In order to solve the problem of grouping of parallel compensation capacity in substation

this paper presents an optimal grouping method that considers the timing characteristics of reactive power demand. The method consists of two steps. The first is to partition the reactive power demand profile into several segments

in which the demand can be regarded as invariable

and the ratio of the sum of the variance in each segment and the sum of the distance between adjacent segments is taken as the objective function to partition the profile; then the partitioning results are clustered using the modified fuzzy C-means clustering （FCM） algorithm. In the clustering

the number of clusters is the number of capacitor groups

and the difference between cluster centers is the capacity of each group. Furthermore

the effect of the number of profile partitioning on the clustering results is studied when the number of capacitor groups is known. Finally

simulation of an equivalent actual power grid and load profile is made to demonstrate the effectiveness and rationality of the presented method. In the simulation

the technical indicators such as energy loss

voltage quality

the number of operations of capacitor and power factor are comprehensively considered

and the results are compared with those of the optimal cover method. The energy loss of the method is close to the optimal cover method

but the number of operations of capacitor is greatly reduced.