Alarm messages with timestamps observed after a fault occurrence in a transmission system consist of rich and useful temporal information. The alarm messages received in a dispatching center are time-tagged and can be expressed as a time series. Based on the time series data mining technique

a fault diagnosing method for power systems is proposed with temporal information of alarm messages taken into account. First

the concept of time series and the similarity matching method are described. Then

the method is modified for addressing the fault diagnosis problem in a power system. The time series model for a transmission system is next developed and the distance of the time series is defined

with the model solved using the subsequence matching method. Comparisons are made between the proposed and two existing transmission system fault diagnosing methods developed in recent years. With the temporal information of alarm messages fully utilized

the proposed method can not only estimate the fault sections

but also identify the malfunctioned protective relays and/or circuit breakers as well as distorted/missing alarms

even for situations with complicated or successive faults. Finally

historical fault scenarios from two actual power systems are used for demonstrating the correctness and efficiency of the fault diagnosing method developed. This work is supported by National Key Technology Research and Development Program (No. 2011BAA07B02) and a project from State Grid Jiangsu Electric Power Company (No. JS2014002).