Fault Detection, Diagnosis and Prediction in Electrical Valves Using Self-Organizing Maps

This paper presents a proactive maintenance scheme for fault detection, diagnosis and prediction in electrical valves. The proposed scheme is validated with a case study, considering a specific valve used for controlling the oil flow in a distribution network. The scheme is based in self-organizing maps, which perform fault detection and diagnosis, and temporal self-organizing maps for fault prediction. The adopted fault model considers deviations either in torque, in the valve’s gate position or in the opening or closing time. The map which performs the fault detection, diagnosis A presentation based on this article was made at the Eleventh IEEE Latin-American Test Workshop, Punta del Este, Uruguay, March 28–31, 2010. Responsible Editor: F. Vargas L. F. Gonçalves (B) · T. R. Balen · M. S. Lubaszewski · E. L. Schneider · R. V. Henriques Departamento de Engenharia Elétrica, Universidade Federal do Rio Grande do Sul, Av. Osvaldo Aranha, 103, Bom Fim, CEP: 90.035-190, Porto Alegre, RS, Brasil e-mail: [email protected] T. R. Balen e-mail: [email protected] M. S. Lubaszewski e-mail: [email protected] E. L. Schneider e-mail: [email protected] R. V. Henriques e-mail: [email protected] J. L. Bosa Instituto de Informática, Universidade Federal do Rio Grande do Sul, Av. Bento Gonçalves, 9500, Agronomia, CEP: 91.509-900, Porto Alegre, RS, Brasil e-mail: [email protected] and prediction, is trained with the energy spectral density information, obtained from the torque and position signals by applying the wavelet packet transform. These signals are provided by a mathematical model devised for the electrical valve. The training is performed by fault injection based on parameter deviations over this same mathematical model. The proposed system is embedded into an FPGA-based platform. Experimental results demonstrate the effectiveness of the proposed approaches.