Abstract: Control operation is an important part of the whole ATC system operation, and the confirmation of its state provides basis for unsafe accidents prevention. Firstly, evaluation variables of control operation state are analyzed and extracted from the aspect of controller workload. Secondly, 6 kinds of system cluster analyzing methods are used to classify the sample data, including single method, complete method, median method, average method,centroid method and Ward method. Summarizing the advantages and disadvantages of each method and their clustering results, final clustering results are obtained by ward method, basing on which, control operation state is divided into control operation healthy state and control operation sub-healthy state, and the latter one is then divided into control operation stable sub-healthy state and control operation unstable sub-healthy state. Analyzing results show that the classification is scientific and reliable, and is of great significance for the theoretical study of safety management.

Key words: control operation, cluster analysis, stable sub-health, unstable sub-health