Alarm Association Rules Mining In Civil Aviation Passenger Service Information System

Civil aviation passenger information service system is an important system of the national civil aviation industry. It carries a lot of important aviation business. The system needs to maintain every day without interruption of continuous operation. Paralysis of the system or even an application service interrupted may lead to paralysis of the civil aviation business. Therefore, the state and Civil Aviation Authority make high demands for the safe and reliable operation of the civil aviation passenger service information system. Once the system produced alarm information, which may lead to other related system produce alarm information. So, a large number of alarm information brings great difficulties to accurately find the root of causing alarm information for system maintenance personnel. Therefore,obtaining accurate alarm information and reasonable analysis of alarm information characteristic and mining alarm association rules are important tasks to be solved. The main work is as follows:Because the alarm data of civil aviation passenger service information system has the unstructured characteristic, this paper puts the data structured method based on XML data format. Extract the attribute values of the alarm data based on this method. Alarm data is distributed in the time dimension, the time sliding window model is used to process the data.For the setting of the time sliding window size, a lot of experiments were carried out and make sure that the time window size depends on the alarm event duration. Then mine alarm association rules on processed data by using Apriori algorithm. There are three kinds of partial ordering relation in alarm sequence: serial relationship, parallel relationship, mixed relationship. Determine the relationship that the alarm rules belong, divide the significance and the application of the alarm association rules, achieve the goal of tracing the source of the fault, compress the redundant alarm information, forecast alarm event. These alarm association rules provide decision support for the system maintenance personnel.