Optimal Power Flow Solutions With Fuzzy Load

The power system secure and economic operation becomes more important withthe development of the interconnected larger power system and power market. As apowerful tool, optimal power flow (OPF) has been applied and researched widely. Theconventional OPF has generally been addressed as a certain optimization problem.However, A lot of uncertain factors exist in the practical power system and this greatlylimits the application of OPF. Therefore, OPF model and algorithm considering fuzzyload are studied in this thesis. The main contents are as follows.This thesis reviews the applications of the fuzzy set theory to solving OPF. Itintroduces the several estimating methods of the trapezoidal fuzzy parameters. Based onthe character that the fitting errors of multiple load samples are normal distribution, anew simple estimating method is presented.Considering active and reactive power combination optimization, this thesis presentsa nonlinear fuzzy optimal power flow model through addressing fuzzy load, in whichthe objective function is the minimum generator cost and the constraints include ACpower flow equations, security limits of line currents and voltage magnitude, limits ofcontrol variables etc. Compared with the existing fuzzy OPF models, which used linearor DC power flow equations and disregarded reactive control variables, the model inthis thesis is more practical.Tow novel optimal strategies are presented to solve the new fuzzy OPF problem.The operation principles of trapezoidal fuzzy numbers is used in the first strategy totransform the fuzzy OPF problem into a certain problem solved by thepredictor-corrector primal-dual interior point algorithm. The weighted sum of fourbreak-point values in the original fuzzy objective is the new objective function. Besidesthe original four break-point constraints, the trapezoidal limits among break points offuzzy control variables are added to new constraints. The strategy is to obtain optimalpossibility distributions of the control variables, which pursues the most profit underensuring security.Another strategy combines the AC fuzzy power flow and the genetic algorithm (GA)to directly solve the new problem. Giving fuzzy load distribution, the initiate populationof control variables is randomly selected. Through calculating AC fuzzy power flow ofevery individual, the corresponding state fuzzy distribution and weighted sum objectiveare known. Then the fitness function is calculated from the feasibility and the objectivevalue to estimate the quality of individual. By the selection,crossover and mutation GAoperations of control variables, the optimal control variables are finally obtained. Comparing the second strategy, the first one can gain more profit, but havedifficulty in practices because of the fuzzy control variables. The second sacrifices afew profit, but it get an only certain set of control variables that can be implementedeasily. Numerical results on IEEE14 test system demonstrate the model and optimalstrategies in this thesis are feasible and effective.