Farid Bourennani

Optimal Design Methods for Hybrid Renewable Energy Systems

Renewable and hybrid energy systems (HESs) are expanding due to environmental concerns of climate change, air pollution, and depleting fossil fuels. Moreover, HESs can be cost effective in comparison with conventional power plants. This article reviews current methods for designing optimal HESs. The survey shows these systems are often developed on a medium scale in remote areas and stand-alone, but there is a global growing interest for larger scale deployments that are grid connected. Examples of HESs are PV–wind–battery and PV–diesel–battery. PV and wind energy sources are the most widely adopted. Diesel and batteries are often used but hydrogen is increasing as a clean energy carrier. The design of an efficient HES is challenging because HES models are nonlinear, non-convex, and composed of mixed-type variables that cannot be solved by traditional optimization methods. Alternatively, two types of approaches are typically used for designing optimal HESs: simulation-based optimization and metaheuristic optimization methods. Simulation-based optimization methods are limited in view of human intervention that makes them tedious, time consuming, and error prone. Metaheuristics are more efficient because they can handle automatically a range of complexities. In particular, multi-objective optimization (MOO) metaheuristics are the most appropriate for optimal HES because HES models involve multiple objectives at the same time such as cost, performance, supply/demand management, grid limitations, and so forth. This article shows that the energy research community has not fully utilized state-of-the-art MOO metaheuristics. More recent MOO metaheuristics could be used such as robust optimization and interactive optimization.

Computing opposition by involving entire population

The capabilities of evolutionary algorithms (EAs) in solving nonlinear and non-convex optimization problems are significant. Among the many types of methods, differential evolution (DE) is an effective population-based stochastic algorithm, which has emerged as very competitive. Since its inception in 1995, many variants of DE to improve the performance of its predecessor have been introduced. In this context, opposition-based differential evolution (ODE) established a novel concept in which, each individual must compete with its opposite in terms of the fitness value in order to make an entry in the next generation. The generation of opposite points is based on the populations current extreme points (i.e., maximum and minimum) in the search space; these extreme points are not proper representatives for whole population, compared to centroid point which is inclusive regarding all individuals in the population. This paper develops a new scheme that utilizes the centroid point of a population to calculate opposite individuals. Therefore, the classical scheme of an opposite point is modified accordingly. Incorporating this new scheme into ODE leads to an enhanced ODE that is identified as centroid opposition-based differential evolution (CODE). The performance of the CODE algorithm is comprehensively evaluated on well-known complex benchmark functions and compared with the performance of conventional DE, ODE, and some other state-of-the-art algorithms (such as SaDE, ADE, SDE, and jDE) in terms of solution accuracy. The results for CODE are promising.

Centroid Opposition-Based Differential Evolution

The capabilities of evolutionary algorithms (EAs) in solving nonlinear and non-convex optimization problems are significant. Differential evolution (DE) is an effective population-based EA, which has emerged as very competitive. Since its inception in 1995, multiple variants of DE have been proposed with higher performance. Among these DE variants, opposition-based differential evolution (ODE) established a novel concept in which individuals must compete with theirs opposites in order to make an entry in the next generation. The generation of opposite points is based on the current extreme points (i.e., maximum and minimum) in the search space. This paper develops a new scheme that utilizes the centroid point of a population to calculate opposite individuals. The classical scheme of an opposite point is modified. Incorporating this new scheme into DE leads to an enhanced ODE that is identified as centroid opposition-based differential evolution (CODE). The accuracy of the CODE algorithm is comprehensively evaluated on well-known complex benchmark functions and compared with the performance of conventional DE, ODE, and other state-of-the-art algorithms. The results for CODE are found to be promising.

Visual integration tool for heterogeneous data type by unified vectorization

Data integration is the problem of combining data residing at different sources, and providing the user with a unified view of these data. One of the critical issues of data integration is the detection of similar entities based on the content. This complexity is due to three factors: the data type of the databases are heterogenous, the schema of databases are unfamiliar and heterogenous as well, and the amount of records is voluminous and time consuming to analyze. As solution to these problems we extend our work in another of our papers by introducing a new measure to handle heterogenous textual and numerical data type for co-incident meaning extraction. Firstly, to in order accommodate the heterogeneous data types we propose a new weight called Bin Frequency - Inverse Document Bin Frequency (BF-IDBF) for effective heterogeneous data pre-processing and classification by unified vectorization. Secondly in order to handle the unfamiliar data structure, we use the unsupervised algorithm Self-Organizing Map. Finally to help the user to explore and browse the semantically similar entities among the copious amount of data, we use a SOM based visualization tool to map the database tables based on their semantical content.

Artificial Neural Networks for Earthquake Anomaly Detection

Earthquakes are natural disasters caused by an unexpected release of seismic energy from extreme levels of stress within the earth’s crust. Over the years, earthquake prediction has been a controversial research subject that has challenged even the smartest of minds. Because numerous seismic precursors and other factors exist that may indicate the potential of an earthquake occurring, it is extremely difficult to predict the exact time, location, and magnitude of an impending quake. Nevertheless, evaluating a combination of these precursors through advances in Artificial Intelligence (AI) can certainly increase the possibility of predicting an earthquake. The sole purpose for predicting a seismic event at a pre-determined locality is to provide substantial time for the citizens to take precautionary measures. With this in mind, Artificial Neural Networks (ANNs) have been promising techniques for the detection and prediction of locally impending earthquakes based on valid seismic information. To highlight the recent trends in earthquake abnormality detection, including various ideas and applications, in the field of Neural Networks, valid papers related to ANNs are reviewed and presented herein.

Optimal Photovoltaic System Design with Multi-Objective Optimization

Recently, several parts of the world suffer from electrical black-outs due to high electrical demands during peak hours. Stationary photovoltaic PV collector arrays produce clean and sustainable energy especially during peak hours which are generally day time. In addition, PVs do not emit any waste or emissions, and are silent in operation. The incident energy collected by PVs is mainly dependent on the number of collector rows, distance between collector rows, dimension of collectors, collectors inclination angle and collectors azimuth, which all are involved in the proposed modeling in this article. The objective is to achieve optimal design of a PV farm yielding two conflicting objectives namely maximum field incident energy and minimum of the deployment cost. Two state-of-the-art multi-objective evolutionary algorithms MOEAs called Non-dominated Sorting Genetic Algorithm-II NSGA-II and Generalized Differential Evolution Generation 3 GDE3 are compared to design PV farms in Toronto, Canada area. The results are presented and discussed to illustrate the advantage of utilizing MOEA in PV farms design and other energy related real-world problems.

OGDE3: Opposition-Based Third Generalized Differential Evolution

Multi-Objective Optimization (MOO) metaheuristics are commonly used for solving complex MOO problems characterized by non-convexity, multimodality, mixed-types variables, non-linearity, and other complexities. However, often metaheuristics suffer from slow convergence. Opposition-Based Learning (OBL) has been successfully used in the past for acceleration of single-objective metaheuristics. The most successful example in this regard is Opposition-based Differential Evolution (ODE). However, OBL was not fully explored for MOO metaheuristics. Therefore, in this paper, to the best of our knowledge, for the first time OBL is successfully adapted for a MOO metaheuristic by using a single population (no coevolution). The proposed MOO metaheuristic is based on the GDE3 method and it is called Opposition-based GDE3 (OGDE3). OGDE3 utilizes OBL for opposition-based population initialization and self-adaptive oppositionbased generating jumping. Furthermore, the new algorithm is compared with seven state-of-the-art MOO metaheuristics using the ZDT test suite. OGDE3 outperformed the other algorithms; the results are explained and discussed in detail.

Leaders and speed constraint multi-objective particle swarm optimization

The particle swarm optimization (PSO) algorithm has been very successful in single objective optimization as well as in multi-objective (MO) optimization. However, the selection of representative leaders in MO space is a challenging task. Most previous MO-based PSOs used exclusively the concept of non-dominance to select leaders which might slow down the search process if the selected leaders are concentrated in a specific region of the objective space. In this paper, a new restriction mechanism is added to non-dominance in order to select leaders in more representative (distributed) way. The proposed algorithm is named leaders and speed constrained multi-objective PSO (LSMPSO) which is an extended version of SMPSO. The convergence speed of LSMPSO is compared to state-of-the-art metaheuristics, namely, NSGA-II, SPEA2, GDE3, SMPSO, AbYSS, MOCell, and MOEA/D. The ZDT and DTLZ family problems are utilized for the comparisons. The proposed LSMPSO algorithm outperformed the other algorithms in terms of convergence speed.

MODEL: Multi-objective differential evolution with leadership enhancement

Differential Evolution (DE) has been successfully used to solve various complex optimization problems; however, it can suffer depending of the complexity of the problem from slow convergence due to its iterative process. The use of the leadership concept was efficiently utilized for the acceleration of Particle Swarm Optimization (PSO) in a single-objective space. The generalization of the leadership concept in multi-objective space is not trivial. Furthermore, despite the efficiency of using the leadership concept, a limited number of multi-objective metaheuristics utilize it. To address these challenges, this paper incorporates the concept of leadership in a multi-objective variant of DE by introducing it into the mutation scheme. The preliminary results are promising as MODEL outperformed the parent algorithm GDE3 and showed the highest accuracy when compared with seven other algorithms.

Clustering Relational Database Entities Using K-means

The fast evolution of hardware and the internet made large volumes of data more accessible. This data is composed of heterogeneous data types such as text, numbers, multimedia, and others. Non-overlapping research communities work on processing homogeneous data types. Nevertheless, from the user perspective, these heterogeneous data types should behave and be accessed in a similar fashion. Processing heterogeneous data types, which is Heterogeneous Data Mining (HDM), is a complex task. However, the HDM by Unified Vectorization (HDM-UV) seems to be an appropriate solution for this problem because it permits to process the heterogeneous data types simultaneously. In this paper, we use K-means and Self-Organizing Maps for simultaneously processing textual and numerical data types by UV. We evaluate how the HDM-UV improves the clustering results of these two algorithms (SOM, K-means) by comparing them to the traditional homogeneous data processing. Furthermore, we compare the clustering results of the two algorithms applied to a data integration problem.