Research On The Distribution Reactive Power Optimization Based On Genetic Algorithm

Due to the rapid development of China’s national economy, the power demand has alsoincreased.Under the protection of the reliability of power supply of distribution network, howto reduce the power losses and improve the supply quality and economical operation of thesystem has drawn increasing attention and concern of The electricity workers.The reactivepower optimization of distribution network can not only stabilize the system voltage level, butalso effectively reduce the electrical energy loss of the distribution network, while plays avery important role to protect the stability and economy of the distribution system.Firstly, in-depth study with theoretical knowledge of reactive power compensation, andanalyze a variety of different power flow calculation algorithms are introduced, currentresearch status of power system reactive power optimization are analyzed as well. Varietiesof traditional optimization algorithms as well as several heuristic optimization algorithms forsolving power system reactive power optimization problem model are discussed for furtheranalysis.Secondly, the recursive backward/forward power flow algorithm is introduced in thisthesis, which is used for related power flow calculations in reactive power optimizationalgorithm for distribution power networks. During the process of distribution network powerflow calculation, the node objects and branch objects are created accordingto the object-oriented programming concepts. A simple graph object to represet thedistribution network topology can easily be created by integrating the node and branch objectswhile taking advantage of the SimpleGraph class provided by the open source JAVApackage JgraphT, and hence thesequence of nodes in power flow calculation can bedetermined by the built-in Dijkstra shortest-path calculation algorithm from the same JAVApackage. Since the tree structure of traditional distribution power networks is intrinsicallyrecursivable, recursive forward/backward calculation method is developed to simplify thealgorithm and improve the behavior of the entire program.A practical optimization model is built to satisfy the optimization of comprehensiveeconomic benefits of the distribution system by minimizing the sum of distributionnetwork power loss costs and reactive power compensation expenses, power flow equationsare used as the equality constraints, while node voltage and maximum reactive compensationshould fulfill the inequality constraints. To solve the reactive power optimization problem forthe practical mathematical model, the genetic algorithm object in Jenetics (another opensource JAVA package) is created by implement the general FitnessFunction interface, while a customized fitness function class is developed to describe the object of the optimizationproblem in this theis.The reactive power optimization program based on the special power flow algorithm andthe object-oriented genetic algorithm introduced in this thesis is applied for several realdistribution networks, optimized reactive power compensation location and amount can bedrawn by the program, which shows good economic benefit, and the effectiveness isvalidated.