<p>The problem of parametric estimation in photovoltaic (PV) modules considering manufacturer information is addressed in this research from the perspective of combinatorial optimization. With the data sheet provided by the PV manufacturer, a non-linear non-convex optimization problem is formulated that contains information regarding maximum power, open-circuit, and short-circuit points. To estimate the three parameters of the PV model (i.e., the ideality diode factor (a) and the parallel and series resistances (R<sub>p</sub> and R<sub>s</sub>)), the crow search algorithm (CSA) is employed, which is a metaheuristic optimization technique inspired by the behavior of the crows searching food deposits. The CSA allows the exploration and exploitation of the solution space through a simple evolution rule derived from the classical PSO method. Numerical simulations reveal the effectiveness and robustness of the CSA to estimate these parameters with objective function values lower than 1 × 10<sup>−28</sup> and processing times less than 2 s. All the numerical simulations were developed in MATLAB 2020a and compared with the sine-cosine and vortex search algorithms recently reported in the literature.</p>
The seagull optimization algorithm (SOA) is a novel swarm intelligence algorithm proposed in recent years. The algorithm has some defects in the search process. To overcome the problem of poor convergence accuracy and easy to fall into local optimality of seagull optimization algorithm, this paper proposed a new variant SOA based on individual disturbance (ID) and attraction-repulsion (AR) strategy, called IDARSOA, which employed ID to enhance the ability to jump out of local optimum and adopted AR to increase the diversity of population and make the exploration of solution space more efficient. The effectiveness of the IDARSOA has been verified using representative comprehensive benchmark functions and six practical engineering optimization problems. The experimental results show that the proposed IDARSOA has the advantages of better convergence accuracy and a strong optimization ability than the original SOA.
AbstractIn this work, a Multi-Objective Evolutionary Algorithm (MOEA) is developed to identify Functional Dependencies (FDEPs) in Complex Technical Infrastructures (CTIs) from alarm data. The objectives of the search are the maximization of a measure of novelty, which drives the exploration of the solution space avoiding to get trapped in local optima, and of a measure of dependency among alarms, which drives the uncovering of functional dependencies. The main contribution of the work is the direct identification of patterns of dependent alarms; this avoids going through the preliminary step of mining association rules, as typically done by state-of-the-art methods which, however, fail to identify rare functional dependencies due to the need of setting a balanced minimum occurrence threshold. The proposed framework for FDEPs identification is applied to a synthetic alarm database generated by a simulated CTI model and to a real large-scale database of alarms collected at the CTI of CERN (European Organization for Nuclear Research). The obtained results show that the framework enables the thorough exploration of the solution space and captures also rare functional dependencies.
One of the most common metrics to evaluate neural network classiﬁers is the area under the receiver operating characteristic curve (AUC). However, optimisation of the AUC as the loss function during network training is not a standard procedure. Here we compare minimising the cross-entropy (CE) loss and optimising the AUC directly. In particular, we analyse the loss function landscape (LFL) of approximate AUC (appAUC) loss functions to discover the organisation of this solution space. We discuss various surrogates for AUC approximation and show their differences. We ﬁnd that the characteristics of the appAUC landscape are signiﬁcantly different from the CE landscape. The approximate AUC loss function improves testing AUC, and the appAUC landscape has substantially more minima, but these minima are less robust, with larger average Hessian eigenvalues. We provide a theoretical foundation to explain these results. To generalise our results, we lastly provide an overview of how the LFL can help to guide loss function analysis and selection.
The problem of the electrical characterization of single-phase transformers is addressed in this research through the application of the crow search algorithm (CSA). A nonlinear programming model to determine the series and parallel impedances of the transformer is formulated using the mean square error (MSE) between the voltages and currents measured and calculated as the objective function. The CSA is selected as a solution technique since it is efficient in dealing with complex nonlinear programming models using penalty factors to explore and exploit the solution space with minimum computational effort. Numerical results in three single-phase transformers with nominal sizes of 20 kVA, 45 kVA, 112.5 kVA, and 167 kVA demonstrate the efficiency of the proposed approach to define the transformer parameters when compared with the large-scale nonlinear solver fmincon in the MATLAB programming environment. Regarding the final objective function value, the CSA reaches objective functions lower than 2.75×10−11 for all the simulation cases, which confirms their effectiveness in minimizing the MSE between real (measured) and expected (calculated) voltage and current variables in the transformer.
Message passing algorithms, whose iterative nature captures well complicated interactions among interconnected variables in complex systems and extracts information from the fixed point of iterated messages, provide a powerful toolkit in tackling hard computational tasks in optimization, inference, and learning problems. In the context of constraint satisfaction problems (CSPs), when a control parameter (such as constraint density) is tuned, multiple threshold phenomena emerge, signaling fundamental structural transitions in their solution space. Finding solutions around these transition points is exceedingly challenging for algorithm design, where message passing algorithms suffer from a large message fluctuation far from convergence. Here we introduce a residual-based updating step into message passing algorithms, in which messages varying large between consecutive steps are given a high priority in updating process. For the specific example of model RB, a typical prototype of random CSPs with growing domains, we show that our algorithm improves the convergence of message updating and increases the success probability in finding solutions around the satisfiability threshold with a low computational cost. Our approach to message passing algorithms should be of value for exploring their power in developing algorithms to find ground-state solutions and understand the detailed structure of solution space of hard optimization problems.
Genome-scale metabolic models are frequently used in computational biology. They offer an integrative view on the metabolic network of an organism without the need to know kinetic information in detail. However, the huge solution space which comes with the analysis of genome-scale models by using, e.g., Flux Balance Analysis (FBA) poses a problem, since it is hard to thoroughly investigate and often only an arbitrarily selected individual flux distribution is discussed as an outcome of FBA. Here, we introduce a new approach to inspect the solution space and we compare it with other approaches, namely Flux Variability Analysis (FVA) and CoPE-FBA, using several different genome-scale models of lactic acid bacteria. We examine the extent to which different types of experimental data limit the solution space and how the robustness of the system increases as a result. We find that our new approach to inspect the solution space is a good complementary method that offers additional insights into the variance of biological phenotypes and can help to prevent wrong conclusions in the analysis of FBA results.
PurposeTravel time at inter-stops is a set of important parameters in bus timetabling, which is usually assumed to be normal (log-normal) random variable in literature. With the development of digital technology and big data analytics ability in the bus industry, practitioners prefer to generate deterministic travel time based on the on-board GPS data under maximum probability rule and mean value rule, which simplifies the optimization procedure, but performs poorly in the timetabling practice due to the loss of uncertain nature on travel time. The purpose of this study is to propose a GPS-data-driven bus timetabling approach with consideration of the spatial-temporal characteristic of travel time.Design/methodology/approachThe authors illustrate that the real-life on-board GPS data does not support the hypothesis of normal (log-normal) distribution on travel time at inter-stops, thereby formulating the travel time as a scenario-based spatial-temporal matrix, where K-means clustering approach is utilized to identify the scenarios of spatial-temporal travel time from daily observation data. A scenario-based robust timetabling model is finally proposed to maximize the expected profit of the bus carrier. The authors introduce a set of binary variables to transform the robust model into an integer linear programming model, and speed up the solving process by solution space compression, such that the optimal timetable can be well solved by CPLEX.FindingsCase studies based on the Beijing bus line 628 are given to demonstrate the efficiency of the proposed methodology. The results illustrate that: (1) the scenario-based robust model could increase the expected profits by 15.8% compared with the maximum probability model; (2) the scenario-based robust model could increase the expected profit by 30.74% compared with the mean value model; (3) the solution space compression approach could effectively shorten the computing time by 97%.Originality/valueThis study proposes a scenario-based robust bus timetabling approach driven by GPS data, which significantly improves the practicality and optimality of timetable, and proves the importance of big data analytics in improving public transport operations management.
This paper presents a novel mathematical formulation in crew scheduling, considering real challenges most railway companies face such as roundtrip policy for crew members joining from different crew depots and stricter working time standards under a sustainable development strategy. In China, the crew scheduling is manually compiled by railway companies respectively, and the plan quality varies from person to person. An improved genetic algorithm is proposed to solve this large-scale combinatorial optimization problem. It repairs the infeasible gene fragments to optimize the search scope of the solution space and enhance the efficiency of GA. To investigate the algorithm’s efficiency, a real case study was employed. Results show that the proposed model and algorithm lead to considerable improvement compared to the original planning: (i) Compared with the classical metaheuristic algorithms (GA, PSO, TS), the improved genetic algorithm can reduce the objective value by 4.47%; and (ii) the optimized crew scheduling plan reduces three crew units and increases the average utilization of crew unit working time by 6.20% compared with the original plan.
Genome-scale metabolic models (GEMs) have been widely used for the phenotypic prediction of microorganisms. However, the lack of other constraints in the stoichiometric model often leads to a large metabolic solution space being inaccessible. Inspired by previous studies that take an allocation of macromolecule resources into account, we developed a simplified Python-based workflow for constructing enzymatic constrained metabolic network model (ECMpy) and constructed an enzyme-constrained model for Escherichia coli (eciML1515) by directly adding a total enzyme amount constraint in the latest version of GEM for E. coli (iML1515), considering the protein subunit composition in the reaction, and automated calibration of enzyme kinetic parameters. Using eciML1515, we predicted the overflow metabolism of E. coli and revealed that redox balance was the key reason for the difference between E. coli and Saccharomyces cerevisiae in overflow metabolism. The growth rate predictions on 24 single-carbon sources were improved significantly when compared with other enzyme-constrained models of E. coli. Finally, we revealed the tradeoff between enzyme usage efficiency and biomass yield by exploring the metabolic behaviours under different substrate consumption rates. Enzyme-constrained models can improve simulation accuracy and thus can predict cellular phenotypes under various genetic perturbations more precisely, providing reliable guidance for metabolic engineering.