Employee Behavior Analysis Of Terminal Building Based On Access Data

With the development of civil aviation, passenger population and aviation business are also increasing rapidly, nowadays many airports have built large terminal buildings. For the perspective of airports, it's really a troublesome work to unify employees' behavior, optimize staff workflow, implement sufficient security management, and identify abnormal behavior for the huge terminal buildings.The key to solve the problem: Access control system is used to implement the control of physical accesses in the terminal building and its internal regions. The access control system of terminal building has a large number of control points. The existing access control systems can achieve unified control and log for the points. So many control points generate tens of thousands records every day. According to these records, we can get path information of each department's staff, find out both normal and abnormal patterns of employees' behavior, to better serve for airport management.In this paper, logs of an access control management system in the airport terminal building are studied, from the aspect of sequential pattern mining and physical intrusion detection. Improved Prefix Span algorithm and subsequence matching position algorithm are put forward. These two algorithms can reasonably exhume the mode of employees behavior.Here is the main work of the paper: A, detailed elaborates pretreatment process of staff path data within the terminal building, including data cleaning, equivalence partitioning, etc. B,before constructing prefix frequent pattern projection database, check whether have been constructed forward or not, to avoid the same frequent prefix patterns constructing projection database repeatedly. C, because of the specificity and uniqueness of the path data, traditional data mining algorithms are not suitable, so this paper put forward the improved PrefixSpan algorithm based on path coding. This algorithm optimizes its time and space complexity,firstly gets original data path encoded, reducing space complexity of mining; secondly check the recently suffix of the projection sequence about prefix only in the recursive mining process, avoiding illegal frequent patterns and their projection database. Stop scanning the projection database when the projection sequence number is less than minimum support number, reducing process of the illegal projection database. This algorithm is used to mine frequent paths of each department, which is also known as the normal behavior of employees(high-frequency behavior) mode. Establishes normal behavior mode tree based on high-frequency behavior mode. Sees if some employee's behavior matches normal by matching it with the normal behavior patterns with the subsequence exact match positioningalgorithm based preorder tree. Finally, applies the normal behavior patterns and abnormal behavior patterns to behavior analysis system, and designs a behavior analysis system for employees of terminal building.