Greenhouse gas (GHG) emissions, especially CO₂, represent a global concern. Among those responsible for CO₂ emissions, buildings stand out due to the consumption of energy from fossil fuels. In this sense, initiatives for the decarbonization of buildings and construction tends to contribute to the achievement of the target defined in the Paris Agreement of limiting the increase in global temperature to 1.5 degrees Celsius above pre-industrial levels, as well as in achieving the Sustainable Development Goals (SDG) and of the Triple Bottom Line (TBL). This article aimed to identify renewable energy generation technologies that can be applied in urban vertical constructions, contributing to the reduction of carbon emissions in the atmosphere. To this end, the following methodology was adopted: a survey of the Conferences of the Parties on climate change; identification of European Union Legislative Directives for the decarbonization of buildings; and a literature review to identify research that deals with renewable energy generation technologies that can be adopted in buildings. The results indicated that there seems to be a correlation between the growth in the number of articles that deal with the topic of decarbonizing buildings and the increase in world concerns about global warming. A hybrid microgrid proposal, combining different sources of renewable energy such as solar photovoltaic, wind, biomass, micro-hydroelectric, and others for vertical buildings with more than five floors, is presented as viable to achieve zero emissions in these buildings, contributing to future research, that can carry out quantitative analyses and feasibility studies, as well as for experiments and applications in existing buildings and the projects of new vertical constructions.

Brazilian legal system prescribes means of ensuring the prompt processing of court cases, such as the principle of reasonable process duration, the principle of celerity, procedural economy, and due legal process, with a view to optimizing procedural progress. In this context, one of the great challenges of the Brazilian judiciary is to predict the duration of legal cases based on information such as the judge, lawyers, parties involved, subject, monetary values of the case, starting date of the case, etc. Recently, there has been great interest in estimating the duration of various types of events using artificial intelligence algorithms to predict future behaviors based on time series. Thus, this study presents a proof-of-concept for creating and demonstrating a mechanism for predicting the amount of time, after the case is argued in court (time when a case is made available for the magistrate to make the decision), for the magistrate to issue a ruling. Cases from a Regional Labor Court were used as the database, with preparation data in two ways (original and discretization), to test seven machine learning techniques (i) Multilayer Perceptron (MLP); (ii) Gradient Boosting; (iii) Adaboost; (iv) Regressive Stacking; (v) Stacking Regressor with MLP; (vi) Regressive Stacking with Gradient Boosting; and (vii) Support Vector Regression (SVR), and determine which gives the best results. After executing the runs, it was identified that the adaboost technique excelled in the task of estimating the duration for issuing a ruling, as it had the best performance among the tested techniques. Thus, this study shows that it is possible to use machine learning techniques to perform this type of prediction, for the test data set, with an R²of 0.819 and when transformed into levels, an accuracy of 84%.

Recent advancements in Artificial Intelligence have yielded promising results in addressing complex challenges within Natural Language Processing (NLP), serving as a vital tool for expediting judicial proceedings in the legal domain. This study focuses on the detection of similarity among judicial documents within an inference group, employing eight NLP techniques grounded in transformer architecture, specifically applied to a case study of legal proceedings in the Brazilian judicial system. The transformer-based models utilised — BERT, GPT-2, RoBERTa, and LlaMA — were pre-trained on general-purpose corpora of Brazilian Portuguese and subsequently fine-tuned for the legal sector using a dataset of 210,000 legal cases. Vector representations of each legal document were generated based on their embeddings, facilitating the clustering of lawsuits and enabling an evaluation of each model’s performance through the cosine distance between group elements and their centroid. The results demonstrated that transformer-based models outperformed traditional NLP techniques, with the LlaMA model, specifically fine-tuned for the Brazilian legal domain, achieving the highest accuracy. This research presents a methodology employed in a real case involving substantial documentary content that can be adapted for various applications. It conducts a comparative analysis of existing techniques focused on a non-English language to quantitatively explain the results obtained with various NLP transformers-based models. This approach advances the current state of the art in NLP applications within the legal sector and contributes to the achievement of Sustainable Development Goals.

The transition to sustainable biofuels has raised interest in microbial electrosynthesis (MES) as a CO2-based strategy for ethanol (EtOH) production. EtOH productivity in MES depends on operational optimization; however, product quantification methods delay decision-making, limiting precise control. Data-driven soft sensors utilize available signals to estimate variables difficult to measure in real-time. This study aimed to develop the first soft sensor to infer EtOH concentration from MES experimental data. Two deep learning methods for time series process data (Long Short-Term Memory-LSTM and Convolutional Neural Network-CNN) were evaluated to estimate EtOH concentration. Both were trained with experimental process data generated over 545 days on a bench-scale platform for MES, collecting data from 14 operational variables and historical EtOH concentration. Five training-validation distributions were evaluated to identify the most robust training strategy. CNN models demonstrated superior performance over LSTM, and the model trained with data from all experimental runs (CNN-E) achieved the best validation performance. CNN-E was further evaluated in a new batch of the same system, using recursive forecasting scenarios in which the model was fed back with its own predictions. CNN-E forecast followed the actual EtOH trend with relatively low errors (RMSE= 5.5571, R2= 0.3286). Further recalibration with updated EtOH laboratory measurements improved accuracy to RMSE= 2.7821 and R2= 0.8919. These results demonstrate the feasibility of CNN-based soft sensors for real-time EtOH monitoring in MES, supporting their potential integration into future control strategies aimed at advancing the digitalization of CO2 bioelectrorecycling processes.

The transition to sustainable biofuels has raised interest in microbial electrosynthesis (MES) as a CO2-based strategy for ethanol (EtOH) production. EtOH productivity in MES depends on operational optimization; however, conventional quantification methods delay decision-making, limiting precise control. Data-driven soft sensors utilize available signals to estimate variables difficult to measure in real-time. This study aimed to develop the first soft sensor to infer EtOH concentration from MES experimental data. Two deep learning methods for time series process data (Long Short-Term Memory-LSTM and Convolutional Neural Network-CNN) were evaluated to estimate EtOH concentration. Both methods were trained with experimental process data generated over 545 days on a bench-scale platform for MES, collecting data from 14 operational variables and historical EtOH concentration in 2 reactors. Different models were developed using five different training-validation data distributions to understand which training strategy fit better, and the best performing model was tested over an unseen dataset. CNN models demonstrated superior EtOH forecasting performance over LSTM. The model trained with data from all experimental runs (CNN-E) achieved the best performance, with validation accuracy of R2= 0.9990. Furthermore, generalization of CNN-E over a new experimental dataset showed prominent high accuracy (R2= 0.9993). SHAP analysis demonstrated that the CNN-E model was primarily influenced by the autoregressive component but refined the forecasting with the CO2 signal, identified as the trigger towards EtOH increase. The high accuracy and explainability on the CNN-E model position it as a suitable real-time EtOH soft sensor potentially capable of feeding MES monitoring and control strategies.