Power System Distributed On-line Fault Section Estimation Using Decision Tree Based Neural Nets Approach

This paper proposes a distributed neural nets decision approach to on-line estimation of the fault section of a transmission and distribution (T&D) system. The distributed processing alleviates the burden of communication between the control center and local substations, and increases the reliability and flexibility of the diagnosis system. Besides, by using the algorithms of data-driven decision tree induction and direct mapping from the decision tree into neural net, the proposed diagnosis system features parallel processing and easy implementation, overcoming the limitations of overly large and complex system. The approach has been practically tested on a typical Taiwan Power (Taipower) T&D system. The feasibility of such a diagnosis system is presented.