scholarly journals Comparing Neural Network Models for Photovoltaic Power Generation Prediction

2021 ◽  
Author(s):  
Carlos Henrique Torres Andrade ◽  
Tiago Figueiredo Vieira ◽  
Ícaro Bezzera Queiroz Araújo ◽  
Gustavo Costa Gomes Melo ◽  
Erick de Andrade Barboza ◽  
...  

Research on alternative energy sources has been increasing for the past years due to environmental concerns and the depletion of fossil fuels. Since photovoltaic generation is intermittent, one needs to predict solar incidence to alleviate problems due to demand surges in conventional distribution systems.Many works have used Long Short-Term Memory (LSTMs) to predict generation. However, to minimize computational costs related to retraining and inference, LSTMs might not be optimal. Therefore, in this work, we compare the performance of MLP (Multilayer Perceptron), Recurrent Neural Networks (RNNs), and LSTMs for the task mentioned above. We used the solar radiance measured throughout 2020 in the city of Maceió (Brazil), taking into account periods of 2 hours for training to predict the next 5-minutes. Hyperparameters were fine-tuned using an optimization approach based on Bayesian inference to promote a fair comparison. Results showed that the MLP has satisfactory performance, requiring much less time to train and forecast. Such results can contribute, for example, to improving response time in embedded systems.

2019 ◽  
Vol 10 (1) ◽  
pp. 1-19
Author(s):  
Matthieu Riou ◽  
Bassam Jabaian ◽  
Stéphane Huet ◽  
Fabrice Lefèvre

Following some recent propositions to handle natural language generation in spoken dialogue systems with long short-term memory recurrent neural network models~\citep{Wen2016a} we first investigate a variant thereof with the objective of a better integration of the attention subnetwork. Then our next objective is to propose and evaluate a framework to adapt the NLG module online through direct interactions with the users. When doing so the basic way is to ask the user to utter an alternative sentence to express a particular dialogue act. But then the system has to decide between using an automatic transcription or to ask for a manual transcription. To do so a reinforcement learning approach based on an adversarial bandit scheme is retained. We show that by defining appropriately the rewards as a linear combination of expected payoffs and costs of acquiring the new data provided by the user, a system design can balance between improving the system's performance towards a better match with the user's preferences and the burden associated with it. Then the actual benefits of this system is assessed with a human evaluation, showing that the addition of more diverse utterances allows to produce sentences more satisfying for the user.


2020 ◽  
Vol 16 (2) ◽  
pp. 74-86 ◽  
Author(s):  
Fatima-Zahra El-Alami ◽  
Said Ouatik El Alaoui ◽  
Noureddine En-Nahnahi

Arabic text categorization is an important task in text mining particularly with the fast-increasing quantity of the Arabic online data. Deep neural network models have shown promising performance and indicated great data modeling capacities in managing large and substantial datasets. This article investigates convolution neural networks (CNNs), long short-term memory (LSTM) and their combination for Arabic text categorization. This work additionally handles the morphological variety of Arabic words by exploring the word embeddings model using position weights and subword information. To guarantee the nearest vector representations for connected words, this article adopts a strategy for refining Arabic vector space representations using semantic information embedded in lexical resources. Several experiments utilizing different architectures have been conducted on the OSAC dataset. The obtained results show the effectiveness of CNN-LSTM without and with retrofitting for Arabic text categorization in comparison with major competing methods.


Land ◽  
2021 ◽  
Vol 10 (3) ◽  
pp. 282
Author(s):  
Alysha van Duynhoven ◽  
Suzana Dragićević

Recurrent Neural Networks (RNNs), including Long Short-Term Memory (LSTM) architectures, have obtained successful outcomes in timeseries analysis tasks. While RNNs demonstrated favourable performance for Land Cover (LC) change analyses, few studies have explored or quantified the geospatial data characteristics required to utilize this method. Likewise, many studies utilize overall measures of accuracy rather than metrics accounting for the slow or sparse changes of LC that are typically observed. Therefore, the main objective of this study is to evaluate the performance of LSTM models for forecasting LC changes by conducting a sensitivity analysis involving hypothetical and real-world datasets. The intent of this assessment is to explore the implications of varying temporal resolutions and LC classes. Additionally, changing these input data characteristics impacts the number of timesteps and LC change rates provided to the respective models. Kappa variants are selected to explore the capacity of LSTM models for forecasting transitions or persistence of LC. Results demonstrate the adverse effects of coarser temporal resolutions and high LC class cardinality on method performance, despite method optimization techniques applied. This study suggests various characteristics of geospatial datasets that should be present before considering LSTM methods for LC change forecasting.


Processes ◽  
2021 ◽  
Vol 9 (7) ◽  
pp. 1121
Author(s):  
Yulim Choi ◽  
Hyeonho Kwun ◽  
Dohee Kim ◽  
Eunju Lee ◽  
Hyerim Bae

Induction furnaces are widely used for melting scrapped steel in small foundries and their use has recently become more frequent. The maintenance of induction furnaces is usually based on empirical decisions of the operator and an explosion can occur through operator error. To prevent an explosion, previous studies have utilized statistical models but have been unable to generalize the problem and have achieved a low accuracy. Herein, we propose a data-driven method for induction furnaces by proposing a novel 2D matrix called a sequential feature matrix(s-encoder) and multi-channel convolutional long short-term memory (s-ConLSTM). First, the sensor data and operation data are converted into sequential feature matrices. Then, N-sequential feature matrices are imported into the convolutional LSTM model to predict the residual life of the induction furnace wall. Based on our experimental results, our method outperforms general neural network models and enhances the safe use of induction furnaces.


Author(s):  
Kau-Fui Vincent Wong ◽  
Guillermo Amador

As society continues advancing into the future, more energy is required to supply the increasing population and energy demands. Unfortunately, traditional forms of energy production through the burning of carbon-based fuels are dumping harmful pollutants into the environment, resulting in detrimental, and possibly irreversible, effects on our planet. The burning of coal and fossil fuels provides energy at the least monetary cost for countries like the US, but the price being paid through their negative impact of our atmosphere is difficult to quantify. A rapid shift to clean, alternative energy sources is critical in order to reduce the amount of greenhouse gas emissions. For alternative energy sources to replace traditional energy sources that produce greenhouse gases, they must be capable of providing energy at equal or greater rates and efficiencies, while still functioning at competitive prices. The main factors hindering the pursuit of alternative sources are their high initial costs and, for some, intermittency. The creation of electrical energy from natural sources like wind, water, and solar is very desirable since it produces no greenhouse gases and makes use of renewable sources—unlike fossil fuels. However, the planning and technology required to tap into these sources and transfer energy at the rate and consistency needed to supply our society comes at a higher price than traditional methods. These high costs are a result of the large-scale implementation of the state-of-the-art technologies behind the devices required for energy cultivation and delivery from these unorthodox sources. On the other hand, as fossil fuel sources become scarcer, the rising fuel costs drive overall costs up and make traditional methods less cost effective. The growing scarcity of fossil fuels and resulting pollutants stimulate the necessity to transition away from traditional energy production methods. Currently, the most common alternative energy technologies are solar photovoltaics (PVs), concentrated solar power (CSP), wind, hydroelectric, geothermal, tidal, wave, and nuclear. Because of government intervention in countries like the US and the absence of the need to restructure the electricity transmission system (due to the similarity in geographical requirements and consistency in power outputs for nuclear and traditional plants), nuclear energy is the most cost competitive energy technology that does not produce greenhouse gases. Through the proper use of nuclear fission electricity at high efficiencies could be produced without polluting our atmosphere. However, the initial capital required to erect nuclear plants dictates a higher cost over traditional methods. Therefore, the government is providing help with the high initial costs through loan guarantees, in order to stimulate the growth of low-emission energy production. This paper analyzes the proposal for the use of nuclear power as an intermediate step before an eventual transition to greater dependence on energy from wind, water, and solar (WWS) sources. Complete dependence on WWS cannot be achieved in the near future, within 20 years, because of the unavoidable variability of these sources and the required overhaul of the electricity transmission system. Therefore, we look to nuclear power in the time being to help provide predictable power as a means to reduce carbon emissions, while the other technologies are refined and gradually implemented in order to meet energy demand on a consistent basis.


2021 ◽  
Author(s):  
Zeynu Shamil Awol ◽  
Rezika Tofike Abate

Abstract Biomass energy is renewable energy source that comes from the material of plants and animals. Forms of biomass energy are bio-ethanol, bio methanol, and biodiesel. Bio-ethanol is one of the most important alternative energy sources that substitute the fossil fuels. The focus of this research is to produce bio-ethanol from waste office paper. Five laboratory experiments were conducted to produce bio-ethanol from wastepaper. The wastepaper was dried in oven and cut in to pieces. Then it passed through dilute acid hydrolysis, fermentation and distillation process respectively. High amount of ethanol was observed at 20 ml/g (liquid to solid ratio) and at the time of 2hr. Cost and economic analysis for ethanol production from wastepaper was performed. Results from the analysis indicated a paper to ethanol plant was feasible from the economic point of view with rate of return (RR) 38.61% and the payback period of 2.2 years.


2016 ◽  
Vol 19 (3) ◽  
pp. 96-109
Author(s):  
Phung Thi Kim Le ◽  
Viet Tan Tran ◽  
Thien Luu Minh Nguyen ◽  
Viet Vuong Pham ◽  
Truc Thanh Nguyen ◽  
...  

Finding alternative energy sources for fossil fuels was a global matter of concern, especially in developing countries. Rice husk, an abundant biomass in Viet Nam, was used to partially replace fossil fuels by gasification process. The study was conducted on the pilot plant fixed bed up-draft gasifier with two kind of gasification agents, pure air and air-steam mixture. Mathematical modeling and computer simulations were also used to describe and optimize the gasification processes. Mathematical modeling was based on Computational Fluid Dynamics method and simulation was carried by using Ansys Fluent software. Changes in outlet composition of syngas components (CO, CO2, CH4, H2O, H2) and temperature of process, in relation with ratio of steam in gasification agents, were presented. Obtained results indicated concentration of CH4, H2 in outlet was increased significantly when using air-steam gasification agents than pure air. The discrepancies among the gasification agents were determined to improve the actual process.


2019 ◽  
Vol 9 (19) ◽  
pp. 3945 ◽  
Author(s):  
Houssem Gasmi ◽  
Jannik Laval ◽  
Abdelaziz Bouras

Extracting cybersecurity entities and the relationships between them from online textual resources such as articles, bulletins, and blogs and converting these resources into more structured and formal representations has important applications in cybersecurity research and is valuable for professional practitioners. Previous works to accomplish this task were mainly based on utilizing feature-based models. Feature-based models are time-consuming and need labor-intensive feature engineering to describe the properties of entities, domain knowledge, entity context, and linguistic characteristics. Therefore, to alleviate the need for feature engineering, we propose the usage of neural network models, specifically the long short-term memory (LSTM) models to accomplish the tasks of Named Entity Recognition (NER) and Relation Extraction (RE). We evaluated the proposed models on two tasks. The first task is performing NER and evaluating the results against the state-of-the-art Conditional Random Fields (CRFs) method. The second task is performing RE using three LSTM models and comparing their results to assess which model is more suitable for the domain of cybersecurity. The proposed models achieved competitive performance with less feature-engineering work. We demonstrate that exploiting neural network models in cybersecurity text mining is effective and practical.


Sign in / Sign up

Export Citation Format

Share Document