A Cyber-Physical monitoring and diagnosis scheme of energy consumption in Plant-Wide chemical processes

Abstract

In the large-scale plant-wide chemical process, energy monitoring and diagnosis have a great impact on energy management and sustainable development. Most monitoring and diagnosis methods focused on the construction of the global model with process data, regardless of the meaningful process knowledge and in-depth analysis. Due to the multi-dimensional, correlative, and uncertain characteristics of the collected industrial data, it is laborious to obtain accurate and reliable energy monitoring and diagnosis results. To address these problems, a novel cyber-physical energy monitoring and diagnosis scheme is proposed in this paper. This scheme constructs a cyber-physical model based on process knowledge and data to describe the cause-effect relationships among the energy variables. For energy monitoring, a distributed monitoring approach is developed for energy state estimation based on the constructed monitoring statistics in both variable and residual spaces. For energy diagnosis, a faulty contribution degree index is further addressed and the corresponding root cause location strategy is discussed. The effectiveness and practicality of the proposed scheme are demonstrated via a numerical simulation example and practical ethylene oxide production. Two preset simulated faults and one practical process fault are used and all the abnormal states and root variables are monitored and diagnosed by the proposed scheme. The energy monitoring and diagnosis results provide great support for energy management and development in large-scale chemical plants.