Recent years have seen an increasing interest in Demand Response (DR) as a means to provide flexibility, and hence improve the reliability of energy systems in a cost-effective way. Yet, the high complexity of the tasks associated with DR, combined with their use of large-scale data and the frequent need for near real-time de-cisions, means that Artificial Intelligence (AI) and Machine Learning (ML) — a branch of AI — have recently emerged as key technologies for enabling demand-side response. AI methods can be used to tackle various challenges, ranging from selecting the optimal set of consumers to respond, learning their attributes and pref-erences, dynamic pricing, scheduling and control of devices, learning how to incentivise participants in the DR schemes and how to reward them in a fair and economically efficient way. This work provides an overview of AI methods utilised for DR applications, based on a systematic review of over 160 papers, 40 companies and commercial initiatives, and 21 large-scale projects. The papers are classified with regards to both the AI/ML algorithm(s) used and the application area in energy DR. Next, commercial initiatives are presented (including both start-ups and established companies) and large-scale innovation projects, where AI methods have been used for energy DR. The paper concludes with a discussion of advantages and potential limitations of reviewed AI techniques for different DR tasks, and outlines directions for future research in this fast-growing area.

/pdf/artificial-intelligence-and-machine-learning-approaches-to-5er9qs2c17.pdf

Artificial Intelligence and Machine Learning Approaches to Energy Demand-Side Response: A Systematic Review

Due to urbanization and the rapid growth of population, carbon emission is increasing, which leads to climate change and global warming. With an increased level of fossil fuel burning and scarcity of fossil fuel, the power industry is moving to alternative energy resources such as photovoltaic power (PV), wind power (WP), and battery energy-storage systems (BESS), among others. BESS has some advantages over conventional energy sources, which include fast and steady response, adaptability, controllability, environmental friendliness, and geographical independence, and it is considered as a potential solution to the global warming problem. This paper provides a comprehensive review of the battery energy-storage system concerning optimal sizing objectives, the system constraint, various optimization models, and approaches along with their advantages and weakness. Furthermore, for better understanding, the optimization objectives and methods have been classified into different categories. This paper also provides a detailed discussion on the BESS applications and explores the shortages of existing optimal BESS sizing algorithms to identify the gaps for future research. The issues and challenges are also highlighted to provide a clear idea to the researchers in the field of BESS. Overall, this paper conveys some significant recommendations that would be useful to the researchers and policymakers to structure a productive, powerful, efficient, and robust battery energy-storage system toward a future with a sustainable environment.

Battery energy-storage system: A review of technologies, optimization objectives, constraints, approaches, and outstanding issues

Prediction of solar energy guided by pearson correlation using machine learning

Prognostics and health management is a promising methodology to cope with the risks of failure in advance and has been implemented in many well-known applications including battery systems. Since the estimation of battery capacity is critical for safe operation and decision making, battery capacity should be estimated precisely. In this regard, we leverage measurable data such as voltage, current, and temperature profiles from the battery management system whose patterns vary in cycles as aging. Based on these data, the relationship between capacity and charging profiles is learned by neural networks. Specifically, to estimate the state of health accurately we apply feedforward neural network, convolutional neural network, and long short-term memory. Our results show that the proposed multi-channel technique based on voltage, current, and temperature profiles outperforms the conventional method that uses only voltage profile by up to 25%–58% in terms of mean absolute percentage error.

/pdf/machine-learning-based-lithium-ion-battery-capacity-2v84fsopg5.pdf

Machine Learning-Based Lithium-Ion Battery Capacity Estimation Exploiting Multi-Channel Charging Profiles

Irregular human behaviors and univariate datasets remain as two main obstacles of data-driven energy consumption predictions for individual households. In this study, a hybrid deep learning model is proposed combining an ensemble long short term memory (LSTM) neural network with the stationary wavelet transform (SWT) technique. The SWT alleviates the volatility and increases the data dimensions, which potentially help improve the LSTM forecasting accuracy. Moreover, the ensemble LSTM neural network further enhances the forecasting performance of the proposed method. Verification experiments were performed based on a real-world household energy consumption dataset collected by the ‘UK-DALE project. The results show that, with a competitive training efficiency, the proposed method outperforms all compared state-of-art methods, including the persistent method, support vector regression (SVR), long short term memory (LSTM) neural network and convolutional neural network combining long short term memory (CNN-LSTM), with different step sizes at 5, 10, 20 and 30 minutes, using three error metrics.

/pdf/a-hybrid-lstm-neural-network-for-energy-consumption-34m43uf3js.pdf

A Hybrid LSTM Neural Network for Energy Consumption Forecasting of Individual Households

As the number of smart meters increases, compression of metering data becomes essential for data transmission, storing and processing perspectives. Specifically, feature extraction can be used for the compression of metering data and further be utilized for smart grid applications such as customer clustering. So far, there are many studies for compression and clustering based on daily load profiles. However, in order to account for long-term characteristics of electricity load, utilizing yearly load profiles (YLPs) is vital for customer load clustering and analysis. In this paper, we propose a deep learning-based YLP feature extraction that jointly captures daily and seasonal variations. By leveraging convolutional autoencoder (CAE), YLPs in 8,640-dimensional space are compressed to 100-dimensional vectors. We apply the proposed CAE framework to YLPs of 1,405 residential customers and verify that the proposed CAE outperforms other dimensionality reduction methods in terms of reconstruction errors, e.g., by 19–40%, or the compression ratio is increased by 130% or higher than other methods for the same reconstruction error. In addition, clustering analysis is performed on the encoded YLPs. Our results confirm that year-round characteristics are well captured during the clustering process and also clearly visualized with load images.

Convolutional Autoencoder Based Feature Extraction and Clustering for Customer Load Analysis

Recent development of artificial intelligence (AI) makes AI applicable to diverse fields, and the smart grid is not an exception. In particular, there have been extensive researches on load forecasting using deep learning. Most existing studies have been conducted on deep neural network (DNN) and recurrent neural network (RNN). Very recently, CNN with shallow network has been studied for short-term load forecasting (STLF). In this paper, we propose a novel framework based on ResNet/LSTM combined model. The proposed model has two steps. First, ResNet extracts latent features of daily and weekly load data. Then, LSTM is applied to train the encoded feature vector with dynamics, and make prediction suitable for volatile load data. By leveraging ResNet and LSTM, the proposed model has the advantage of forecasting load data that has both regularity and inconsistency. To demonstrate the performance, we compare the proposed model with other deep learning models: multi-layer perceptron (MLP), ResNet, LSTM and ResNet/MLP combined model. The results show that the proposed ResNet/LSTM combined model has 21.3% of MAPE improvement in overall, and 25.8% of MAPE improvement for the bottom 25% group in terms of MAPE compared to MLP.

Short-Term Load Forecasting based on ResNet and LSTM

In energy data analytics, load profile clustering is essential for various smart grid applications such as demand response, load forecasting, and tariff design. Most of the conventional clustering techniques are based on a representative time domain load profile within a certain period, and the daily and seasonal variations are not well captured. In this paper, we propose a deep learning based customer load profile clustering framework that jointly captures daily and seasonal variations. By leveraging convolutional autoencoder (CAE), the yearly load profile in the time domain is converted into a representative vector in the smaller dimensional encoded space. The clusters are then determined based on the vectors encoded by the CAE. We apply the proposed framework to 1,405 households' yearly load profiles and verify that the trained CAE can encode those load profiles into approximately 100 times smaller dimensional space. The encoded load profiles can be decoded by the CAE with a negligible loss between 1–3%. The clustered load images can visualize both daily and seasonal variations, and clustering in the encoded space speeds up the clustering process by almost three orders of magnitude.

Residential Load Profile Clustering via Deep Convolutional Autoencoder

Recent advances in battery technologies have reduced the financial burden of using the energy storage system (ESS) for customers. Peak cut, one of the benefits of using ESS, can be achieved through proper charging/discharging scheduling of ESS. However, peak cut is sensitive to load-forecasting error, and even a small forecasting error may result in the failure of peak cut. In this paper, we propose a two-phase approach of day-ahead optimization and real-time control for minimizing the total cost that comes from time-of-use (TOU), peak load, and battery degradation. In day-ahead optimization, we propose to use an internalized pricing to manage peak load in addition to the cost from TOU. The proposed method can be implemented by using dynamic programming, which also has an advantage of accommodating the state-dependent battery degradation cost. Then in real-time control, we propose a concept of marginal power to alleviate the performance loss incurred from load-forecasting error and mimic the offline optimal battery scheduling by learning from load-forecasting error. By exploiting the marginal power, real-time ESS charging/discharging power gets close to the offline optimal battery scheduling. Case studies show that under load-forecasting uncertainty, the peak power using the proposed method is only 22.4% higher than the offline optimal peak power, while the day-ahead optimization has 76.8% higher peak power than the offline optimal power. In terms of profit, the proposed method achieves 77.0% of the offline optimal profit while the day-ahead method only earns 19.6% of the offline optimal profit, which shows the substantial improvement of the proposed method.

https://www.mdpi.com/1996-1073/12/6/1098/pdf

Hyungeun Choi

Papers

Convolutional Autoencoder Based Feature Extraction and Clustering for Customer Load Analysis

Short-Term Load Forecasting based on ResNet and LSTM

Residential Load Profile Clustering via Deep Convolutional Autoencoder

Practical Operation Strategies for Energy Storage System under Uncertainty