模糊树鲁棒回归算法的研究及其应用

发布时间：2018-06-02 16:52

  本文选题：模糊树 + 局部异常因子　； 参考：《动力工程学报》2017年05期

【摘要】：针对实际工程中噪声难以避免和预测的问题,提出了鲁棒性较强的加权模糊树(W-FT)算法,采用基于局部异常因子(LOF)的加权最小二乘法代替最小二乘法学习模糊规则的后件参数,通过2个典型的非线性例子验证了该算法的有效性.应用W-FT算法建立了电站锅炉NOx排放特性模型,并与其他建模方法所建模型进行了对比.结果表明:所提出的W-FT算法能够有效地辨识噪声和异常值,具有较强的鲁棒性,所建立的模型预测精度较高,泛化能力较强.
[Abstract]:Aiming at the problem that the noise is difficult to be avoided and predicted in practical engineering, a robust weighted fuzzy tree algorithm is proposed. The weighted least square method based on local anomaly factor (LOF) is used instead of the least square method to learn the parameters of fuzzy rules. The effectiveness of the algorithm is verified by two typical nonlinear examples. The model of NOx emission characteristics of utility boiler is established by using W-FT algorithm and compared with other modeling methods. The results show that the proposed W-FT algorithm can identify noise and outliers effectively and has strong robustness. The proposed model has high prediction accuracy and strong generalization ability.
【作者单位】： 华北电力大学新能源电力系统国家重点实验室;华北电力大学工业过程测控新技术与系统北京市重点实验室;
【基金】：国家重点基础研究发展计划(973计划)资助项目(2012CB215203) 中央高校基本科研业务费专项资金资助项目(2015MS33)
【分类号】：TB53;TM621.2
，

本文编号：1969487