Unit Commitment Solution using Agglomerative and Divisive Cluster Algorithm - An Effective New Methodology

N.M. Reddy, N.V. Ramana, and K.R. Reddy (India)


Power System operation and control, Unit Commitment Solution, Cluster Algorithms, Objective Function and Constraints.


A new methodology for the most complex unit commitment problem using agglomerative and divisive hierarchical clustering is presented. Euclidean costs, which is a measure of difference in fuel cost and start-up costs of any two units are first calculated. Depending upon the value of Euclidean costs, similar type of units are placed in a cluster. This cost is also useful for preparing priority lists for the units in a cluster and forming the different clusters. Proposed methodology has two individual algorithms. While the load is increasing, agglomerative cluster algorithm is proposed. Divisive cluster algorithm is used when the load is decreasing. The search is carried for an optimal solution from a minimal number of clusters and cluster data points. The performance of the method is evaluated on a standard ten thermal unit power system for a period of 24 hours. Numerical results are presented and compared with the existing popular methods in most judicious way.

Important Links:

Go Back