Abstract
Motivated by initiatives such as the UN Sustainable Development Goals (SDG), particularly SDG 1-Poverty Eradication and SDG 7-Clean and Accessible Energy, the search for solutions aiming to mitigate poverty has been recurrent in several studies. This paper main objective is to evaluate the dynamics of global research on the use of photovoltaic projects for poverty alleviation (PVPA) from 2003 to 2022. We use a bibliometric analysis to identify publication patterns and consequently list research trends and gaps of the area. A total of 336 publications from Scopus database are identified and complemented by a state-of-the-art study, where the articles are investigated and classified according to: Business model and financing and evaluation of PVPA results. The results show that PA is often associated with PV power and its application in rural areas. “Biomass” and “application in developing countries” have become a trend. Urban areas application, aiming to reduce poverty, and the need for a synergetic integration of energy and urban planning, to mitigate the risks associated with energy flow and efficiency, are the most relevant gaps identified. Most of the publications focus on macropolicies effects involving PV technology; papers on projects construction and ex-post are not identified.

Abstract
This article presents a methodology to estimate the evolution of QoS indices, based on investments and maintenance costs carried out in the DN. The indices were estimated at various disaggregated levels, including the global index, 3 different QoS zones (urban, semi-urban and rural) and 278 municipalities, thereby facilitating the mitigation of QoS asymmetries by allocating investments and maintenance actions to specific regions. To achieve this objective, an optimization problem was formulated to allocate investments and maintenance costs to municipalities with higher improvement benefit-cost ratios, potentially exhibiting lower levels of QoS. This methodology was adopted by the Portuguese DSO to establish the future investments plan from 2023 to 2027. The results demonstrate estimations of good performance, considering the stochastic nature of the phenomena affecting QoS (e.g. atmospheric conditions), which are included in this study, thus developing confidence levels for the global indices.

Abstract
A common problem in distribution planning is the scarcity of historic data (training examples) relative to the number of variables, meaning that most data-driven techniques cannot be applied in such situations, due to the risk of overfitting. Thus, the suitable regression techniques are restrained to efficient models, preferably with embedded regularization features. This article compares three of these techniques: LASSO, Bayesian and CMLR (Conditioned multi-linear regression - a new approach developed within the scope of a project with a distribution company). The results showed that each technique has its own advantages and limitations. The Bayesian regression has the main advantage of providing inherent confidence intervals. The LASSO is a very economic and efficient regression tool. The CMLR is versatile and provided the best performance.A common problem in distribution planning is the scarcity of historic data (training examples) relative to the number of variables, meaning that most data-driven techniques cannot be applied in such situations, due to the risk of overfitting. Thus, the suitable regression techniques are restrained to efficient models, preferably with embedded regularization features. This article compares three of these techniques: LASSO, Bayesian and CMLR (Conditioned multi-linear regression - a new approach developed within the scope of a project with a distribution company). The results showed that each technique has its own advantages and limitations. The Bayesian regression has the main advantage of providing inherent confidence intervals. The LASSO is a very economic and efficient regression tool. The CMLR is versatile and provided the best performance.

Abstract
This paper explores the application of Deep Learning techniques to forecast electricity market prices. Three Deep Learning (DL) techniques are tested: Dense Neural Networks (DNN), Long Short-Term Memory Networks (LSTM) and Convolutional Neural Networks (CNN); and two non-DL techniques: Multiple Linear Regression and Gradient Boosting (GB). First, this work compares the forecast skill of all techniques for electricity price forecasting. The results analysis showed that CNN consistently remained among the best performers when predicting the most unusual periods such as the Covid19 pandemic one. The second study evaluates the potential application of CNN for automatic feature extraction over a dataset composed by multiple explanatory variables of different types, overcoming part of the feature selection challenges. The results showed that CNNs can be used to reduce the need for a variable selection phase.

Abstract
Nontechnical losses in electricity distribution networks are often associated with a countries' socioeconomic situation. Although the amount of global losses is usually known, the separation between technical and commercial (nontechnical) losses will remain one of the main challenges for DSO until smart grids become fully implemented and operational. The most common origins of commercial losses are energy theft and deliberate or accidental failures of energy measuring equipment. In any case, the consequences can be regarded as consumption anomalies. The work described in this paper aims to answer a request from a DSO, for the development of tools to detect consumption anomalies at end-customer facilities (HV, MV and LV), invoking two types of assessment. The first consists of the identification of typical patterns in the set of consumption profiles of a given group or zone and the detection of atypical consumers (outliers) within it. The second assessment involves the exploration of the load diagram evolution of each specific consumer to detect changes in the consumption pattern that could represent situations of probable irregularities. After a representative period, typically 12 months, these assessments are repeated, and the results are compared to the initial ones. The eventual changes in the typical classes or consumption scales are used to build a classifier indicating the risk of anomaly.