Design And Implementation Of Monitoring And Analysis System For Control Characteristics Of Power Grid Load And Frequency

With the deepening of the digital transformation process of national power grid,the accumulation of power grid data is becoming more and more mature,which makes the analysis,application and corresponding technology upgrading of power grid big data become a trend.How to get valuable information from massive grid data and apply it to the future grid production is the main problem facing the grid development.In this paper,the monitoring and analysis system of power grid frequency control characteristics is designed based on the frequency prediction after primary frequency modulation.Using the historical data of power grid as the data set to train the long-term and short-term memory neural network to realize the frequency prediction model of primary frequency modulation.The model improves the accuracy of frequency prediction,ensures the real-time prediction,and provides reliable decision support when the power grid responds to risks.The main contents of this paper are as follows:First,based on the historical data,the frequency prediction model and short-term load prediction model are established.The short-term frequency prediction after primary frequency modulation is a core function of the system.The primary frequency prediction model is constructed by using long-term memory neural network.It is developed as an independent module in Python platform to provide frequency prediction service for other modules of the system,so as to facilitate algorithm implementation and other module calls.Finally,the original data is divided into training set,test set and verification set.The training set is used to train the frequency prediction model,and the test set is used to measure the prediction accuracy,prediction time,convergence degree and other indicators of the model to determine whether to continue the model iteration.When the model evaluation fails,the verification set is used to adjust the super parameters of the model to get a relatively better prediction model.Through the above process,the frequency prediction model of primary frequency modulation is successfully constructed,and the test set is used to verify that the model can better reflect the frequency change trend after primary frequency modulation.On the basis of the frequency prediction model,combined with Bayesian optimization algorithm,optimize the LSTM network model,get a set of optimal model parameters,use the grid load data combined with the parameter set training model,realize short-term load prediction.Secondly,the design and implementation of the monitoring and analysis system of power grid frequency control characteristics.This system uses Java programming language and spring Boot development framework is developed and implemented.The system includes real-time state monitoring of the whole network,real-time state monitoring of sub areas,future state T+15 minutes and T+60 minutes,anti-interference ability prediction of the whole network,performance evaluation of primary frequency modulation and other main functions.On the premise of defining the overall goal of system development,the detailed demand analysis of each functional module of the system is carried out,the flow chart of each module is designed to sort out the interaction between modules,the entity relationship diagram is constructed according to the data items involved in the system,and the database table is designed.According to the system design,the system function is realized,the class diagram and sequence diagram are used to show the process of system realization,and the test case table is used to test the function of the system,finally the realization effect diagram of the system is shown.In the system implementation process,the restful architecture style is used to access the front and back end requests,and the token mechanism is used to ensure the security of the system.Finally,the system has been put into use in a power grid company,which can well meet the needs of users,improve the prediction accuracy of short-term frequency after a disturbance of the power grid,and meet the real-time performance when the system processes the real-time data of the power grid for state monitoring.Making the historical data of power grid play a role is a successful exploration for the future development of smart grid,which provides decision support and risk early warning for the field of power grid automation.