Laura Cornejo-Bueno

A grouping genetic algorithm — Extreme learning machine approach for optimal wave energy prediction

In this paper we propose an approach for feature selection in a problem of significant wave height prediction, to improve the exploitation of marine energy. The method that we present, a Grouping Genetic Algorithm - Extreme Learning Machine approach (GGA-ELM), mainly tries to improve the prediction performance of the regressors, providing more effective predictors and good performance in the final significant wave height prediction. In this method, the GGA looks for several subsets of features, and the ELM provides the fitness of the algorithm, through its accuracy on significant wave height prediction. The GGA is able to evolve different groups of features in parallel, which may improve the performance of the prediction obtained. After the feature selection process with the GGA-ELM, the final results are obtained by applying an ELM and also by a Support Vector Regressor algorithm, both working on the best GGA groups of features previously evolved. In the experimental part of the paper, we show the performance of the proposed approach in a real problem of significant wave height prediction at the West Coast of the USA, using variables directly obtained from several measuring buoys.

Bayesian optimization of a hybrid system for robust ocean wave features prediction

In the last years, Bayesian optimization (BO) has emerged as a practical tool for high-quality parameter selection in prediction systems. BO methods are useful for optimizing black-box objective functions that either lack an analytical expression, or are very expensive to evaluate. In this paper, we show that BO can be used to obtain the optimal parameters of a prediction system for problems related to ocean wave features prediction. Specifically, we propose the Bayesian optimization of a hybrid Grouping Genetic Algorithm for attribute selection combined with an Extreme Learning Machine (GGA-ELM) approach for prediction. The system uses data from neighbor stations (usually buoys) in order to predict the significant wave height and the wave energy flux at a goal marine structure facility. The proposed BO methodology has been tested in a real problem involving buoys data in the Western coast of the USA, improving the performance of the GGA-ELM without a BO approach.

Combining Reservoir Computing and Over-Sampling for Ordinal Wind Power Ramp Prediction

Wind power ramp events (WPREs) are strong increases or decreases of wind speed in a short period of time. Predicting WPREs is of vital importance given that they can damage the turbines in a wind farm. In contrast to previous binary approaches (ramp versus non-ramp), a three-class prediction is proposed in this paper by considering: negative ramp, non-ramp and positive ramp, where the natural order of the events is clear. The independent variables used for prediction include past ramp function values and meteorological data obtained from physical models (reanalysis data). The proposed methodology is based on reservoir computing and an over-sampling process for alleviating the high degree of unbalance of the dataset (non-ramp events are much more frequent than ramps). The reservoir computing model is a modified echo state network composed by: a recurrent neural network layer, a nonlinear kernel mapping and an ordinal logistic regression, in such a way that the order of the classes can be exploited. The standard synthetic minority oversampling technique (SMOTE) is applied to the reservoir activations, given that the direct application over the input variables would damage its temporal structure. The performance of this proposal is compared to the original dataset (without over-sampling) and to nominal logistic regression, and the results obtained with the oversampled dataset and ordinal logistic regression are found to be more robust.

A Hybrid Neuro-Evolutionary Algorithm for Wind Power Ramp Events Detection

In this work, a hybrid system for wind power ramps events prediction in wind farms is proposed. The system is based on modelling the prediction problem as a binary classification problem from atmospheric reanalysis data inputs. On the other hand, a hybrid neuro-evolutive algorithm is proposed, which combines Artificial Neuronal Networks such as Extreme Learning Machines, with evolutionary algorithms to optimize the trained models. The phenomenon under study occurs with a very low probability, for this reason the problem is so unbalanced, and it is necessary to resort to techniques focused on obtain good results by means of a reduction of the samples from the majority class, as the SMOTE approach. A feature selection is performed by the evolutionary algorithm in order to choose the best trained model. Finally, this model is evaluated by a test set and its accuracy performance is given. The accuracy obtained in the results is quite good in terms of classification performance.

Optimal placement of distributed generation in micro-grids with binary and integer-encoding evolutionary algorithms

This paper discuses the performance of two different Evolutionary Algorithms (EAs) in a problem of Optimal Placement of Distributed Power Generation (OPDPG) in Micro-Grids (MGs). Specifically, the problem consists of choosing the node/nodes to locate a number of different distributed generators with different technologies (such as micro wind turbines, photovoltaic panels, etc.), in such a way that the electrical power losses along a given time period (T) in the MG are minimized. We consider a situation where the network topology is already defined and where each node can have a load with different profiles allocated. The consumption profiles are real measurements of different types (residential, industrial, etc.) and will be hourly evaluated. The generations profiles are also real measurement data from different generation technologies. We consider two different encodings the EAs: first a binary-encoding approach, where each wind generator is represented by 2 bits and each solar generator by N bits, where N is the number of nodes that form the MG; and second, an integer-encoding approach, where both wind and PV generators are represented by 1 and 4 integer elements, respectively. Experiments are performed by considering three different MG topologies, with different number of nodes, in order to test the behavior of the algorithms with search spaces of increasing size. In these experimental scenarios we show how the binary approach attains better solutions than the integer-encoding approach, tough the computational time of the former is higher.

Feature Selection with a Grouping Genetic Algorithm – Extreme Learning Machine Approach for Wind Power Prediction

This paper proposes a hybrid algorithm for feature selection in a Wind Power prediction problem, based on a Grouping Genetic Algorithm-Extreme Learning Machine (GGA-ELM) approach. The proposed approach follows the classical wrapper method where a global search algorithm looks for the best set of features which minimize the output of a given predictors. In this case a GGA searches for several subsets of features and the ELM provides the fitness of the algorithm. Moreover, we propose to use variables from atmospheric reanalysis data as predictive inputs for the system, which opens the possibility of hybridizing numerical weather models with Machine Learning (ML) techniques for wind power prediction in real systems. The ERA-Interim reanalysis from the European Center for Medium-Range Weather Forecasts has been the one used in this paper. Specifically, after the process of feature selection, we have tested the ELM and Gaussian Processes (GPR) to solve the problem. Experimental evaluation of the prediction system in real data from three wind farms in Spain has been carried out, obtaining excellent prediction results when the ELM is applied after the feature selection but not enough in the case of the GPR algorithm.

Energy Flux Range Classification by Using a Dynamic Window Autoregressive Model

This paper tackles marine energy prediction from the classification point of view, by previously discretising the real objective variable into a set of consecutive categories or ranges. Given that the range of energy flux is enough to obtain an approximation of the amount of energy produced, the purpose of this discretisation is to simplify the prediction task. A special kind of autoregressive models are considered, where the category to be predicted depends on both the previous values of energy flux and a set of meteorological variables estimated by numerical models. Apart from this, this paper introduces two different ways of adjusting the order of the autoregressive models, one based on nested cross-validation and the other one based on a dynamic window. The results show that these kind of models are able to predict the time series in an acceptable way, and that the dynamic window procedure leads to the best accuracy without needing the additional computational cost of adjusting the order of the model.

Nested evolutionary algorithms for joint structure design and operation of micro-grids under variable electricity prices scenarios

This paper proposes to tackle the structure design and operation of a Micro-Grid in a jointly way, by means of a novel nested Evolutionary Algorithms (EAs) approach. Specifically, in an scenario of variable electricity prices in an hourly basis, we apply different EAs, nested, to obtain optimal values for the sizing of generators and Energy Storage System (ESS), also to obtain the optimal values for each access tariff periods (structure part of the MG), and ESS scheduling (operational part of the MG). The proposed nested EAs starts from an initial solution for the ESS scheduling given by a deterministic approach (DA algorithm), from which an initial structure part is obtained by means of a first evolution. This part is set, and a different EA is then applied to obtain an improved ESS scheduling, which will be set to apply a different EA for the structure part. This scheme is applied in a sequential fashion for a number of evolutions. We will show that the proposed evolution scheme is able to obtain excellent results in terms of MG design, better than those by a single EA with the same number of function evaluations.