论文部分内容阅读
根据多模型可以改善模型估计精度,提高泛化性的思想,提出了1种粗糙分类器的多模型软测量建模方法。该方法采用聚类、分类相结合的方式对数据进行分组训练,在一定程度上消除了矛盾样本点可能对模型精度造成的影响。对各组样本利用支持向量回归机建立回归子模型,得到多模型软测量系统。同时,通过向粗糙集引入相似度作为评价样本间相似性的指标,解决了传统粗糙集无法识别训练样本集中未出现过的模式的问题。通过引入概率测度,利用概率公式作为粗糙集分类的决策规则,简化了算法。基于上述理论构造的粗糙分类器,有效地提高了分类器的分类精度,确保了各子模型的估计精度。将该方法应用于双酚A生产过程的质量指标软测量建模,仿真结果表明了该算法的有效性。
According to the idea that multi-model can improve the accuracy of model estimation and improve generalization, a multi-model soft-sensing modeling method based on a rough classifier is proposed. The method uses clustering and classifying methods to group data training, to a certain extent, eliminating the impact of contradictory sample points on model accuracy. For each group of samples, a regression sub-model was built by using support vector regression to get a multi-model soft measurement system. At the same time, this paper solves the problem that the traditional rough set can not recognize the patterns that have not appeared in the training sample set by introducing similarity to the rough set as an index to evaluate the similarity between the samples. By introducing the probability measure and using probability formula as the decision rule of rough set classification, the algorithm is simplified. The rough classifier constructed based on the above theory effectively improves the classification accuracy of the classifier and ensures the estimation accuracy of each sub-model. The method was applied to the soft index modeling of the quality index of BPA production process. The simulation results show the effectiveness of the algorithm.