The Study And Software Development Of The Load Characteristics Indicator And Load Forecasting

The content of this thesis is one of the main components of a project of Shanxi Electric Power Company. This thesis was put forward for improving the security and economy of the power production and grid operation, arranging the power grid construction reasonably. In the period of the rapid development of electric power enterprises, the research and analysis for the load characteristic indicators is important for the stable operation and economic production. Accurate medium-and long-term forecasting is important for the power system development plan and the electric power plan.Firstly, the author analyzed and researched the load characteristic indicators systematically, and then developed a set of load characteristic indicators analysis and calculation software on the platform of Visual Basic and Access. In the software, the author used the Visual Basic to complete the man-machine interface and used the Access to manage the large number of load data (the24points load data was used in this thesis).By calculating the load characteristic indicators and drawing the load characteristic curve of a city of Shanxi in2010, we knew that the software is easy to use and have a friendly interface.Secondly, because of the medium-and long-term load forecasting using the less historical data and a single data will have a great impact on the forecasting result, the author take advantage of the GRNN’s strong approximation ability and fast learning speed feature to develop a set of software which make the Visual Basic and MATLAB connection seamlessly. The case study shows that compared to the traditional data processing methods, the neural network complement has the higher data accuracy.Thirdly, because of the medium-and long-term load forecasting was influenced by many factors (for example the economy factors, the policy factors etc.), he author analyzed the uncertainty forecasting model of the medium-and long-term load forecasting. The classical proportional coefficient method and regression analysis method were improved by the blind number theory in this thesis. And then a city’s load was forecasted by the proportional coefficient method which based on the blind number theory and its relative error is3.23%. The result shows that the uncertain model has the higher prediction accuracy. The GM(1,1) model and its improvement models were also studied in this thesis. And the software which was divided into yearly、quarterly and monthly software forecasting package and includes the traditional models、GM(1,1) model and the uncertain model which based on the blind number theory was developed.Finally, based on the medium-and long-term load forecasting, the combination forecasting models were studied in this thesis. The average weights combination forecasting model and the variance-covariance combination forecasting model were compiled into software. The fuzzy variable weight combination forecasting model was studied deeply. The examples show that the fuzzy variable weight combination forecasting has the higher accuracy than the traditional combination models.The load characteristic indicators calculation and analysis software and medium-and long-term load forecasting software were developed in this thesis. The software has already passed the examination and applied in a power company. Practical application shows that these two sets of software has a friend man-machine interface and better applicability.