Optimal Fault Diagnosis of Electric Power Systems Using MSE Learning

10.3966/199115992017102805022

Enjie Shi；Qun Li；Tianyu Ren；Chao Wang；Xiaohu Wang

electric power systems ； fault diagnosis ； matrix representation ； MSE learning

電腦學刊

28卷5期（2017 / 10 / 01）

246 - 254

英文

The electric power systems (EPSs) is a complex system with lots of bus bars, transmission lines, and transformers. The primary goal of maintaining an EPS is providing sustainable and stable power supply for customers. However, the quality of service of an EPS suffers from the failures of its system sections. Fast and accurate fault diagnosis is the prerequisite to bring the system back in normal state, thus has attracted much attentions from power engineers. In this paper, we present a minimum square error (MSE) learning based optimal fault diagnosis algorithm, in which the operation state of system sections and protections are formulated using matrices representations. Optimization model is developed to find most probably state of the system sections. The method requires no complex logic designing, nor any historical operation records, and it is simple and fast for implementation. Test results prove that the proposed method has satisfactory diagnosis performance compared with other existing methods.