With the rapid growth of industrial equipment monitoring data, ensuring data security and efficient classification has become a critical challenge. This paper proposes a novel data security monitoring model that integrates sensor data classification and a grading strategy with Eclipse Dataspace Components (EDC) technology, ensuring the trusted transmission of monitoring data. Taking the cement roller press as a case study, a convolutional neural network (CNN) is employed for local fault diagnosis, and meanwhile a transfer learning method is used for heterogeneous fault diagnosis. Experimental results demonstrate that the proposed model significantly enhances the processing efficiency of production data and improves fault diagnosis accuracy. This study provides a practical solution for industrial equipment data security management and fault monitoring, and shows high potential for real-world applications and engineering value.