基于SVR算法的燃气轮机功率预测研究

Gas Turbine Power Prediction Based on Support Vector Regression

对燃气轮机未来几小时的功率预测是跳闸等故障预警的关键,而国内该方面研究尚少。采用支持向量回归模型,并融合多变量预测,以提高预测的准确性。以某电厂燃气轮机运转的实际数据为例,设计多组对比实验,详细阐述了实验流程以及重要参数的选取方法,最终验证了该方法的有效性。

Forecasting for the next few hours of the gas turbine power is the key to predict trips,while the domestic related researches are still few. Though using support vector regression（SVR） model, and integration of multivariate fore casting, it improves the accuracy of the forecasts. These experimental data from the real data of a power plant, through some comparative experiments, describes the experimental procedure and selection methods of important parameters in detail. The results verify the effectiveness of the support vector regression techniques applied to practical power prediction.