热声不稳定主动控制的SVM时间序列预测模型  

SVM Time Series Prediction Model for Active Control of Thermoacoustic Instability

作　　者：周昊[1] 丁芳[1] 黄燕[1] 

机构地区：[1]浙江大学能源清洁利用国家重点实验室,杭州310027

出　　处：《动力工程学报》2016年第3期178-184,共7页Journal of Chinese Society of Power Engineering

基　　金：国家重点基础研究发展计划资助项目(2015CB251501);国家自然科学基金资助项目(51476137)

摘　　要：针对目前热声主动控制方法中存在的滞后问题,提出采用支持向量机算法进行热声振动时间序列的模糊预测.在自行设计的Rijke管热声不稳定实验台上进行扬声器扰动实验,采集热声振动动态压力序列,利用相空间重构法则构建输入输出数据集,并采用支持向量回归机建立预测模型,用测试集进行校验,定性分析了嵌入维数和延迟时间对时间序列预测精度的影响.结果表明:所用方法能得到较高的预测精度,模型在较高的延迟时间下仍能保持决定系数为0.95的高拟合度,验证了该时间序列预测方法的可行性.To overcome the time delay problem of present thermoacoustic active control system, a method of fuzzily predicting the time series of thermoacoustic instability was proposed using support veetor ma- chine （SVM）. In the self-developed Rijke tube thermoacoustic instability test bench, the experiments of speaker disturbance were conducted to capture the dynamic pressure sequence under thermoacoustic oscilla- tion, and to build the input and output datasets using phase space reconstruction theory, following which a prediction model was established by SVM and verified with test datasets, and the effects of embedding di- mension and time delay on predietion accuracy of the model were qualitatively analyzed. Results show that via the model, high prediction accuracy can be achieved, and the coefficient of determination can get up to 0.95 even if the time delay is large, indicating high degree of fitting, proving the time series prediction method to be effeetive.

关 键 词：热声振动 RIJKE管 时间序列 支持向量机 预测 

分 类 号：TK16[动力工程及工程热物理—热能工程]