Support vector machine (SVM) is a statistical learning method based on structural risk minimization principle. Compared with traditional neural network, the generalization error rate is low and the result is easy to explain. When SVM is applied to load forecasting, the inaccurate selection of parameters will result in poor prediction performance. A short-term load forecasting method of support vector machine based on grasshopper optimization algorithm is proposed. SVM is trained with historical data such as load and weather in a certain area. The parameters of support vector machine are selected by GOA, and then the GOA-SVM load forecasting model is established with the optimal parameters obtained. The example analysis shows that the GOA-SVM model has better convergence performance and higher prediction accuracy than GA-SVM and PSO-SVM models.