Research On Distributed State Estimation Of Active Distribution Network With Distributed Generation

The large amount of distributed generation(DG)penetration is due to the continuous expansion of the traditional distribution network and the increasingly complex structure,which has become an active distribution network(ADN)with active control and management functions.Due to the complicated scale and structure of ADN and the uncertainty of DG’s output,the traditional centralized distribution network state estimation obviously cannot cope with this change.Failure to accurately estimate the operation state of the distribution network in time will result in the inability to timely respond to the distribution network.The real-time operation of the power distribution network is controlled,which seriously affects the safe,stable and economical operation of the distribution network.Therefore,it is necessary to establish an active distribution network distributed state estimation considering the uncertainty of distributed power output to improve the observation and control capabilities of ADN and ensure the safe,stable and economic operation of the distribution network.Aiming at how to solve the impact of the large scale,complex structure of the active distribution network and the uncertainty of DG output,this article mainly conducts research from the following aspects:First,the basic theory of active distribution network is introduced,including the basic structure and basic characteristics of active distribution network,and the mathematical model of active distribution network is described from the two perspectives of the measurement system in the active distribution network and the power network.Finally,based on the above theory,the mathematical model of traditional distribution network state estimation is analyzed,including weighted least square method and fast estimation method.Second,the basic principles of decoupling of active distribution networks are introduced.Based on the above principles,combined with the characteristics of active distribution network distributed state estimation,a multi-index two-step zoning method based on the extended subarea method is proposed,and the distribution method is introduced.The basic structure framework of power grid distributed state estimation.Third,in view of the current characteristics of distributed power access to the distribution network and the highly uncertain characteristics of the output of new energy power generation,the principles of photovoltaic and wind power generation and output influence factors are analyzed,and the output prediction model of distributed power and new energy power generation is established.And carry out the calculation example simulation.Finally,in view of the problems of insufficient utilization of big data in current centralized distribution network state estimation,limited calculation speed and insufficient calculation accuracy,a distributed state estimation method for active distribution network based on transfer learning convolutional neural network is proposed.And through multiple sets of simulation test analysis,proved the correctness and superiority of the proposed method.