Optimization Research For Self-healing And Dispatch In Active Distribution Network Based On Intelligent Algorithm

Recently,governments around the world made new plants to strongly develop the renewable clean resource, especially wind power and photovoltaic generation. Wind power and photovoltaic generation always insert in distribution network by the way of distribution generation, and changes a passive distribution network to an active distribution network. In this paper, artifical intelligent algorithms such as: genetic algorithm、Q-Learning algorithm、firefly algorithm and so on were used to slove the problems in active distribution network.Self-healing optimization in active distribution network considering distribution generation and reactive power compensation was researched. With the development of distribution network, distribution generationand reactive power compensation were introduced in active distribution network and their power flow model was builded. The model, whose objective is to minimize the power loss, could not only heal fault but also minimize power loss in active distribution network. Fault was healed by network reconfiguration, and genetic algorithm was employed to reconfigure the network topology. To meet the demand of the self-healing reconfiguration problem, genetic algorithm was improved as follows: hybrid coding strategy that adopts decimal code and binary code; special chromosome design according to the characters of loop network; special initialization, crossover and mutation operation were used to improve efficiency of the algorithm; crossover and mutation probability were optimized by cloud model to improve the convergence quality of genetic algorithm. Simulations validated the validity of the proposed model and algorithm.Power dispatch optimization considering intermittent clean power resource reliability in active distribution network was researched. Wind generation, photovoltaic array were introduced in active distribution network, to take into account the effect of intermittent clean power resource in power dispatch, dispatch model that includs wind power, photovoltaic generation, coal-fired units and mutative load were built. The dispatch model formed penalty function and satisfaction index to measure the realibility of wind power and photovoltaic generation. The multi-objective consisted of economic objective and emission objective. On algorithm, Q-Learning algorithm, which improves the online learning capacity of the system, was used to get the Pareto front. Then analysis method based on fuzzy was employed to find the reference solution from the Pareto front. Simulations testified the effectiveness of the proposed model and algorithm, and the realation between satisfaction index and reliability was clarified.Power dispatch optimization with energy storage system in active distribution network was researched. The inherent intermittency and volatility of wind generation caused the great impact in active distribution network. To relieve the impact, energy storage system was introduced in active distribution network. Dispatch model included wind generation, energy storage system and cyclic loads. Net output power and gross output power of wind generation and energy storage system, were regarded as objective respectively. Firefly algorithm is a new optimization technique that possesses advantages of the fast-searching speed, the fewer adjustable parameters, and it was refined by elitist strategy to deal with the nonlinear coupled dispatch problem, which considering energy storage system. Simulations verified that the energy storage system could effectively relief the impact caused by wind power and smooth the gross output and the net output respectively.