Study On Load Forecasting Technology Based On User Behavior Pattern With Forward Analysis

With the rapid development of intelligent communication technology(ICT) and the widely applications of intelligent terminal equipment in smart grid, the collecting datastored in the power system become one of the big data actually. How to transform the big data into the information for the improvement of the power system operation is very interesting problem for the power engineers. Different with the conventional one, the load forecasting in smart grid has been changed from the simple technology to the precision service management problem. The load forecasting demands more accurateand higher density results. It increase the difficulty of load forecasting. Load forecasting combined with big data technology is a big challenge for the smart grid, which is also the novel trend and demand.The new information platform and ICT are applied into the load forecasting in the thesis. A novel load forecasting approach based on user behavior pattern with forward analysis is present here. The feature library of user behavior pattern are established for different industries with ICT. Then the dynamic clustering are used todistinguish the types of the customers. With the type of customer and the behavior pattern, the load forecasting is realized finally, which is called forward analysis. This approach do not only counton the historical load data of the customers, but also on the relevance between different type of data, which is one core idea of big data technology. With cloudy platform, the load forecasting system is developed for Yangjiang grid. The core technologies include the cloud computation and mass data storage. By the application in two customers, the effectiveness of the approach is verified. At the same time, the load forecasting result on the bus in the substation can also obtained by the system, which is very important for the planning and operation in the power system.