Research On Short Term Power Load Forecasting Method Based On Data Mining Technology In Nanjing Area

Power load forecasting is one of the important work of the power sector, accurate load forecast, can be economical and reasonable arrangements for the internal power generating unit start stop, to maintain security and stability of power grid operation, reduce unnecessary spinning reserve capacity, unit maintenance scheduling for reasonable arrangements to ensure the social normal production and life, effectively reduce the cost of power generation, increase the economic benefit and social benefit. The results of load forecasting, but also can help to determine the future of the new generation unit installation, decided to size, location and time of installed capacity, determine the construction and development of power grid.In this paper, based on data mining technique for clustering analysis technology, using k-means clustering analysis method is applied to clustering analysis of historical load data, the initial clustering number is set to 2, good clustering results with the heavy load of working day and holiday type counterpart. The results of cluster analysis based on the detection of abnormal data on actual load data to detect the abnormal data by using the grey theory GM(1, 1) model to achieve the correction of abnormal data. The meteorological department of meteorological data query response date, the meteorological data and the characteristic parameters for the clustering step 6 before pushing the historical load values and the corresponding neurons as input, neural network analysis. Finally realize the forecast of the whole point load. The effectiveness of the method is proved by the results of the actual load forecasting in Jiangning District of Nanjing.