Research on power demand forecasting based on attention mechanism and deep learning network

Combining actual conditions, power demand forecasting is affected by various uncertain factors such as meteorological factors, economic factors, and diversity of forecasting models, which increase the complexity of forecasting. In response to this problem, taking into account that different time step states will have different effects on the output, the attention mechanism is introduced into the method proposed in this paper, which improves the deep learning model. Improved models of convolutional neural networks (CNN) and long short-term memory (LSTM) that combine the attention mechanism are proposed respectively. Finally, according to the verification results of actual examples, it is proved that the proposed method can obtain a smaller error and the prediction performance are better compared with other models.

Assessment of Electric Consumption Forecast Using Machine Learning and Deep Learning Models for the Industrial Sector

Power demand forecasting is one of the fields which is gaining popularity for researchers. Although machine learning models are being used for prediction in various fields, they need to upgrade to increase accuracy and stability. With the rapid development of AI technology, deep learning (DL) is being recommended by many authors in their studies. The core objective of the chapter is to employ the smart meter's data for energy forecasting in the industrial sector. In this chapter, the author will be implementing popular power demand forecasting models from machine learning and compare the results of the best-fitted machine learning (ML) model with a deep learning model, long short-term memory based on RNN (LSTM-RNN). RNN model has vanishing gradient issue, which slows down the training in the early layers of the network. LSTM-RNN is the advanced model which take care of vanishing gradient problem. The performance evaluation metric to compare the superiority of the model will be R2, mean square error (MSE), root means square error (RMSE), and mean absolute error (MAE).

Robotic Expert System for Energy Management in Distributed Grid Ecosystem

A robotic expert system (RES) for energy management (EM) in community-based micro-grids is developed using a fuzzy computational scheme. Within the micro-grid multi-dimensional space, embedded algorithms for residential homes, sectors and central controller units are introduced to perform EM in a collaborative manner. Demand response and load shedding are carried out within the community micro-grid to ascertain the behavioral responses based on changes in power demand levels. Various tests are carried out with an observable low error margin. It was observed that the system reduced the total power demand on the micro-grid by 20% of the total distributed power. Micro-grid RES, neuro-fuzzy control (NFC), and support vector regression (SVR) evaluations are compared considering the home units at 40kW of the generated capacity. The results gave a 35.79%, 31.58% and 32.63% energy demand, respectively. Consequently, RES provides a grid look-ahead prediction, annotated-self healing, and stability restoration.

Modeling and Analysis of STATCOM for Renewable Energy Farm to Improve Power Quality and Reactive Power Compensation

Power Quality (PQ) improvement in grid-integrated photovoltaic (PV) and wind energy hybrid systems for effective power transfer is presented in this paper. Due to interlinked hybrid renewable energy resources and nonlinear loads, various issues arise which affect the power quality, i.e., voltage sag, harmonic distortion increases, and also reactive power demand. In order to mitigate these issues, flexible alternating current transmission system (FACTS) devices are utilized. In this paper, hysteresis band current controller (HBCC)-based static synchronous compensator (STATCOM) is modeled to reduce PQ problems. HBCC is a robust and simple technique to improve voltage profile, reduce total harmonic distortion (THD) and fulfill the reactive power demand. Two case scenarios of the hybrid system, i.e., (I) grid integrated hybrid system without HBCC (II) grid integrated hybrid system with HBCC, are tested. Results demonstrate that under scenario II, load bus voltage is regulated at 1.0 p.u., THD of system voltage and current is reduced 0.25% and 0.35%, respectively, and reactive power demand of 30 kVAR is fulfilled. The HBCC was designed for reducing THD of the system with the limits specified by standards IEEE 519-1992 STATCOM using hysteresis band current controller to improve power quality in the distribution system which is simulated using MATLAB/SIMULINK. After that, the performance of the system is better in terms of power quality.

Power Management for Connected EVs Using a Fuzzy Logic Controller and Artificial Neural Network

In recent years, the electric vehicles (EVs) power management strategy has been developed in order to reduce battery discharging power and fluctuation when an EV requires high and rapid discharging power due to frequent stop-and-go driving operations. A combination of lithium-ion batteries and a supercapacitor (SC) as the EV’s energy sources is known as a hybrid energy storage system (HESS) and is a promising solution for fast discharging conditions. Effective power management to extensively utilize HESS can be developed if future power demand is accessible. A vehicular network as a typical form of the currently developed internet of things (IoT) has made future information obtainable by collecting information on surrounding data. This paper proposes a power management strategy for the HESS with the support of IoT. Since the obtained information from vehicular network could not directly be used to improve HESS, a two levels control structure has been developed to perform future data prediction and power distribution. A fuzzy logic controller (FLC) is utilized in the level one control structure to manage a HESS power split based on future information. Since FLC requires future information as a reference input, the future information is obtained by using an artificial neural network (ANN) in a level two control structure. The ANN prediction is direct, which could approximate the future power demand prediction with the assumption that the vehicular network scenario that is used to obtain surrounding information is deployed. Simulation results demonstrate that the average discharging battery power and power variation are reduced by 46.1% and 52.3, respectively, when compared to the battery-only case.

A Power Demand Analytical Model of Self-Propelled Vessels

The article herein presents a closed-form mathematical equation by which it is possible to estimate the propulsion power demand of ships as a function of the propeller parameters and total Resistance. The validation of the derived model is conducted by use of the Series 60 Model data and of the Korea Research Institute of Ships and Ocean Engineering (KRISO) Very Large Crude-oil Carrier 2 (KVLCC2) data. In all the cases tested, the derived model explained more than 99.9% of the data variability. Furthermore, the paper describes a practical method for quantifying changes in hull and propeller performance and provides an application example.