the tripping during internal faults. The frequency environment of power system has been made more complicated and the quantity of 2nd frequency component in inrush state has been decreased because of the improvement of core steel. Then, traditional approaches will likely be maloperated in the case of magnetizing inrush with low second harmonic component and internal faults with high second harmonic component. This paper proposes a new relaying algorithm to enhance the fault detection sensitivities of conventional techniques by using a fuzzy logic approach. The proposed fuzzy-based relaying algorithm consists of flux-differential current derivative curve, harmonic restraint, and percentage differential characteristic curve. The proposed relaying was tested with relaying signals obtained from Salford EMTP simulation package and showed a fast and accurate trip operation.

Fuzzy logic-based relaying for large power transformer protection

This paper suggests the possibility of improving digital power transformer protection. The establishment of inrush in power transformers is becoming unreliable in existing numerical protection. An artificial neural network (ANN) was applied to inrush detection. The saturation of protective current transformers (CT) cannot be totally eliminated despite proper dimensioning. ANN was used for the reconstruction of distorted secondary CT currents due to saturation. In both cases, an ANN was included in the protection algorithm as an extension of the existing methods, which improved the reliability of the protection operation. The paper presents the digital protection algorithm completed in this way and the laboratory equipment by means of which experimental results were obtained. The results confirm faster and more reliable recognition of transformer inrush, as well as satisfactory reconstruction of the distorted secondary CT currents.

Improved operation of power transformer protection using artificial neural network

The complete design of a stand-alone prototype digital protective relay for three-phase power transformers is described. The major emphasis is on the detailed description of the hardware and software of the prototype relay. The protection functions implemented include: a percentage differential protection with a second-harmonic restraint for magnetizing inrush and a fifth-harmonic restraint for overexcitation conditions, and a separate protection for high-impedance primary and secondary ground faults. The relay design is tested with the Fourier algorithm, and any other relay algorithm can be used by replacing only one subroutine. The relay hardware consists of a data acquisition board and a digital processing board which is based on the TMS320E15 processor. Sample real-time test cases are discussed. The results show that the relay never misoperated and correctly identified all the faults. >

I. Hermanto

Papers

A Stand-alone Digital Protective Relay for Power Transformers