Short Term Load Forecasting of Offshore Oil Field Microgrids Based on DA-SVM

Abstract

Abstract Prediction accuracy is a basic indicator for short-term load forecasting, which is particularly crucial for the microgrid of offshore oilfield groups. A method of support vector machine based on the dragonfly algorithm (DA-SVM) is proposed to predict the short-term load of the microgrid in an offshore oil field. This method combines the penalty factor and kernel function of support vector machine as the solution position of dragonfly. The prediction accuracy of the algorithm is employed as the current fitness value of dragonfly. The optimal location of the dragonfly is the optimal parameters of the support vector machine. The DA-SVM algorithm was used to predict the short-term load of an offshore oil field microgrid in the Bohai sea, China, and is compared with the prediction of PSO-SVM, GA-SVM and BPNN models. The results show that the DA-SVM algorithm has more straightforward steps, better global search ability, higher prediction accuracy and better computing speed.