论文部分内容阅读
针对水泥熟料质量指标游离氧化钙含量难以实时检测的问题,提出一种基于局部pso-lssvm的软测量建模方法。在构建局部建模数据集时,同时考虑了数据样本之间的加权欧氏距离与向量的夹角,使得训练数据的选取更加具有实际意义。由于局部建模的时间开销较大,首先,采用加权k均值聚类算法对历史数据库进行聚类分析得到若干子样本集;其次,在与当前输入数据加权欧氏距离最小的子样本集中,建立基于pso-lssvm算法的局部软测量模型,计算得到当前游离氧化钙含量值;最后,应用实际数据进行训练和验证,结果表明,该方法较全局建模具有更好的泛化能力,能够满足水泥熟料游离氧化钙含量检测的实时性要求,对于实现水泥烧成系统的优化控制,提高能源利用率具有重要意义。
In view of the difficulty of real-time detection of free calcium oxide content in cement clinker quality index, a soft-sensing modeling method based on local pso-lssvm is proposed. When constructing the local modeling dataset, the included angle between the weighted Euclidean distance and the vector is taken into account simultaneously, which makes the selection of training data more practical. Due to the large time cost of local modeling, firstly, a weighted k-means clustering algorithm is used to cluster the historical database to obtain a number of sub-sample sets. Secondly, in the sub-sample set with the weighted Euclidean distance to the current input data, Based on the local soft sensing model of pso-lssvm algorithm, the current free CaO content was calculated. Finally, the actual data were used for training and verification. The results show that this method has better generalization ability than the global modeling, The requirement of real-time detection of free calcium oxide content in clinker is of great significance for optimizing the control of cement burning system and improving the energy utilization rate.