Process monitoring and abnormal reason tracing of coal mill based on instance migration

Abstract

Coal mill is the main equipment in the industrial pulverizing system. It is the key to ensure the stable and efficient operation of the pulverizing system to effectively use the production operation data to establish the process monitoring model and timely investigate the abnormal cause variables. In view of the problem that it is difficult to establish the process monitoring model for the newly put into use target domain coal mill due to the insufficient operation data, and combined with the advantages of kernel principal component analysis in non-linear industrial process monitoring, this paper proposes an instance migration-based coal mill process monitoring and abnormal cause variable tracing method. Firstly, multiple similar off-site coal mills are selected as source domain coal mills. By minimizing the mean difference and variance difference between source domain and target domain process data in high dimensional feature space, the source domain samples are weighted to match the distribution of source domain and target domain. Then, the kernel principal component analysis monitoring model is established based on the process data of source domain and target domain; the model is applied to monitor the samples and the monitoring results are obtained. Finally, the contribution graph method is used to find out the abnormal cause variables. The proposed method is applied to the actual coal mill production process, and the results verify the effectiveness of the proposed method.