Distributed dictionary learning for industrial process monitoring with big data

Abstract

With the development of sensor and communication technology, industrial systems have accumulated a large amount of data. This data has provided new perspectives and methods for industrial system analysis, monitoring and control, which is proven to be of great significance. However, with the collection and storage of industrial data in a 7 × 24 manner, the computing and information processing capabilities of edge controllers and computers at industrial sites face new challenges. Therefore, this paper proposes a distributed dictionary learning algorithm based on the MapReduce framework. The dictionary learning method can efficiently extract useful information from high-dimensional data for process monitoring. In addition, deploying the algorithm under the MapReduce framework can achieve the purpose of parallel distributed computing, which would solve the issue that the ability of calculation and information processing is limited at industrial sites. Based on extensive numerical experiments, the proposed method can improve the effectiveness and robustness of process monitoring for industrial processes.