Development And Application Of Short-Term Load Forecast Platform

Power grid short-term load forecasting is of great significance to scientifically arrange the power generation output of power plants,ensure the balance of supply and demand,and ensure the security and stability of power system.It is an urgent task for power grid enterprises to improve prediction accuracy and work efficiency.Under the premise of ensuring the security of data and information,it is imperative to develop and apply the power grid short-term load forecasting platform by applying new technologies and methods such as big data.The main contents of this paper include;1.Describes the basic method of short-term load forecasting.It includes common load forecasting methods and implementation steps,point-to-point multiple ratio method,multiple ratio smoothing method,overlapping curve method and other"normal day" load forecasting methods based on the same type of day.Considering the influence of temperature and holidays on load,the principle and implementation steps of artificial neural network(ANN)and chaos theory are analyzed respectively.On this basis,the structure of convolutional neural network(CNN)in deep learning method is described,and its implementation in MATLAB is given.2.Combined with the new trend of power big data mining and application,explore the application of big data technology in power grid short-term load forecasting-Based on the analysis of MapReduce data processing logic,a time-sharing distributed short-term load forecasting method based on Hadoop is proposed.Based on Hadoop big data architecture,a load forecasting model based on big data is built.The city company is the main node to identify the leading factors.The county company is taken as the slave node.The load pattern library is established by K-means clustering.The random forest algorithm is used for pattern matching to carry out local load forecasting.Then,the data is centralized through MapReduce process to realize time-sharing distributed negative Dutch prediction.3.Based on the actual demand of load forecasting by the regulatory departments of local power supply companies,a short-term load forecasting platform for power grid is developed.In order to ensure the information security in the process of data transmission,this paper proposes to use Paillier homomorphic encryption algorithm suitable for big data and cloud computing to protect the privacy of power grid load data.On this basis,various load forecasting methods and load curve analysis are implemented in the platform.The platform is developed with Visual C++object-oriented programming language.The data storage and access rely on SQL Server database.The realization of prediction algorithm relies on MATLAB software.Through the interface technology of VC,SQL server and MATLAB,various functions are presented to users.A simple,practical and friendly operation interface is established to realize the integration of various functions of daily load forecasting in the system.The structure and functions of the platform are introduced in detail.In summary,based on Hadoop big data architecture,time-sharing distributed short-term load forecasting can effectively improve the accuracy of load forecasting considering the differences of regional characteristics;the Paillier homomorphic encryption algorithm is used to protect the privacy of power grid load data to effectively protect the data and information security of load forecasting under big data and cloud computing framework;the platform for short-term load forecasting of power grid is developed and applied,and Establish a friendly load forecasting and load curve analysis interface to achieve accurate and efficient short-term load forecasting.