A Data-Driven Multi-Regime Approach for Predicting Energy Consumption

Abstract

There has been increasing interest in reducing carbon footprints globally in recent years. Hence increasing share of green energy and energy efficiency are promoted by governments. Therefore, optimizing energy consumption is becoming more critical for people, companies, industries, and the environment. Predicting energy consumption more precisely means that future energy management planning can be more effective. To date, most research papers have focused on predicting residential building energy consumption; however, a large portion of the energy is consumed by industrial machines. Prediction of energy consumption of large industrial machines in real time is challenging due to concept drift, in which prediction performance deteriorates over time. In this research, a novel data-driven method multi-regime approach (MRA) was developed to better predict the energy consumption for industrial machines. Whereas most papers have focused on finding an excellent prediction model that contradicts the no-free-lunch theorem, this study concentrated on adding potential concept drift points into the prediction process. A real-world dataset was collected from a semi-autonomous grinding (SAG) mill used as a data source, and a deep neural network was utilized as a prediction model for the MRA method. The results proved that the MRA method enables the detection of multi-regimes over time and provides a highly accurate prediction performance, thanks to the dynamic model approach.

Dynamic Modeling With Integrated Concept Drift Detection for Predicting Real-Time Energy Consumption of Industrial Machines

“…According to the [42], multi-layer perceptron (MLP) networks are superior for predicting energy consumption compared with standard regression models. However, according to the no-free-lunch theory, there are no specific models for certain problems [13].…”

“…Most companies have started reducing their energy expenses to ensure a better profit margin and reduce emissions to minimize their environmental impacts. Furthermore, load forecasting enables better energy management, and thus lower costs [5], [13].…”

“…It is difficult to explicitly observe or detect any specific patterns or sudden shifts in machine energy consumption. Uncertainties in streamed datasets may include variances in the data known as concept drift [13] or model degradation [14]. While the allocation of the input variable may or may not vary, concept drift is defined as a change in the conditional distribution of the target variable [15].…”

“…Furthermore, many researchers use synthetic datasets, in which all concept drift points are known in advance. However, working with real-world industrial datasets is far more challenging as they typically include many missing values or unrecorded features, as well as unknown concept drifts and machine running conditions [13], [19].…”

Shaping energy cost management in process industries through clustering and soft sensors

“…Shi et al (2021) designed a model based on convolutional neural networks to predict coal and electricity consumption simultaneously, and this model also eliminated the negative effects of the coupling between variables. Kahraman et al (2021) proposed a data-driven method based on the deep neural network, which provided a highly accurate prediction performance for energy consumption of industry machines. The NARXNN adopted in this study is a neural network that combines autoregression and exogenous input series, and this model has the additional advantage of handling nonlinear time series compared to the ARMAX model.…”