Distributed control system(DCS) often needs to monitor the extreme value of some important measuring points. The traditional approach is to import specialized external computing system

or build complex function blocks through configuration tools to calculate the extreme values of measuring points

which may affect the stability

real-time and reliability of DCS. In order to improve the efficiency of monitoring and reduce the security risk of the system

a real-time calculation method of clustering extreme values based on graphic configuration is proposed. In this method

the measuring points' data is firstly divided into k initial clusters by the distance-based kmeans algorithm in the historical station

then the density-based local outlier factors(LOF) cluster analysis is used to calculate the extreme points of the k initial clusters. Finally

a real-time memory index library is built on the engineer station to update the information of points

and display the extreme value information in front of the user through the monitoring graphic. At present

this method has been applied in DCS of many thermal power plants. It has been proved by actual engineering that this method can avoid the safety problem of communication between the external extreme value calculation system and the DCS

reduce the complexity of user configuration logic configuration

and ensure the accuracy

real-time and stability of extreme value calculation for measurement points.