一种基于扩散映射的化工过程IWO-FCM数据挖掘算法

An IWO-FCM data mining algorithm of chemical industrial process based on diffusion mapping

针对实际化工过程数据具有高维、非线性等特征而难以进行聚类分析的问题,提出一种基于扩散映射的IWO-FCM算法.该算法先利用扩散映射提取高维数据的低维流形特征,整合数据的局部特征使原始数据的几何信息得以保留,然后用IWO-FCM算法对低维流形数据进行聚类分析.试验通过对TE过程多个故障数据集进行测试,与获取数据低维特征后使用FCM算法相比,结果表明,IWO-FCM算法具有较强的稳定性和鲁棒性,比FCM算法具有更强的寻优能力和更好的收敛效果,聚类效果明显改善,能够快速有效地识别故障特征,验证了其有效性和优越性.

Because there is difficulty in cluster analysis of practical chemical industrial process data with high dimension and nonlinearity, an IWO-FCM data mining algorithm was proposed based on diffusion mapping. In this algorithm, firstly the diffusion mapping was used to extract low-dimensional manifold characteristics from high-dimensional data, and the local characteristics of data was conformed so that geometric information of original data was retained. Then the IWO-FCM algorithm was used in cluster analysis of the low-dimensional manifold data. TE process multiple fault data set test was performed and the experimental results demonstrated that the IWO-FCM algorithm would have stronger stability and robustness and better optimal ability and convergence effect than FCM algorithm with extracted low-dimensional manifold characteristics. The clustering effect was improved obviously, and the proposed algorithm could be used to identify fault features quickly and effectively, confirming its validity and superiority.