Genetic Programming On Power Short-term Load Forecasting For The Application

In recent years, the load forecasting of technical research have been more and more, and achieved some practical effect. Load forecast is the basis of power system's operation and control. Accurate load forecasting will benefit for the grid internal generator starting and stopping in economic arrangement, reduce unnecessary spinning reserve capacity, reasonably arrange the overhaul plan, ensure the normal production and social life etc. From the traditional predictive technology to today's artificial intelligence technology, various load forecasting technology have emerged.Conventional short-term load forecasting technology mainly includes exponential smoothing prediction method, regression model prediction technology, time series prediction technology and gray prediction technology, etc. Intelligent prediction technology mainly includes expert system, artificial neural network prediction technology, wavelet analysis prediction technology and fuzzy logic system, etc. The current load forecasting majority is aimed at load and temperature, wind, humidity, the relationship between individual meteorological factors that can not correctly reflect the defect is all weather information, easy to cause the prediction results appear error. With the development of economy and the improvement of people's living standard, the summer cooling load and winter heating load the proportion of more and more biger, and is also causes the seasonal is one of the main reasons for lack of electricity. And the feeling of mankind to external environment is not just by a single meteorological factors such as temperature or humidity affect to determine, meaning that the human body on the outside of feeling is the comprehensive effect factors of the meteorological results, which puts forward the comprehensive meteorological index -- human comfort index. Human body is measurable index of temperature, humidity, wind speed, the comprehensive reflection of meteorological factors in high summer's temperature, thermal comfort degree of different will directly influence the air-conditioning cooling and refrigeration equipment, thus affecting the use of how much power load usage. Papers to the comprehensive consideration of the temperature, humidity, wind speed of meteorological index factors such as comprehensive as a starting point, and with it a comprehensive meteorological index model for variables.Paper uses genetic programming as a method of forecasting model of schema, and its essence is the level of generalized computer program description of the problem, this computer program according to environmental conditions and size of the dynamic changes in its structure. Articles are based on genetic programming of short-term power load characteristics of two different input variables forecast model: one is the use of direct meteorological factors as variables, including temperature, humidity and wind speed as input modeling three meteorological factors, and the other species of human comfort index is used as the feature variables to model. In addition, the two models and compare the prediction error, the experiment proved that the introduction of human comfort in the daily load forecasting model prediction accuracy is improved, and the program downsizing, reducing the possible evolution of space exploration, reduction in the time predicted, that is, the convergence rate of process has been improved.