Microgrid Load Forecasting Research Based On Improved Support Vector Machine

Microgrid load forecasting is an important part of microgrid power system, impacting on the stability of the microgrid operation and economy, it is the premise of distributed energy intelligent management and the basis for improving the load optimization of micro grid in the user side. However, due to the microgrid load has the characteristics of simple structure, strong volatility, nonlinear load data, the load factors are difficult to count etc. Therefore, the main research content of this paper is microgrid load forecasting based on improved support vector machine algorithm.Firstly, in this paper, the historical load data of the microgrid community is chosen as the main body to analyze microgram load forecasting, Multidimensional data variables are constructed based on weather factors, the types of the week, real-time price, and other external factors. In the process of data acquisition and transmission will exist noise data and affect the accuracy of load forecasting, so for this problem, the wavelet threshold method is proposed to carry out data denoising. Through the matlab simulation analysis, the wavelet denoising method can not only remove the most useless information from the load data, but also retain the integrity of the load characteristics as far as possible. Secondly, in the processing of abnormal data, the fuzzy C mean clustering algorithm is proposed to extract the characteristic daily load curve, which is used to identify and correct the abnormal data.Moreover, in view of the many factors that affect the microgrid load, the dimension is difficult to choose, the principle of principal component analysis is proposed to solve the problem, the individual contribution rate and cumulative contribution rate are calculated by extracting the characteristic function of each dimension variable, according to these two indicators to determine the dimension of variables and to achieve the purpose of dimension reduction.Finally, the kernel function of support vector machine is difficult to be selected, so the radial basis function is used as the kernel function by the control variable method. In the process of load forecasting, the SVM needs to adjust the operating parameters, and the optimal operating parameters can play a decisive role in improving the accuracy of the load forecasting. To solve this problem, an improved genetic algorithm is proposed to optimize the parameters of support vector machine, parameter optimization is improved in local search ability and global search ability, avoid falling into local extreme value, and then improve the precision and generalization ability of the support vector machine in the microgrid load.