Research of cuckoo search algorithm based SVM and its application on load forecasting

摘要

The arrival of power system big data era provides stable data foundation and reliable information support for power load forecasting. In this context, more reasonable and relative train data can be obtained to train the SVM model. A new heuristic algorithm named cuckoo search is applied to optimize parameters of SVM since penalty factor (C) and kernel parameter (δ) have a great relationship with the performance of SVM. Test results of Jiangsu power grid show the accuracy of proposed multi-scale load forecasting model and performance of optimized SVM algorithm.