论文部分内容阅读
针对氧化铝生产过程中原矿浆制备工序碳酸碱浓度机理模型难以建立,无法实现在线检测的问题,通过对铝酸钠溶液温度和电导率的特性分析,提出一种基于同步聚类和稳定学习的碳酸碱浓度模糊建模方法。首先采用同步聚类算法将建模数据分类,并对聚类中心及分类进行离线修正,然后对每类数据进行模糊TS(Takagi-Sugeno)建模,并采用稳定学习算法对模型参数进行在线校正,保证辨识误差有界。将该方法应用于氧化铝生产过程碳酸碱浓度的软测量,现场实际数据预测结果表明了方法的有效性。
Aiming at the problem that the model of carbonate concentration mechanism in the process of ore slurry preparation in alumina production is difficult to set up and online detection can not be achieved, a temperature and conductivity characterization of sodium aluminate solution is proposed. Based on synchronous clustering and steady learning, Alkali concentration fuzzy modeling method. First of all, we use the synchronous clustering algorithm to classify the modeling data and make offline correction to the clustering center and classification. Then we model Takagi-Sugeno for each type of data, and apply the stable learning algorithm to online calibrate the model parameters , To ensure that the identification error is bounded. The method was applied to the soft carbonaceous concentration measurement in the alumina production process. The actual field prediction results show the effectiveness of the method.