Development Of Power Load Analysis And Forecast System Based On Agent

The paper develops a system using multi-agent theory that analysis the relevance between power load and weather factors and then predict daily peak load. Firstly, the paper introduce the analysis thought and method, mainly including error data elimination using wavelet denoising theory, load natural growth rate elimination, weather factors quantification using meteorological indexes, weather thermal incubation effect elimination, grey correlation analysis between load and weather, load-weather scatter fitting and sensitivity analysis using cubic liner regression, load forecasting using support vector machine and etc. After making sure of the analysis thought, the paper designs the system structure with multi-agent thought including structure, cooperation, transmission, package and so on. The paper uses data warehouse to store analysis data, Visual Basic to realize the system. Finally, the paper analysis the results of relevance, sensitivity and forecast produced by the software.