Short-term Load Forecasting Based On Data Mining

Power system short-term load forecasting is one of the important works of thepower system, high precision of prediction need good prediction model and theprediction method to guarantee but also to consider all factors. In recent years,experts and scholars put forward many of the intelligent algorithm is applied to powersystem load forecasting example, there are also several intelligent algorithms used incombination, make obtained the rapid development of power system load forecasting,prediction accuracy has improved significantly. This paper introduces the theory ofdata in the power system load forecasting, the use of short-term load forecastingalgorithms in data mining structure model.Data mining theory is a relatively new concept, in recent years gradually appliedin the relevant aspects of power system, using the algorithms in data mining theory toanalyze and integrate load data, rather than using only a single method on the data ofload forecasting, so we can better is, for each algorithm has a better load forecastingalgorithm are obtained. In this paper, using multiple linear regression in data miningand neural network to forecast, also used the analytic hierarchy process (AHP), andwill also be very classic simulated annealing is introduced into the model, in thisarticle, when considering load forecasting factors not only consider the historical loaddata, also considered the effect of temperature, the innovation is considered thePM2.5, finally combining example verification, can be predicted comparison results.Fully meet the requirements, the results show that the prediction accuracy andprediction results contrast figure after each algorithm can be obtained by usingalgorithm of optimal prediction algorithm.