Research On Data Driven Analysis Model Of Distribution Network With Distributed Generation And Its Application

The distribution network is the most terminal part of the entire power system.The connected transmission network and power supply users are closely related to work activities.With the continuous development and upgrading of the power system,the working capacity of the power system is also expanding.The distribution network structure is also increasingly complex,and the work mode is diversified.Many problems are also raised for the daily work adjustment,planning and design of the distribution network power flow design.Therefore,it is necessary to scientifically excavate and study the large amount of information collected and saved in the power flow measurement side of distribution network for a long time,so as to achieve the goal of improving the working mode of power flow design of distribution network,enhancing power supply stability and reducing network losses.The realization of observation and fine modeling on the distribution network side is an effective guarantee for the safe,reliable and economic operation of the power system.This paper mainly studies the distribution network power flow analysis model based on the real data obtained by the distribution system and the data-driven method.It is mainly divided into three parts.First of all,the first part of this paper proposes a generation model of load,photovoltaic and wind power curves based on generation of confrontation network,aiming at the limited scale of data available for distribution network.This paper briefly introduces the basic principle of generating confrontation network,improves the discriminator and generator according to the load,photovoltaic and wind power characteristics of the distribution network,selects a reasonable loss function,establishes and trains the neural network structure suitable for the load power,photovoltaic power and wind power characteristics of the distribution network,and explains the idea and process of generating confrontation network applied to the generation of distribution network load curve and new energy power curve.The indicators used to evaluate the quality of generated data are proposed,and the data generated by the method of generating data are respectively verified to have good consistency in the contour,power probability density function curve and spatial correlation.Combined with the actual data,the validity and feasibility of the generation model of load new energy power curve for the generation of countermeasures network are verified.Then in the second part of the study,considering that the change of low voltage line resistance is greatly affected by the change of the surrounding environment,an electrothermal coupling model is established to calculate the line resistance accurately.However,in view of the limited measurement level of the low voltage distribution network measurement on the environmental conditions,there are some areas that can not effectively obtain the environmental operating parameters.Based on this,A method based on nonlinear regression is proposed to identify the line parameters of low-voltage distribution network.This method can be used to identify the line conditions in the distribution network.Finally,the parameters of the distribution network in the actual WF area are identified,which provides an effective method for identifying the parameters of multiple lines in the distribution network,and provides an effective support for updating the line parameters in the power flow calculation in the following text.Finally,in the third part of this paper,a multi-objective optimization model considering the dynamic characteristics of the distribution network is established for the establishment of the distribution network,and a solution to the problem of distributed new energy capacity allocation in the distribution network is proposed.The problem is solved by the improved particle swarm optimization algorithm.First,the daily load curve and the daily output of new energy are extracted from the data generated in the second chapter through the Monte Carlo method,Then the distribution network line parameters are updated by the calculation method in Chapter 3,and then the updated data is calculated by the forward and backward generation power flow to obtain the objective function required by particle swarm optimization,and then the calculated objective is optimized and updated by changing the corresponding capacity of each particle through the improved particle swarm algorithm,and finally a solution set of three objectives is obtained.The rationality of this method is determined by comparing the voltage distribution before and after optimization by analyzing the calculation results of the actual power grid in WF area of SD Province,IEEE33-bus standard calculation example and IEEE69bus standard calculation example.An effective solution is proposed for the capacity allocation problem faced by the new energy access of the distribution network,which is conducive to the planner’s reasonable planning of the new energy access capacity of the distribution network and the scientific nature of the distribution network planning.This method can effectively reduce the network loss,improve the voltage stability,and provide simulation verification for energy conservation and loss reduction and improve the voltage qualification rate of the distribution side.