Search

In this paper we propose and discuss different Deep Learning-based ensemble algorithms for a problem of low-visibility events prediction due to fog. Specifically, seven different Deep Learning (DL) architectures have been considered, from which multiple individual learners are generated. Hyperparameters of the models, including parameters concerning data preprocessing, models architecture and training procedure, are randomly selected for each model within a pre-defined discrete range. Also, every model is trained with slightly different data sampled randomly, assuring that every models introduce variety in the ensemble. Then, three different information fusion techniques are employed to build the ensemble models. The influence of the filtering process and the elitism level (the percentage of the individual models entering the ensemble) is also assessed. The performance of the proposed methodology have been tested in two real problems of low-visibility events prediction due to orographical and radiation fog, at the north of Spain. Comparison with different Machine Learning, alternative DL algorithms and meteorological-based methods show the good performance of the proposed deep learning ensembles in this problem. [ABSTRACT FROM AUTHOR]

Highlights • Hourly-resolved model to investigate highly-renewable electricity generation in Spain. • Correlation analysis of time series to create 900 combinations used in simulations. • Transition paths evaluated based on security of supply, CO 2 emissions, and renewable share. • Short-term phase-out of nuclear and coal power plants proven to be feasible. Abstract Two alternative paths to achieve highly-renewable electricity generation in peninsular Spain are investigated in this paper. Every transition path comprises a description of the installed and decommissioned generation and storage capacities, from 2017 to 2030, as well as a hypothesis on the evolution of the electricity demand. The electricity mix for every hour within the transition path is determined through a dispatch algorithm that prioritizes electricity from renewable energy sources. The simulation is run for 900 different combinations of time series representing the hourly capacity factors of different technologies, as well as the electricity demand. This robust approach allows the evaluation of the transition paths based on the statistical distribution of several defined assessment criteria, such as security of supply, CO 2 emissions or renewable share in electricity generation. The feasibility of a Spanish power system with high renewable penetration is investigated not only in a future reference year but throughout the transition path. In particular, a progressive and simultaneous phase-out of nuclear and coal power plants in the short-term is proven to be feasible. Furthermore, the results sensitivity is analyzed including scenarios with a delayed nuclear phase-out, lower hydroelectricity generation due to more frequent and severe droughts caused by climate change and higher annual increment for the electricity demand. [ABSTRACT FROM AUTHOR]

Abstract: This paper presents an evolutionary algorithm for wind speed reconstruction from synoptic pressure patterns. The algorithm operates in a search space formed by grids of pressure measures, and must classify the different situations into classes, in such a way that a measure of wind speed in a given point is minimized among patterns assigned to the same class. Then, each class is assigned a mean wind speed and direction, so the wind speed reconstruction is possible for a new grid of synoptic pressures. In this paper we present the problem model and the specific description of the evolutionary algorithm proposed to solve the problem. We also show the good performance of the proposed method in the reconstruction of the average wind speed in six wind towers in Spain. The proposed method is applicable to wind speed reconstruction or reconstruction of wind missing data of wind series, specially when there is no other variable or related measure available. [Copyright &y& Elsevier]

Abstract: In recent years, fuzzy models have been acknowledged as a suitable approach for species distribution modelling due to their transparency and their ability to incorporate the ecological gradient theory. Specifically, the overlapping class boundaries of a fuzzy model are similar to the transitions between different environmental conditions. However, the need for ecological expert knowledge is an important constraint when applying fuzzy species distribution models. Moreover, the consistency of the ecological preferences of some fish species across different rivers has been widely contested. Recent research has shown that data-driven fuzzy models may solve this ‘knowledge acquisition bottleneck’ and this paper is a further contribution. The aim was to analyse the brown trout (Salmo trutta fario L.) habitat preferences based on a data-driven fuzzy modelling technique and to compare the resulting fuzzy models with a commonly applied modelling technique, Random Forests. A heuristic nearest ascent hill-climbing algorithm for fuzzy rule optimisation and Random Forests were applied to analyse the ecological preferences of brown trout in 93 mesohabitats. No significant differences in model performance were observed between the optimal fuzzy model and the Random Forests model and both approaches selected river width, the cover index and flow velocity as the most important variables describing brown trout habitat suitability. Further, the fuzzy model combined ecological relevance with reasonable interpretability, whereas the transparency of the Random Forests model was limited. This paper shows that fuzzy models may be a valid approach for species distribution modelling and that their performance is comparable to that of state-of-the-art modelling techniques like Random Forests. Fuzzy models could therefore be a valuable decision support tool for river managers and enhance communication between stakeholders. [Copyright &y& Elsevier]

Abstract: We present in this paper the Generic RAdiative traNsfer AnD non-LTE population Algorithm (GRANADA). This model is able to compute non-LTE populations for vibrational, rotational, spin (i.e., NO and OH), and electronic (i.e., O2) states in a given planetary atmosphere. The model is very flexible and can be used for computing very accurate non-LTE populations or for calculating reasonably accurate but at high speed non-LTE populations in order to implement it into non-LTE remote sensing retrievals. We describe the model in detail and present an update of the non-LTE collisional processes and their rate coefficients for the most important molecules in Earth''s atmosphere. In addition, we have applied the model to the most important atmospheric infrared emitters including 13 species (H2O, CO2, O3, N2O, CO, CH4, O2, NO, NO2, HNO3, OH, N2, and HCN) and 460 excited vibrational or electronic energy levels. Non-LTE populations for all these energy levels have been calculated for 48 reference atmospheres expanding from the surface up to 200km, including seasonal (January, April, July and October), latitudinal (75°S, 45°S, 10°S, 10°N, 45°N, 75°N) and diurnal (day and night) coverages. The effects of the most recent updates of the non-LTE collisional parameters on the non-LTE populations are briefly described. This climatology is available online to the community and it can be used for estimating non-LTE effects at specific conditions and for testing and validation studies. [Copyright &y& Elsevier]

Abstract: We present a novel transmission expansion planning model that explicitly considers a multi-year planning horizon. The proposed formulation is a mixed-integer linear problem that can be solved using commercially available software. The algorithm determines the best overall transmission expansion plan, reflecting both investment and operational costs. To analyze the results obtained from the model, we use a set of efficiency metrics to appraise the effect of the expansion among generators and demands, as well as congestion metrics to measure changes in nodal prices and line congestions. This paper also provides a case study that considers the expansion plan for a transmission system based on the one of mainland Spain. [Copyright &y& Elsevier]

Abstract: Land cover monitoring using remotely sensed data requires robust classification methods which allow for the accurate mapping of complex land cover and land use categories. Random forest (RF) is a powerful machine learning classifier that is relatively unknown in land remote sensing and has not been evaluated thoroughly by the remote sensing community compared to more conventional pattern recognition techniques. Key advantages of RF include: their non-parametric nature; high classification accuracy; and capability to determine variable importance. However, the split rules for classification are unknown, therefore RF can be considered to be black box type classifier. RF provides an algorithm for estimating missing values; and flexibility to perform several types of data analysis, including regression, classification, survival analysis, and unsupervised learning. In this paper, the performance of the RF classifier for land cover classification of a complex area is explored. Evaluation was based on several criteria: mapping accuracy, sensitivity to data set size and noise. Landsat-5 Thematic Mapper data captured in European spring and summer were used with auxiliary variables derived from a digital terrain model to classify 14 different land categories in the south of Spain. Results show that the RF algorithm yields accurate land cover classifications, with 92% overall accuracy and a Kappa index of 0.92. RF is robust to training data reduction and noise because significant differences in kappa values were only observed for data reduction and noise addition values greater than 50 and 20%, respectively. Additionally, variables that RF identified as most important for classifying land cover coincided with expectations. A McNemar test indicates an overall better performance of the random forest model over a single decision tree at the 0.00001 significance level. [Copyright &y& Elsevier]

Abstract: This paper describes an application of the Strength Pareto Evolutionary Algorithm to the multi-objective optimization of a stand-alone PV–wind–diesel system with batteries storage. The objectives to be minimized are the levelized cost of energy (LCOE) and the equivalent carbon dioxide (CO2) life cycle emissions (LCE). Each solution of the Pareto front is a possible solution for the PV–wind–diesel-batteries system, which can supply the load, but each one has a different LCOE and LCE. Some solutions have low LCOE values, but high LCE values, and vice versa. Results show that the photovoltaic (PV) generator is the most important source of electrical energy for stand-alone systems in Spain and Southern Europe, not only environmentally, but also economically. In some cases, PV is almost the only source of energy, as some solutions of the best Pareto front do not include wind turbines and diesel generators run only a few hours during the year. [Copyright &y& Elsevier]

Abstract: This paper presents a signal analysis methodology to validate (detect) and reconstruct the missing and false data of a large set of flow meters in the telecontrol system of a water distribution network. The proposed methodology is based on two time-scale forecasting models: a daily model based on a ARIMA time series, while the 10-min model is based on distributing the daily flow using a 10-min demand pattern. The demand patterns have been determined using two methods: correlation analysis and an unsupervised fuzzy logic classification, named LAMDA algorithm. Finally, the proposed methodology has been applied to the Barcelona water distribution network, providing very good results. [Copyright &y& Elsevier]

Abstract: This paper presents a decision support system for automatic keep-clear signs management. The system consists of several modules. First of all, an acquisition module obtains images using a vehicle equipped with two recording cameras. A recognition module, which is based on Support Vector Machines (SVMs), analyzes each image and decides if there is a keep-clear sign in it. The images with keep-clear signs are included into a Geographical Information System (GIS) database. Finally in the management module, the data in the GIS are compared with the council database in order to decide actions such as repairing or reposition of signs, detection of possible frauds etc. We present the first tests of the system in a Spanish city (Meco, Madrid), where the systems is being tested for its application in the near future. [Copyright &y& Elsevier]

Abstract: In this paper we present a segmentation algorithm, inspired from an image-processing region-merging algorithm, for the delineation of discrete contiguous management zones in agriculture. The algorithm is unique in that it is applicable to high- or low-density irregular data sets, such as yield data. The algorithm is described and a brief example presented using unprocessed sensor-derived grain yield data. A comparison between the segmentation algorithm and a common classification algorithm (k-means clustering) was done using an aerial normalised differences vegetation index (NDVI) image collected on a 200ha vineyard in Olite, Southern Navarre, Spain. Classification was performed as a univariate (NDVI) analysis and a spatially constrained analysis. Segmentation and classification were run to find 2, 4, 6, …, 24 levels and the effectiveness of the outputs determined by how well it explained the variance in vine trunk circumference, a correlated but independent measurement. The results obtained demonstrated that for a given number of manageable (effective) zones the segmentation outputs were equivalent or superior to the classification outputs for partitioning vine circumference variance. The segmentation output also generated more coherent management units that should facilitate differential management. The algorithm presented is a first generation segmentation algorithm and several aspects still need to be developed, in particular methods for eliminating edge effects and converting management zones into management (treatment) classes. The results of the segmentation algorithm presented here would indicate that with further development, segmentation might provide an alternative and possibly preferable approach to delineating management zones. [Copyright &y& Elsevier]

Abstract: Since 2002, the Galician Administration (NW Spain) has been working in the development of a decision support system for Agricultural Production Planning. To this end, an initial decision tool, Gestop®, was developed. However, the tool had a number of limitations for practical use because of the lack of differentiated data about the real situation of farms. With a view to improving the initial tool, a combinatorial method (CM) was developed. Such a combinatorial method used official census data to classify farms based on the different production processes, sizes and geographical locations of farms, among other factors. The combinatorial method proved very useful as a source of differentiated data for the decision support system. Yet, validating the results using robust statistical methods was considered essential in order to guarantee the suitability of the model for use. This paper uses discriminant analysis (DA) techniques to validate the types of dairy farms obtained from the combinatorial method for Galicia. Results clearly validate the combinatorial method and its use as a data source for decision support systems in Agricultural Production Planning. [Copyright &y& Elsevier]

Abstract: The validation of coarse satellite-derived products from field measurements generally relies on up-scaling the field data to the corresponding satellite products. This question is commonly addressed through the generation of a reference high-resolution map of an area covering several moderate resolution pixels. This paper describes a method by which reference leaf area index (LAI) maps can be generated from ground-truth LAI measurements. The method is based on a multivariate ordinary least squares (OLS) algorithm which uses an iteratively re-weighted least squares (IRLS) algorithm. It provides an empirical relationship (i.e. a transfer function) between in situ measurements and concomitant radiance values from high-resolution HRVIR2/SPOT4 imagery. The band combination composed of green (G), red (R) and near infrared (NIR) proved to be appropriate for LAI prediction, with a cross-validation error lower than 0.7. The convex hull technique has been proposed to assess the extrapolation error of the up-scaling model. Although the model showed to be sensitive to pixels outside the convex hull, the error associated with extrapolated pixels was moderate (i.e. root mean square error (RMSE)≈0.3). A close agreement was found between LAI maps derived from different field based datasets despite the notable differences in the sampling constraints imposed by the use of different instruments. It confirms the validity of a stratified random sampling strategy and suggests the suitability of using around 50 elementary sampling units (ESUs) to characterize the study area. A sensitivity analysis has confirmed that this method is suited to capture the variability across the site extent and, at the same time, minimize field efforts. Knowledge gained from this work provides guidance for an efficient sampling strategy of LAI in a cropland area. [Copyright &y& Elsevier]

This paper shows the results of a low level controller that has been applied to an autonomous robotic system using a WiFi-based Partially Observable Markov Decision Process (POMDP). These observations provide a clue for global robot localization from the first iteration of the POMDP algorithm. Due to the noise channel of WiFi measures, it becomes necessary to make observations at a well known location within the environment. Therefore, a robust local navigator is needed in order to place the robot in an optimal position for making the observation. The system has been tested with two Pioneer 2AT robots in the premises of the Department of Electronics at the University of Alcalá. [Copyright &y& Elsevier]

The blind separation problem of linear time-dependent mixtures is addressed in this paper. We have developed a new algorithm based on the minimization of the mutual information as well as a penalized term which ensures an a priori normalization of the estimated sources (outputs) and so, avoids the scale indeterminacy. The criterion minimization is done using a well-known gradient approach. Finally, some numerical results are shown to illustrate the performance of the penalized algorithm compared to the Babaie-Zadeh approach presented in (Proceedings of IWANN, Granada, Spain, June 2001, pp. 834–842). [Copyright &y& Elsevier]