Analysis And Research On Short-term Forecast Of Power System Load

Short-term forecasting of the electrical power load is an important research topic in the power system dispatching and system state monitoring.The fast and accurate load forecasting is really important to ensure the security,the reliability and the economy in the power system.Support Vector Machine(SVDvM)is a common method of data prediction,but there are some problems in parameter selection which need to be solved urgently.In this thesis,by analyzing the factors affecting pow'er load,we establishes a preprocessing model of load sequence,and develops a support vector regression(SVR)forecasting model with the parameter optimization based on cuckoo search(CS)algorithm and realizes the short-term load forecasting in Hanzhong City,Shaanxi Province.In this thesis,aiming at the random noise interference in power load series,we establishes a noise suppression model based on the VMD decomposition;secondly,considering the two directions of neighborhood and similar day,we realizes the repair of missing data based on AR regression model;using t-test criterion,we regards the abnormal data as the gross error and achieves the detection and correction of the abnormal data.Simulation and experimental results verify the effectiveness of the preprocessing model.By considering the temperature,humidity,weather,seasonal factors,holidays,near-point loads and similar daily loads,the SVR short-term load forecasting model is established.The experimental results show that the penalty factor and the kernel parameters of the radial basis function are correlated,which have great influence on the prediction performance,and the optimal kernel parameters of the different data are quite different.Aiming at the disadvantage of traditional CS algorithm which is sensitive to the initial value,combined with the bilinear search algorithm,an improved CS algorithm is established to optimize the parameters of SVR.Experiments on power load forecasting in different seasons and dates show that,the improved CS algorithm not only improves the convergence speed of the algorithm,but also improves the accuracy of the forecasting model.The research results have certain scientific significance and practical value for improving the accuracy and reliability of short-term load forecasting.