Modeling of electric power transformer with on-load tap changer voltage control using complex-valued neural networks

Use of Complex-Valued Neural Networks (CVNN) is proposed for modeling of power transformer. An advantage of this approach is possibility to build accurate and precise models, training the network with previously simulated or with measured real data (transformer's voltages and currents). Inherent capability of CVNN to handle complex values makes it convenient to deal with electrical values in their common representation as rotating vectors during modeling. Due to neural networks' nature, it is possible to take into account different spontaneous factors which hardly can be precisely considered in analytical models. The paper describes modeling of a transformer system with On-Load Tap Changer (OLTC) voltage control using analytical method and new CVNN-based method. In the first part of the paper analytical model is introduced. The model is based on conventional transformer's equations complemented with nonlinearities in magnetizing system, ambient temperature influence on windings and OLTC voltage stabilization. Typical day-long load curve is used for the simulation. The second part of the paper describes basics of CVNNs and the application of the approach for modeling of the transformer system. Data generated by analytical model is used for training and verification of derived CVNN-based model. Verification shows that CVNN is capable to track nonlinear dynamics of power equipment. Proposed method can be considered as basics for further developments of CVNN use in the field of electrical equipment modeling.

[1]  Tülay Adali,et al.  Fully Complex Multi-Layer Perceptron Network for Nonlinear Signal Processing , 2002, J. VLSI Signal Process..

[2]  S. V. Minchev,et al.  Neural networks for modeling of dynamic systems with hysteresis , 2002, Proceedings First International IEEE Symposium Intelligent Systems.

[3]  S.V. Minchev,et al.  Modeling of nonlinear dynamic systems via discrete-time recurrent neural networks and variational training algorithm , 2004, 2004 2nd International IEEE Conference on 'Intelligent Systems'. Proceedings (IEEE Cat. No.04EX791).

[4]  Alexey Minin,et al.  Comparison of the Complex Valued and Real Valued Neural Networks Trained with Gradient Descent and Random Search Algorithms , 2011, ESANN.

[5]  R.P. Maheshwari,et al.  Power Transformer Differential Protection Based On Optimal Probabilistic Neural Network , 2010, IEEE Transactions on Power Delivery.

[6]  Ralph Neuneier,et al.  How to Train Neural Networks , 2012, Neural Networks: Tricks of the Trade.

[7]  B. A. D. H. Brandwood.  A complex gradient operator and its applica-tion in adaptive array theory , 1983 .

[8]  Henry Leung,et al.  The complex backpropagation algorithm , 1991, IEEE Trans. Signal Process..

[9]  Alexey Minin,et al.  Historical Consistent Complex Valued Recurrent Neural Network , 2011, ICANN.

[10]  Alois Knoll,et al.  Modeling of electric power transformer using complex-valued neural networks , 2011 .

[11]  A. El-Hag,et al.  A cascade of artificial neural networks to predict transformers oil parameters , 2009, IEEE Transactions on Dielectrics and Electrical Insulation.