Model Predictive Control Approach to Two-Stage Stochastic Power System Optimization Problems Open Access
Downloadable ContentDownload PDF
The optimal-power-flow-based problems, such as economic dispatch, unit commitment, and optimal power flow, are subject to various uncertainties which include, but are not limited to, demand fluctuation, generation/ transmission outages, adverse weather conditions, and electricity pricing. The large integration of renewable energy resources, such as wind and solar, has further caused additional uncertainties due to the variable and unpredictable nature of these resources. However, the next generation smart power systems are also equipped with enabling technologies such as control, communication, and powerful computing capabilities that could be utilized to better deal with these uncertainties and operate the power system at a stable, reliable, and economic operating point. Stochastic programming is a powerful tool that enables power system operators to deal with such uncertainties. However, the stochastic programming needs a large number of scenarios to generate an accurate solution. Considering a large number of scenarios in stochastic programming problem increases the computation time, and may lead to an in-tractable problem. Scenario reduction is usually used to reduce the number of scenarios as a tradeoff between computational time, and solution accuracy.The objective of this dissertation is to formulate and solve two-stage stochastic programming problems that consider a large number of scenarios, without the need for using a scenario reduction technique that compromises the accuracy of the solution. To this end, a multi-stage model predictive control (MPC) approach is proposed to compensate the forecast error in a scenario-based two-stage stochastic problem through a feedback mechanism. Reformulating the problem as a finite moving-horizon optimal control problem, the proposed approach decelerates the growth of the number of scenarios by updating the system as uncertainties are gradually realized. Scenario decomposition approach is used to divide this large-scale optimization problem into several subproblems. Since each subproblem is an independent problem, the optimal solution is obtained by solving these independent problems in parallel, in a coordinated fashion. Consequently, the computation time is reduced, since the computation is distributed among these parallel independent problems. Therefore, the problem is solved without the need for using a scenario reduction technique that compromises the accuracy of the solution. To exhibit the computational efficiency of the proposed model and multi-stage MPC approach, numerical experiments are conducted on the IEEE 118-bus system with 54 units.