在热工过程建模中,所建模型精度与过程数据精度密切相关,当现场数据中存在坏值时,将严重降低过程模型精度.而且数据库中大量的冗余数据增加建模成本而不能提高模型精度.为此,应用拟合LOWESS曲线方法进行数据处理,剔除数据坏值,然后应用模糊聚类方法精简原始样本,减少冗余数据.试验结果证明了改进方法的有效性.%The model accuracy is closely related to the process data accuracy in the thermal process modeling.The paper applied the method of fitting LOWESS curve for data processing,eliminated the bad data values,then applied the method of fuzzy clustering to stream line the original sample and reduced redundant data.The test results demonstrate the effectiveness of the proposed method.
展开▼