Research On Topology Adaptive Identification Of Distribution Network Based On Data-driven

Accurate topology and line parameters are the basis for analysis methods such as power system state estimation,power flow calculation and stability control,which are prerequisites for grid planning,optimal operation and stability control.In the current topology identification methods,most algorithms focus only on the identification of the steady-state topology of the distribution network lacking the identification of line parameters and topology changes.Meanwhile,the line topology and parameters can change dynamically due to the influence of certain objective elements.Therefore,there is an immediate demand to propose a new method for adaptive identification of distribution network topology considering topology changes due to uncertainties.Based on the limited measurement data of the distribution network,the research work on the adaptive identification of distribution network topology is carried out as follows.First of all,the distribution network is abstracted into a network topology diagram based on graph theory principles.To address the problems of topological confusion and large errors of line parameters in distribution networks,in the case that voltage phase angle information of all nodes cannot be collected in some distribution networks.The power and voltage data available from the existing measurement equipment in the distribution network is analyzed and the physical system is mapped digitally to visualize the distribution network topology and line parameters.According to the tidal equation,a linear regression topology and line parameter identification model is established based on the measured data under multiple time sections,and the least squares method is used to solve the model for the derivative parameters.Considering the uncontrollable factors that may lead to missing data during the actual data acquisition process,a method is proposed to use minimum variance to fill in missing data to ensure the completeness of information.Secondly,a correction model is established,and an improved Newton-Raphson method is used to iteratively adjust the initial values of the topology and line parameters to refine the accuracy of the topology structure and line parameters.By identifying the parameters,a topology change detection model is constructed based on the phase angle data obtained.The topology dynamic perception model is constructed based on the phase angle data obtained from identifying parameters,which can quickly and easily detect whether the topology has changed or not.Finally,the feasibility and robustness of the method are verified using the IEEE 33-node test system.Secondly,to address the problem of topological confusion in the distribution station area,based on the principle of energy conservation between the station area and the users.Using the collected energy time-series data of distribution network nodes,a combination of principal component analysis and convex optimization algorithm is used to extract the main features and achieve the goal of data dimensionality reduction.Establish a topology identification model for the low-voltage distribution substation,the regression matrix is solved by the interior point method to realize the identification of the relationship between users and the distribution substation ownership.Two-and multi-layer models are built to verify the feasibility of the algorithm,and the simulation results show that the proposed optimized algorithm has higher recognition speed and correct rate compared with the traditional method.Finally,the practical engineering application of data-driven distribution network topology identification is based on the specific needs of the topology identification platform.Through the distribution network real concrete case,from the network architecture and hardware composition,data collection,analysis,and other aspects of the introduction,to achieve the correct identification of topology results.The applicability and feasibility of the method in distribution networks of different scales are verified,and proves the topology adaptive identification method has some engineering application.