The Design And Implementation Of Electric Load Forecasting And Analyzing System For Meishan Power Grid

Load forecasting of the electric power system is very necessary, and availablecapacity of generators within the system, under normal operating conditions, should beable to meet the requirements of the system load at any time. Load forecasting can beused to determine the most economical way to meet safety requirements, operatingconstraints, as well as the Operation scheme limited by natural, environmental reasons.The only way to make sure that the power system is operating economically is to usereal-time load forecasting to arrange the generation capacity and transmission mode.The load forecasting is not only essential to the plan of capacity increasing, but also thekey information to raise funds for construction and purchase of equipment correctly. Forany of the electricity department, the accuracy of load forecasting is always crucial,because it determines the running time of most of the devices within the system.Appropriate load forecast is closely related to construction investment, and it involves ifthe investment is correct and whether the investment can be recovered in a timelymanner, and to obtain greater economic benefits. With the commercialization andmarketization of power systems, the accuracy of power load forecasting has greatsignificant meaning to the security and economic operation of power system securityand development of national economy. The level of power load forecasting has becomeone of the significant signs of the modernization of a power enterprise management.The main works of this paper are as follows:1) This paper gives an overview of the characteristics of the power system load, aswell as short term load forecasting research status and problems. A variety of traditionalmethods and modern methods of power system short-term load forecasting are reviewed.After presenting the constitution and classification of Power system load, the paper listseveral different treatments to correct different types of bad historical data, and somedata preprocessing needed in load forecasting. And some prediction error analysisindicators are given in this paper.2) The paper introduces the history of the development of the wavelettransformation, continuous wavelet transformation and discrete wavelet transformation. Then it specifies the basic concept and method of wavelet multi-resolution analysis, andanalyzes the reason why wavelet transformation can be used in load forecasting.Moreover, this paper elaborates time series models in load forecasting. Finally thewavelet decomposition and time series models are combined to forecast power load.Finally, the effectiveness of this method can be seen through simulation. This result andthe result of BP neural network and time series model are compared.3) Detailed introduction of the development of artificial neural network and BPneural network load forecasting used in this article is presented. The nodes number ofBP neural network hidden layer, as well as some methods and guidelines to selectnetwork parameters is given. Then, a simulation is performed, and the validation of BPneural network is analyzed. Finally, wavelet transformation is used in BP neuralnetwork to reduce the noise in training data. And the simulation result confirms itssignificance in improving the forecasting accuracy.4) Design and implementation of the Meishan load forecasting and analysis system.First of all, the paper analyzes the data requirements and feature requirements. And thispaper gives a detailed introduction of the system design which includes systemarchitecture, overall architecture and network architecture. Then the feature design ofthe system satisfies the needs of software users including forecasing, analyzing,inquiring and counting. Similarly, Oracle11g is chosen to serve as the database and theinterface problem of the system is solved. After spesifying the developing environmentand running environment, the main interface, inquiring interface, error contrast interfaceand waveform contrast interface is introduced. At last, a system test is performed andthe forecasting results are then examed. The forecasting results meet the accuracyrequirement.