Nonlinear probabilistic modeling of incipient fault in underground cables

• Designing a suitable laboratory setup for creating incipient faults in the modeled cable. • Presentation of a polynomial model with eight variants for modeling the incipient fault in the cable. • Selection of the optimum models’ orders. • Finding the best probability distribution function for the models’ parameters. The incipient fault in underground cables is a well-known fault that occurs because of cable insulation failure and water penetration. For assessment of the suggested detection methods for such a fault, one could use practical data from the real field or a suitable model for the generation of data in different conditions. This paper presents a model for incipient faults in underground cables, by which the required data close to the real records can be provided. Although there are various arc models for different arc types, there is no specific model for incipient faults in underground cables. Eight models based on polynomial functions are proposed here for incipient fault modeling. Actual records based on a laboratory setup are used in the models’ identification process. The proposed models are nonlinear and dynamic, in which the model parameters are changed every half cycle based on some probability functions. The optimal model orders are attained for the proposed models. For every model parameter, several probability distribution functions (PDFs) are tested and the most fitted one with the actual data is chosen.