基于支持向量机和模拟退火算法对供暖热负荷的预测方法

Thermal Network Load Forecasting Based on SVM Algorithm and Simulated Annealing

准确的热网负荷预测能够提高热网的经济效益。本文对支持向量机模型进行了分析,对于支持向量机预测时需要人为确定参数这一缺点提出了利用模拟退火算法来自动优化参数这一方法。实例验证的结果表明,本文所提出的方法对参数的选取更有效,降低了单纯支持向量机的预测误差,与默认参数的支持向量机模型进行对比,该方法有效地提高了负荷预测精度。

Accurate thermal network load forecasting can improve the economic efficiency of the heating network. The model of support vector machine（SVM）is analyzed, in order to overcome the shortcoming of the SVM, its parameters need to be determined by experience. A new method is proposed that the simulated annealing algorithm is applied to optimize the parameters. An example can prove that this method is more efficient in selecting the parameters. It can reduce the testing error of the simple support vector machine （SVM） model. Compared with the default parameters, the accuracy of the load forecasting is improved effectively.