One Place Of Xinjiang Short Term Load Intelligent Forecasting Techniques

Short term load forecasting of power system is an integral part of the dispatching, and is a very important power sector’s daily work. It includes sure start stop unit, hydropower and thermal power of coordination, exchange of tie-line power, economic load distribution, reservoir for various operations such as scheduling and equipment maintenance on a regular basis to provide data, forecast accuracy level can directly affect the safety and economic behavior. Many factors affect the load forecasting precision, including time, weather, and other factors, mainly cyclical time factors and growth; main meteorological factors include weather temperature and air humidity; other factors mainly for special events (such as large gatherings, celebrations and other activities).This paper analyzes the short-term load forecasting of power system and the status of all kinds of traditional forecasting methods, in view of the traditional method of strong subjectivity, influence factor analysis is not thorough enough, proposes a method of empirical mode decomposition (EMD) and artificial neural network (ANN) method for short term load forecasting using analysis method.Using empirical mode decomposition (EMD) will load sequence automatic parse out the load variation characteristics of different intrinsic mode components (IMF), then on the various components are analyzed separately, accurately grasp the effect of each component of the factors, then uses the appropriate BP neural network prediction model, the model prediction results were fitted, draw the ultimate load forecasting results.Because (EMD) decomposition is based on the signal sequence of the local characteristic time scales, and thus the adaptability, there is a strong learning ability of BP neural network and adaptive capacity, handling load forecast of such nonlinear problems to have had certain advantages. Through analysis of real-time load data in a region, a combination of methods than a single method of BP neural network in prediction accuracy has a higher increase.