Source Load Probability Distribution And Matching Method For Power Field

In recent years,with the continuous development of new energy industry,analyzing the characteristics of new energy will be conducive to the stability of power grid and economic scheduling,so modeling the uncertainty of new energy becomes more and more important.The existing methods build the probability model of wind power output based on the real load data by randomly generating the scene,and then generate the scene by sampling,but the accuracy of the model is not high and the calculation complexity is high.The accuracy of the model is poor and the calculation complexity is high.Therefore,a method of generating wind power output random scene based on conditional variational automatic encoder is proposed.Therefore,a probability load distribution method based on conditional variable auto encoding is proposed.Compared with the existing probabilistic modeling method,the method in this paper can learn the characteristics of wind load data without supervision,and generate data that accord with observation characteristics effectively,without complicated scene simplification.At the same time,in order to improve the absorption capacity of new energy,it is necessary to match the value between the source and load from the perspective of power grid planning,so as to achieve the local and nearby absorption capacity of new energy.Firstly,this paper describes the significance and background of the research on source load in the field of power,and introduces the current research situation at home and abroad.Secondly,the basic knowledge,neural network and its classification are summarized,and the forward propagation and back propagation algorithms of neural network are introduced in detail.This paper introduces the process and idea from self coding to variational self coding,and expounds the calculation process of variational inference and the selection of loss function.Then this paper mainly introduces the establishment of the source load probability distribution model based on the variational self coding.By using the deep learning model variational self coding,the probability model is built based on the measured wind power data.This method can learn thecharacteristics of wind power data unsupervised,and can generate new data that meet the observation characteristics according to the conditions effectively.Moreover,it does not need the process of scene reduction,which greatly improves the efficiency and quality of scene generation.Next,in view of many uncertain factors(such as weather,wind speed,etc.),this paper proposes a source charge value matching method,which is to match the source charge value by calculating the matching degree,to improve the nearby wind power consumption capacity of new energy,so as to achieve the purpose of improving the utilization rate of new energy.Six schemes are designed and tested to verify the reliability and practicability of the method.In this paper,it is measured by the size of the matching degree.The bigger the matching degree is,the better.The scheme with the larger matching degree is the best scheme for nearby wind power consumption.Combined with the theoretical basis of the previous chapters,this paper uses Python and other technologies to build a Django based new energy and multi load value matching software system,and at the end of the paper,summarizes and prospects the work of this paper in the source load probability distribution and matching methods.The experimental results show that the theoretical method proposed in this paper is effective and feasible.It is helpful for the analysis of new energy and load to some extent,and provides a new idea and experience for the future research in this field.