A Real-Time Estimation Framework of Carbon Emissions in Steel Plants Based on Load Identification

Abstract

In recent years, extreme weather disasters have occurred frequently, so reducing greenhouse gas emissions is an urgent task. How to accurately estimate corporate carbon emissions in real-time directly affects this task. Therefore, a real-time estimation framework of corporate carbon emissions based on load identification is proposed in this paper. In the proposed framework, the total carbon emission consists of two parts: direct carbon emission and indirect carbon emission. First, a Convolutional Neural Network - Bidirectional Long Short-Term Memory (CNN-BLSTM) model is presented and employed to monitor the states of devices in a factory in realtime. Then, the direct carbon emission is estimated according to the state and related carbon emission intensity. Meanwhile, the indirect carbon emission can be obtained through multiplying the marginal carbon emission factor by the electricity consumption in the factory. The proposed framework was used in a case study to estimate the carbon emission of a steel plant in real-time, which proved its effectiveness and accuracy.