Extremal Optimization Research Articles

Short-Term Wind Speed Forecasting Based on Ensemble Online Sequential Extreme Learning Machine and Bayesian Optimization

Short-term wind speed forecasting is crucial to the utilization of wind energy, and it has been employed widely in turbine regulation, electricity market clearing, and preload sharing. However, the wind speed has inherent fluctuation, and accurate wind speed prediction is challenging. This paper aims to propose a hybrid forecasting approach of short-term wind speed based on a novel signal processing algorithm, a wrapper-based feature selection method, the state-of-art optimization algorithm, ensemble learning, and an efficient artificial neural network. Variational mode decomposition (VMD) is employed to decompose the original wind time-series into sublayer modes. The binary bat algorithm (BBA) is used to complete the feature selection. Bayesian optimization (BO) fine-tuned online sequential extreme learning machine (OSELM) is proposed to forecast the low-frequency sublayers of VMD. Bagging-based ensemble OSELM is proposed to forecast high-frequency sublayers of VMD. Two experiments were conducted on 10 min datasets from the National Renewable Energy Laboratory (NREL). The performances of the proposed model were compared with various representative models. The experimental results indicate that the proposed model has better accuracy than the comparison models. Among the thirteen models, the proposed VMD-BBA-EnsOSELM model can obtain the best prediction accuracy, and the mean absolute percent error (MAPE) is always less than 0.09.

Hybrid self-adaptive biobjective optimization of multiple robot scheduling problem for mixed-model assembly lines considering energy savings

With the wide application of robots in the material distribution process on the assembly lines, single robot scheduling cannot meet the actual production needs. However, the high degree of mechanization also brings about environmental problems. Therefore, this article aims to develop a scheduling methodology to accomplish material supply tasks using multiple robots with energy consumption consideration. Meanwhile, a targeted mathematical model to minimize total weighted penalty costs and total energy consumption is developed. Due to the NP-hard nature of the problem, an adaptive hybrid mutation population extremal optimization multi-objective algorithm based on uniform distribution selection is proposed to solve multi-objective problems. Furthermore, a new coding method for initialization is designed to optimize the whole iterative process. The performance of the proposed algorithm is evaluated by comparing with three benchmark multi-objective algorithms. Computational experiments are represented to prove the validity and feasibility of the proposed algorithm.

Nonlinearity in stock returns: Do risk aversion, investor sentiment and, monetary policy shocks matter?

This paper examines the nonlinear effect of risk aversion, monetary shocks and sentiment changes, on index returns for a set of developed countries. We employ a smooth transition model, in which, these leading stock return drivers are separately considered as a threshold variable. Results show that index returns fall as investors become more risk averse following a positive shock to the volatility index for most of the markets. A restrictive monetary policy negatively affects index returns, in the low regime for some countries, and such an effect turns to be larger in the high regime for more liquid markets. When investors exhibit extreme pessimism or optimism, the market smoothly switches from bear to bull states according to the heterogeneous reactions of market participants and the degree of compensation for taking additional trading risks. Assessment of the predictive performance provides compelling evidence for the supremacy of the model with monetary shocks as a transition variable over competing models.

Time-variant system reliability analysis method for a small failure probability problem

This paper proposes a time-variant system reliability analysis method by combining multiple response Gaussian process (MRGP) and subset simulation (SS) to solve the small failure probability problem. One common method for time-variant reliability analysis is based on the double-loop procedure where the inner loop is the optimization for extreme values and the outer loop is extreme-value-based reliability analysis. In this paper, a new single-loop strategy is firstly proposed to decouple the double-loop procedure by using the best value in current initial samples to approximate the extreme value, thus the extremal optimization in inner loop can be avoided. Then the MRGP model is used to construct the surrogate model of extreme value response surface for time-variant system reliability analysis based on the approximated extremums. Meanwhile, the Kriging model is also constructed based on the initial samples to assist in searching the new sample point. Furthermore, for selecting the new point that resides as close to the extreme value response surface as possible from the Monte Carlo simulation (MCS) sample pool, three learning functions (U-function, EFF-function and H-function) are respectively used to find the new random variable sample point based on the MRGP model and the expected improvement (EI) function is used to find the new time sample point based on the Kriging model. Finally, for reducing the size of candidate sample pool and the computing burden, the SS method is combined with the MRGP model to deal with the small failure probability problem. The effectiveness of the proposed method is also demonstrated by several examples.

Optimal design of robust resilient automatic voltage regulators

Uncertainties in the plant model parameters and perturbations in the controller gains imposed by implementation errors represent a challenge to ensure robust stability and controller non-fragility simultaneously. Optimal design of robust non-fragile proportional–integral–derivative (PID) controller is presented for an automatic voltage regulator (AVR). The PID design relies basically on Kharitonov theorem and optimization by future search algorithm (FSA). The proposed algorithm has low computational complexity and fast convergence rate because it utilizes both local and global search methods. Further, FSA can improve the exploration characteristic and prevent trapping in local optima by updating its random initial. The PID controller is optimized by FSA to cope with expected parametric uncertainties of the plant model and tolerate its gain perturbations such that robust stability and controller non-fragility are simultaneously met. An interval plant model is suggested to account for model uncertainties where only eight extreme plants derived by Kharitonov theorem are considered in design. FSA-based PID optimization is constrained by the stability conditions of Kharitonov’s plants derived using Routh–Hurwitz. A new figure-of-demerit (FoD) based performance index is suggested to enforce simultaneous minimization of the time domain specifications. The suggested objective function is represented by a weighted sum of FoD of nominal response and the sum of reciprocals of the perturbation radii of PID gains. The output results of the recommended design are compared to that of artificial bee colony (ABC) algorithm and teaching–learning based optimization (TLBO) algorithm, multi-objective extremal optimization (MOEO), and non-dominated sorting genetic algorithm II (NSGA II). The results can confirm better response of the suggested technique measured up to other techniques where robustness and non-fragility are simultaneously ensured.

Measurement of Overconfidence, Illusion of Control and Optimism in the Context of Behavioral Finance: Case of University Students

Behavioral finance is concerned with analyzing market anomalies as well as biases that cause these anomalies. There are many biases that are stated to have an effect on the decision processes of individuals in behavioural finance literature. It is noteworthy that the biases emphasized among them are overconfidence, illusion of control and extreme optimism. This study aimed at measuring the scale that consists of statements about these three biases on college students in Turkey and reliability and validity of this scale was tested. There are 19 items in the first version of the scale that aims to measure biases and used in this study. The questionnaire, including the demographic information of the students, was applied to a total of 522 students studying at different universities. As a result of the analyzes, it was determined that the final version of the scale Cronbach's Alpha value is 0.825 and the items in the scale were collected under 4 factors.

Short-term wind speed interval prediction based on artificial intelligence methods and error probability distribution

At present, environmental pollution, climate warming and other problems are becoming more and more serious. And wind energy is pollution-free and never be exhausted, so it can make a major contribution to the global energy transformation. However, its random fluctuations and uncertainties bring adverse effects to the power system and endanger the safety of the power grid. Therefore, this paper combines artificial intelligence methods with statistical knowledge, and proposes a new interval prediction model based on the Fast Correlation Based Filter (FCBF) algorithm, the optimized Radial Basis Function (RBF) model and Fourier distribution for wind speed. Firstly considering environmental factors, this paper studies multi-factor wind speed prediction and applies the FCBF algorithm to filter the factors that affect the wind change. After that, this paper applies the idea of the Extremal Optimization (EO) to improve the Particle Swarm Optimization (PSO) and constructs a new EPSO optimization model for optimizing the RBF model. Next, using the Fourier function to fit the error probability distribution, and the wind speed interval is estimated based on point prediction results. Finally, the actual data of Changma Wind Farm is used for experiments to verify the feasibility and effectiveness of the proposed model. And through experimental results and comparison, it can be concluded: (1) Using the FCBF algorithm to select input variables can reduce redundant variables and lay a good foundation for subsequent prediction; (2) Applying the constructed EPSO-RBF model to predict wind speed, and the maximum and average value of the prediction error are only 0.8430 m/s, 0.1749 m/s, which is significantly better than several other traditional neural network models; (3) Introducing the Fourier function into the wind speed interval prediction, even at the 80% confidence level, the average width of the interval prediction is less than 3 m/s, and the coverage rate is higher than 90%.

A single-loop strategy for time-variant system reliability analysis under multiple failure modes

This paper proposes a single-loop strategy for time-variant system reliability analysis by combining multiple response Gaussian process (MRGP) and Kriging model. Firstly, a new strategy is provided to decouple the double-loop strategy of extreme-value-based time-variant system reliability analysis, where the extreme value response in the double-loop strategy is approximated by the best value in the initial points to avoid the inner-loop extremal optimization. Then, the extreme value response surface is directly built based on the approximated extreme value using the MRGP model. Meanwhile, a Kriging model based on initial points is also proposed to search a new sample point. Furthermore, in order to update the initial sample points and improve the accuracy of the proposed strategy, three learning functions (U-function, EFF-function and H-function) and expected improvement (EI) function are combined to select a new point that resides as close to the extreme value response surface as possible. Finally, several examples are presented to demonstrate the effectiveness of the proposed strategy.

52‐4: Late‐News‐Paper: Development of 240Hz Refresh Rate ADS Mode Gaming Product

The ADS display mode shows significant advantages of high transmittance, wide viewing angle, small color shift at large view angle, etc., and thus has been widely used in all kinds of display product such as mobile phone, tablet, notebook, monitor and television. However, ADS mode shows shortcomings of slow response speed. In this work, we made extreme optimization research and breakthroughs in improving ADS mode response speed from the aspects of LC material, cell gap, and pixel ITO slit angle. Based on the research, 27 inch FHD 240Hz refresh rate gaming product was successfully developed. It has been achieved gray to gray average response time 1.9ms with over driving as well as contrast ratio 1214:1 fulfilling static CR requirements of VESA HDR standard.

Resilience Design of Healthcare Resources Supply Network Based on Self-Organized Criticality.

The healthcare resources supply network design for resilience is an effective way to deal with uncertainty disruption. In this article we propose a model of supply network self-organization evolution, and establish self-organized criticality as a cause of cascade failure. Our main purpose is to keep the system in a resilient range, i.e., critical state. A network structural design with smaller degree distribution exponent can achieve better absorptive capacity at macro level. An interactive rule design with extremal optimization has better adaptive capacity at micro level. Using macro statistic and indicator micro performance indicator, we demonstrate that our design can slow the development to a supercritical state and can improve the resilience of the supply network.

Optimization of Fractional Order Controllers for AVR System Using Distance and Levy-Flight Based Crow Search Algorithm

ABSTRACT This paper proposes a Distance and Levy-flight based Crow Search Algorithm (DLCSA) for optimization of three Fractional-order controllers: Fractional Order Proportional-Integral-Derivative (FOPID), Fractional Order Proportional-Integral (FOPI) and Fractional Order Proportional-Derivative (FOPD) controllers. Novelties of the proposed method are fourfold; (a) distance-based Flight length (Dfl)to improve the convergence rate, (b) Levy-flight based evasion to improve the efficiency of the algorithm, (c)adaptive awareness probability(AP) to set a tradeoff between local and global search and (d) Fuzzy based adaptive objective function to adjust the solution weights automatically. With such adaptations, the new approach preserves solution diversity and improves the convergence to global optima. To demonstrate the effectiveness of the contributions, proposed methods: DLCSA-FOPID, DLCSA-FOPI, and DLCSA-FOPD are implemented on an Automatic Voltage Regulator system and results are compared with few other well-established techniques like Zeigler–Nichols (Z-N), Artificial Bee Colony (ABC) Optimization (ACO), Multi-Objective Extremal Optimization (MOEO), Genetic Algorithm (GA), Chaos Ant Swarm (CAS) algorithm, Real Coded Extremal Optimization (RCEO), etc. Further to investigate robustness of proposed methods, the results are obtained for set-point tracking, noise suppression, load disturbance rejection, control effort, and modeling errors. The results demonstrate the better performance of the proposed method as compared to other state of the art techniques.

An Energy Efficient e-Healthcare Framework Supported by Novel EO-μGA (Extremal Optimization Tuned Micro-Genetic Algorithm)

The edge/fog computing has the potential to gear up the healthcare industry by providing better and faster health services to the patients. In healthcare systems where every second is crucial, the edge computing can be helpful to reduce the time between data capture and analytics in a powerful manner. In edge computing, the network edge devices are configured in such a manner that they can handle critical analysis and make necessary decisions instead of sending the captured health data directly to the cloud. However, lifetime of the edge network is a critical factor and thus an energy efficient network architecture has to be designed to achieve the above mentioned goal. In this regard, this research presents a new extremal optimization tuned micro genetic algorithm (EO-μGA) based clustering technique for the sake of efficient routing and prolonging network lifetime by saving the battery power of network edge devices. Moreover, a novel fitness function with a set of relevant criteria of edge devices such as energy factor, average intra-cluster distance, average distance to cluster leader over data analytics center, average sleeping time, and computational load has been considered for the selection of the cluster leader which will be responsible for managing intra-cluster and inter-cluster data communication. The simulation results show that the proposed EO-μGA based clustering model offers a higher network lifetime and a least amount of transmission energy consumption per node as compared to various state of the art optimization algorithms.

Technical Revisions Reveal Overconfidence in US and Canadian Reserves Estimates

Summary In this paper, we present methodology to quantify biases in reserves estimates using technical revisions (TRs) listed in reserves-reconciliation reports filed with regulators in the US and Canada. Using this methodology, we assessed the reliability of reserves estimates for 34 companies filing in Canada and 32 companies filing in the US from 2007 to 2017. Filers in both Canada and the US overestimated proved (1P) reserves, and US filers overestimated 1P reserves (51% positive TRs instead of 90%) more often than Canadian filers (72% positive TRs). Canadian filers underestimated proved-plus-probable (2P) reserves slightly (54% positive TRs instead of 50%). Considering the entire reserves distribution, Canadian filers were moderately overconfident (underestimated uncertainty) and slightly pessimistic. US filers, who report only 1P, were somewhere between the combination of extreme overconfidence and neutral directional bias (DB) and the combination of moderate overconfidence and extreme optimism. Three groups of professionals can benefit from this study: estimators, who can use the methodology to track their TRs over time, calibrate them, and use this information to improve future estimation procedures; investors, who can analyze reported reserves estimates to compare volumes fairly; and regulators, to whom the paper provides quantitative methodology to suggest to filers to help them ensure compliance with appropriate criteria for 1P and 2P reserves and avoid significant reserves write-downs later.

Ground State Properties of the Diluted Sherrington-Kirkpatrick Spin Glass

We present a numerical study of ground states of the dilute versions of the Sherrington-Kirkpatrick (SK) mean-field spin glass. In contrast to so-called "sparse" mean-field spin glasses that have been studied widely on random networks of finite (average or regular) degree, the networks studied here are randomly bond diluted to an overall density p, such that the average degree diverges as ∼pN with the system size N. Ground state energies are obtained with high accuracy for random instances over a wide range of fixed p. Since this is an NP-hard combinatorial problem, we employ the extremal optimization heuristic to that end. We find that the exponent describing the finite-size corrections ω varies continuously with p, a somewhat surprising result, as one would not expect that gradual bond dilution would change the T=0 universality class of a statistical model. For p→1, the familiar result of ω(p=1)≈2/3 for the SK model is obtained.

A Comparison of Extremal Optimization, Differential Evolution and Particle Swarm Optimization Methods for Well Placement Design in Groundwater Management

When designing a well field for efficiently extracting groundwater or petroleum, it is common for designers to rely on computational optimization methods to determine the optimal placement of wells. The goal of these methods is to find a well-field solution that maximizes the value of a defined objective function, and to do so while utilizing the least amount of computational effort. To achieve this, researchers have developed algorithms based on a wide range of heuristics. Within groundwater management, popular methods include particle swarm optimization (PSO) and genetic algorithms such as differential evolution (DE). This study seeks to investigate a recently developed method called Extremal Optimization for Well Placement Problems (EO-WPP), and to compare its performance with established methods like PSO and DE. EO-WPP is an optimization method based on the extremal optimization (EO) algorithm. EO optimizes by iteratively identifying and modifying the least effective components of a solution. By following this heuristic, the EO algorithm has the potential to quickly find optimal solutions while requiring minimal computational effort. To test this, the performance of DE, PSO and EO-WPP was compared on four benchmark problems. Two of these are the Rastrigin and the Rosenbrock benchmark functions. These functions were used because of their quick evaluation and their popularity in optimization literature. The third benchmark is a synthetic groundwater model, built to test the methods under the context of groundwater management. The final benchmark is a field problem using the Aberdeen groundwater model in South Dakota. The results reveal that EO-WPP was able to outperform DE and PSO on all tested benchmarks. EO-WPP is an effective and efficient optimization tool for well placement design in groundwater management.

An Integrated Shadow-Adjusted Snow-Aging Index for Alpine Regions

Detecting the variations in snow cover aging over undulating alpine regions is challenging owing to the complex snow-aging process and shadow effect from steep slopes. This study proposes a novel snow-cover status index, namely shadow-adjusted snow-aging index (SASAI), portraying the integrated aging process within the Manas River Basin in northwest China. The Environment Satellites HJ-1A/B optical images and in-field measurements were used during the snow ablation and accumulation periods. The in-field measurements provide a reference for building a candidate library of snow-aging indicators. The representative aging samples for training and validation were obtained using the proposed time-gap searching method combined with the target zones established based on the altitude of snowline. An analytic hierarchy process was used to determine the snow-aging index (SAI) using multiple optimal snow-aging indicators. After correction by the extreme value optimization algorithm, the SASAI was finally corrected for the effects of shading and assessed. This study provides both a flexible algorithm that indicates the characteristics of snow aging and speculation on the causes of the aging process. The separability of the SAI/SASAI and adaptability of this algorithm on multiperiod remote sensing images further demonstrates the applicability of the SASAI to all the alpine regions.

Automated proton treatment planning with robust optimization using constrained hierarchical optimization.

We present a method for fully automated generation of high quality robust proton treatment plans using hierarchical optimization. To fill the gap between the two common extreme robust optimization approaches, that is, stochastic and worst-case, a robust optimization approach based on the p-norm function is used whereby a single parameter, , can be used to control the level of robustness in an intuitive way. A fully automated approach to treatment planning using Expedited Constrained Hierarchical Optimization (ECHO) is implemented in our clinic for photon treatments. ECHO strictly enforces critical (inviolable) clinical criteria as hard constraints and improves the desirable clinical criteria sequentially, as much as is feasible. We extend our in-house developed ECHO codes for proton therapy and integrate it with a new approach for robust optimization. Multiple scenarios accounting for both setup and range uncertainties are included (13scenarios), and the maximum/mean/dose-volume constraints on organs-at-risk (OARs) and target are fulfilled in all scenarios. We combine the objective functions of the individual scenarios using the p-norm function. The p-norm with a parameter or result in the stochastic or the worst-case approach, respectively; an intermediate robustness level is obtained by employing -values in-between. While the worst-case approach only focuses on the worst-case scenario(s), the p-norm approach with a large p value ( ) resembles the worst-case approach without completely neglecting other scenarios. The proposed approach is evaluated on three head-and-neck (HN) patients and one water phantom with different parameters, . The results are compared against the stochastic approach (p-norm approach with ) and the worst-case approach, as well as the nonrobust approach (optimized solely on the nominal scenario). The proposed algorithm successfully generates automated robust proton plans on all cases. As opposed to the nonrobust plans, the robust plans have narrower dose volume histogram (DVH) bands across all 13 scenarios, and meet all hard constraints (i.e., maximum/mean/dose-volume constraints) on OARs and the target for all scenarios. The spread in the objective function values is largest for the stochastic approach ( ) and decreases with increasing toward the worst-case approach. Compared to the worst-case approach, the p-norm approach results in DVH bands for clinical target volume (CTV) which are closer to the prescription dose at a negligible cost in the DVH for the worst scenario, thereby improving the overall plan quality. On average, going from the worst-case approach to the p-norm approach with , the median objective function value across all the scenarios is improved by 15% while the objective function value for the worst scenario is only degraded by 3%. An automated treatment planning approach for proton therapy is developed, including robustness, dose-volume constraints, and the ability to control the robustness level using the p-norm parameter , to fit the priorities deemed most important.

Legendre Neural Network Method for Several Classes of Singularly Perturbed Differential Equations Based on Mapping and Piecewise Optimization Technology

In this paper, we develop a novel neural network model with mapping and piecewise optimization technology for several classes of the linear singularly perturbed initial value and boundary value differential equations with variable coefficients. First, the Legendre polynomials are selected as the activation function of the artificial neural network, the mapping technology is employed to transform the original uniform partition points and the piecewise optimization technology is used to improve the calculation accuracy. Then, the solution of the linear singularly perturbed differential equations is solved by using the extreme learning machine optimization algorithm. Finally, the numerical experiments show that the developed method can effectively improve the accuracy of the calculation.

Time-variant reliability analysis for industrial robot RV reducer under multiple failure modes using Kriging model

This paper proposes a time-variant reliability method for an industrial robot rotate vector (RV) reducer with multiple failure modes using a Kriging model. Firstly, the limit state functions of the industrial robot RV reducer are built by considering time-variant load and material degradation based on the failure physic method. Secondly, a time-variant reliability analysis method for multiple failure modes is proposed based on a double-loop Kriging model. The inner loop is the extremal optimization for each limit state function based on the efficient global optimization (EGO). The outer loop is the active learning reliability analysis by combining multiple response Gaussian process model (MRGP) and the Monte Carlo simulation (MCS). Furthermore, three learning functions (U-function, EFF-function and H-function) are individually adopted to choose a new sample point until the convergence is satisfied. Case studies are finally provided to illustrate the effectiveness of the proposed method.

Resources Allocation at the Physical Layer for Network Function Virtualization Deployment

Resource Allocation (RA) is one of the important factors in network function virtualization (NFV) deployment. As physical (PHY) layer resources are limited, e.g., transmitted energy and channel uses, the RA problem at the PHY layer for NFV deployment has become a fast-growing problem, especially for supporting ultra-reliable and low-latency communications (URLLC). Moreover, different nodes in NFV have different requirements for end-to-end communication, e.g., a controller has more stringent reliability requirements than does a logical node. There is a need for efficient and robust RA algorithms at the PHY layer for NFV deployment. To illustrate these challenges, we consider an up-link (UL) transmission protocol for NFV deployment, in which wireless transmissions with short packets are considered, and both the packet length and the transmission power are adjustable. Then, for three NFV deployment scenarios, we formulate three RA problems as three optimization problems to obtain the optimal parameters. Since these optimization problems are highly non-convex and they include excessive constraint conditions, the global optimal solutions are hard to obtain and are even infeasible for the conventional heuristic algorithms due to their low convergence efficiency. To address these problems, in this paper, the intelligent scheme of the modified shuffled frog-leaping algorithm (MSFLA) based on improved extremal optimization (EO) is applied to design RA algorithms. Three RA algorithms are designed for three NFV deployment scenarios to evaluate the quality of the solutions produced by the MSFLA-EO scheme. We perform simulations of three proposed RA algorithms in terms of various performance parameters. The experimental results are encouraging and demonstrate the efficiency of the proposed RA algorithms.