This paper gives a solution for the unit commitment problem (UCP) for an power system to minimize the total production cost over a period of time by considering the load as a random(Probabilistic) variable instead of a deterministic one. The UCP is a mixed integer large scale non linear program. If the number of unit is more and the time horizon is large, then the analysis become more complex. That is the reason still researches are active in finding the solution for unit commitment problem. In this paper the proposed methodology by considering the load as a probabilistic variable instead of regular deterministic one. By considering the load in this way we can solve the problem of UCP as a real time problem. In the proposed methodology LR method is used to separate the problem in to a primal and dual problem and each sub-problem is solved by using dynamic programming. The economic dispatch problem is solved by particle swarm optimization. By comparing the results with conventional method the proposed methodology gives the schedule of generator with minimum production cost. © 2011 IEEE.