Estimation Of Distribution Algorithm Research Articles

Modelling and solving a cost-oriented resource-constrained multi-model assembly line balancing problem

A line balancing problem considers the assignment of operations to workstations in an assembly line. While assembly lines are usually associated to mass production of standardised goods, their advantages have led to their widespread use whenever a product-oriented production system is applicable and the benefits of the labour division and specialisation are significant, even when some of its characteristics may deviate from classical assembly lines. In this work, we study a line balancing problem found in the textile industry in which the line must be balanced for multiple types of goods taking into account resource requirements. In order to solve the problem, a hybrid method that combines classical methods for line balancing with an Estimation of Distribution Algorithm is proposed. Computational experiments show that the new procedure improves upon the state of the art when compared using a benchmark set derived from the literature, as well as when compared using data from the manufacturer that originated this research work.

Anisotropic adaptive variance scaling for Gaussian estimation of distribution algorithm

Traditional Gaussian estimation of distribution algorithms (EDAs) are confronted with issues that the variable variances decrease fast and the main search direction tends to become perpendicular to the improvement direction of the fitness function, which reduces the search efficiency of Gaussian EDAs (GEDAs) and makes them subject to premature convergence. In this paper, a novel anisotropic adaptive variance scaling (AAVS) technique is proposed to improve the performance of traditional GEDAs and a new GEDA variant named AAVS-EDA is developed. The advantages of AAVS over the existing variance scaling strategies lie in its ability for tuning the variances and main search direction of GEDA simultaneously, which are achieved by anisotropically scaling the variances along different eigendirections based on corresponding landscape characteristics captured by a simple topology-based detection method. Besides, AAVS-EDA also adopts an auxiliary global monitor to ensure its convergence by shrinking all the variances if no improvement is achieved in a generation. The evaluation results on 30 benchmark functions of CEC2014 test suite demonstrate that AAVS-EDA possesses stronger global optimization efficiency than traditional GEDAs. The comparison with other state-of-the-art evolutionary algorithms also shows that AAVS-EDA is efficient and competitive.

Multiagent Rendezvous With Shortest Distance to Convex Regions With Empty Intersection: Algorithms and Experiments.

This paper presents both algorithms and experimental results to solve a distributed rendezvous problem with shortest distance to convex regions. In a multiagent network, each agent is assigned to a certain convex region and has information about only its own region. All these regions might not have an intersection. Through local interaction with their neighbors, multiple agents collectively rendezvous at an optimal location that is a priori unknown to each agent and has the shortest total squared distance to these regions. First, a distributed time-varying algorithm is introduced, where a corresponding condition is given to guarantee that all agents rendezvous at the optimal location asymptotically for bounded convex regions. Then a distributed tracking algorithm combined with a distributed estimation algorithm is proposed. It is first shown that for general possibly unbounded convex regions, all agents rendezvous in finite time and then collectively slide to the optimal location asymptotically. Then it is shown that for convex regions with certain projection compressibility, all agents collectively rendezvous at the optimal location in finite time, even when the regions are time varying. The algorithms are experimentally implemented on multiple ground robots to illustrate the obtained theoretical results.

Unreliable communication in high-performance distributed multi-agent systems: A ingenious scheme in high computing

Designing of distributed consensus algorithms featuring accuracy, robustness, reliability, and speed of convergence is in high demand for various multi-agent applications. In this research, it has been investigated to device a novel design of distributed estimation algorithm which can tackle the problem of unreliable communication among multi-agents to achieve consensus on the average value of their initial values and must be capable of computing the total number of agents in the system under dynamically changing interaction topologies. A dynamically changing network topology is considered in this research with unreliable communication links, and four different scenarios are established to be analyzed for the proposed consensus-based distributed estimation algorithm. This study established a consensus for a dynamically changing interaction topology among agents, for addition of agents in the network with dynamically switching topology at any instant in communication, for removal of agents from the network with dynamically switching topology at any instant in communication, and for a fixed topology with link failure and a reconnection with the same agent after each iteration. The proposed algorithm paces up the rate of convergence by reducing the number of iteration, along with sure convergence of the designed algorithm using the concepts of stochastic differential equation theory, control system theory, algebraic graph theory, and algebraic matrix theory. Finally, in the end, simulation results are provided which are clear evidence to validate the effectiveness of theoretical results of the proposed algorithm in comparison to previously known consensus algorithms in terms of different performance parameters.

Multi‐objective hybrid estimation of distribution algorithm‐interior point method‐based meter placement for active distribution state estimation

This study proposes a new multi-objective hybrid estimation of distribution algorithm (EDA)-interior point method (IPM) algorithm to obtain the optimal location of measuring devices for state estimation (SE) in active distribution networks. The objective functions to be minimised are, the total network configuration cost, the average relative percentage error of bus voltage magnitude and angle estimates. As the objectives are conflicting in nature, a multi-objective Pareto-based non-dominated sorting EDA has been proposed in this study. Moreover, due to poor exploitation capability of the EDA, it is hybridised with IPM to improve its local searching ability in the search space. The hybridisation of EDA and IPM brings a higher degree of balance between the exploration and exploitation capability of the algorithm during the search process. Furthermore, the loads and generators are treated as stochastic variable and the impact of different type of distributed generations on SE performance has also been investigated. The efficiency of the proposed algorithm is tested on PG&E 69-bus system and Indian 85-bus radial distribution network. The obtained results are compared with conventional EDA, particle swarm optimisation, non-dominated sorting genetic algorithm and also with existing techniques in the literature such as dynamic programming and ordinal optimisation algorithm.

Bayesian hypothesis testing detector for one bit diffusion LMS with blind missing samples

This paper proposes a sparse distributed estimation algorithm when missing data occurs in the measurements over adaptive networks. Two classes of measurement models are considered. First, the traditional linear regression model is investigated and second the sign of the linear regression model is studied. The latter is referred to as one-bit model. We utilize the diffusion LMS strategy, in the proposed methods, where a set of nodes cooperates with each other to estimate a vector model parameter. In both models, it is shown that replacing the missing sample with a simple estimate is equivalent to removing the missing sample from the distributed diffusion algorithm. We consider two cases, where in the first case the positions of missing samples are known (non-blind) and in the second case the positions of missing samples are unknown (blind). In the linear regression model scenario, a Bayesian hypothesis testing (BHT) is used for detection of the missing samples. In the one-bit model, in addition to BHT detector, a simple heuristic detector, based on mean square error (MSE), is also suggested. Simulation results show the effectiveness of the proposed detector-assisted distributed algorithms.

Optimal Hyper-Parameter Tuning of SVM Classifiers With Application to Medical Diagnosis

Proper tuning of hyper-parameters is essential to the successful application of SVM-classifiers. Several methods have been used for this problem: grid search, random search, estimation of distribution Algorithms (EDAs), bio-inspired metaheuristics, among others. The objective of this paper is to determine the optimal method among those that recently reported good results: Bat algorithm, Firefly algorithm, Fruit-fly optimization algorithm, particle Swarm optimization, Univariate Marginal Distribution Algorithm (UMDA), and Boltzmann-UMDA. The criteria for optimality include measures of effectiveness, generalization, efficiency, and complexity. Experimental results on 15 medical diagnosis problems reveal that EDAs are the optimal strategy under such criteria. Finally, a novel performance index to guide the optimization process, that improves the generalization of the solutions while maintaining their effectiveness, is presented.

Estimation of Distribution Algorithm for Initial Orbit Determination of Too-short-arc Based on Kernel Density Estimation

A new approach of the initial orbit determination for too-short-arc with angular measurements is implemented by building the probabilistic model in the solution space with the estimation of distribution algorithm. Without any assumption about distribution, the non-parametric kernel density estimation is employed in the model building. The method, unlike other evolutionary algorithms, such as the genetic algorithm and particle swarm optimization, considers the fitness as well as the characteristics of the solution space. Numerical experiments with real observations indicate that without any constraints, the proposed technique has a good performance for the observations of usual accuracy.

Resilient Estimation and Control on k-Nearest Neighbor Platoons: A Network-Theoretic Approach

Abstract This paper is concerned with the network-theoretic properties of so-called k-nearest neighbor intelligent vehicular platoons, where each vehicle communicates with k vehicles, both in front and behind. The network-theoretic properties analyzed in this paper play major roles in quantifying the resilience and robustness of three generic distributed estimation and control algorithms against communication failures and disturbances, namely resilient distributed estimation, resilient distributed consensus, and robust network formation. Based on the results for the connectivity measures of the k-nearest neighbor platoon, we show that extending the traditional platooning topologies (which were only based on interacting with nearest neighbors) to k-nearest neighbor platoons increases the resilience of distributed estimation and control algorithms to both communication failures and disturbances.

Multi-mode resource leveling in projects with mode-dependent generalized precedence relations

In real-life project management, resource leveling is an important technique to ensure the effective use of resources, in which activities (a) can often be executed in alternative modes and (b) are constrained by precedence relations with minimum and maximum time lags that can be modeled using generalized precedence relations (GPRs). In addition, the values of the time lags tend to depend on activity modes. The resource leveling problem with multiple modes and mode-dependent GPRs (MRLP-GPR) is a generalization of the classic NP-hard resource leveling problem. To our knowledge, no literature exists regarding the MRLP-GPR.We propose several heuristics for the MRLP-GPR built upon two solution approaches: (a) a steepest descent algorithm and a fast descent algorithm that are based on a decomposition approach and (b) a hybrid estimation of distribution algorithm (EDA), which is based on an integration approach. Extensive computational experiments on a large number of benchmark instances are conducted to evaluate the proposed heuristics. A comparison of the results shows that our EDA outperforms or is competitive with three baseline heuristics (a random search algorithm and two variants of a genetic algorithm that is the best-performing metaheuristic for the single-mode resource leveling problem with GPRs). Our results can serve as a benchmark for future research. Our model and solution algorithms provide an automatic tool for the project manager's multi-mode resource leveling decision-making.

An estimation of distribution algorithm for scheduling problem of flexible manufacturing systems using Petri nets

• This is the first report on applying estimation of distribution algorithm (EDA) to the studied problem. • A kind of PN-based deadlock controllers for FMSs is imbedded to exclude infeasible individuals. • An effective voting procedure is adopted to construct the probabilistic model of EDA. • The longest common subsequence is also embedded in the model for mining excellent genes. • A new modified variable neighborhood search is developed as an efficiency enhancement of EDA. Based on the place-timed Petri net models of flexible manufacturing systems (FMSs), this paper proposes a novel effective estimation of distribution algorithm (EDA) for solving the scheduling problem of FMSs. A candidate solution is represented as an individual with two sections: the first contains the route information while the second is a permutation with repetition for parts. The feasibility of individuals is checked and guaranteed by a highly permissiveness deadlock controller. A feasible individual is interpreted into a deadlock-free schedule while the infeasible ones are amended. The probabilistic model in EDA is constructed via a voting procedure. An offspring individual is then produced based on the model from a seed individual, and the set of seed individuals is extracted by a roulette method from the current population. The longest common subsequence is also embedded into the probabilistic model for mining good genes. A modified variable neighborhood search is applied on offspring individuals to obtain better solutions in their neighbors and hence to improve EDA’s performance. Computational results show that our proposed algorithm outperforms all the existing ones on benchmark examples for the studied problem. It is of important practice significance for the manufacturing of time-critical and multi-type products.

A distributed estimation algorithm for collective behaviors in multiagent systems with applications to unicycle agents

In this paper, we propose a distributed estimation algorithm for the detection of collective behaviors in multiagent systems. Through local information exchange, the proposed algorithm can estimate the average of the integrals of local time-varying behavior signals of individual agents, such as network moments, which could serve as a feature indicator for group collective behaviors. Under the assumption that the communication network among agents is connected and bidirectional, the algorithm convergence is rigorously analyzed with an explicitly defined bounds of estimation errors. The event-triggering mechanism for information transmission is further employed for reducing the communication load. As a case study, the proposed distributed estimation algorithm is applied to detect the collective behaviors of unicycle agents. Simulation results on the estimation of network centroids are provided to illustrate the effectiveness of the method.

Multi-target localisation and circumnavigation by a multi-agent system with bearing measurements in 2D space

ABSTRACTThis paper investigates the problem of multi-target localisation and circumnavigation by a multi-agent system in two-dimensional space, where each target's bearing measurement can be obtained by at least one agent. Both stationary and slowly-moving targets are considered. An estimator is designed to estimate the target position; then a distributed estimation algorithm is designed for each agent to cooperatively estimate the centroid of multi-target; and finally, based on the first two steps, a circumnavigation controller is devised to guarantee that each agent circumnavigates all targets around the multi-target centroid with a desired radius. The convergence of the estimation algorithm and the stability of the circumnavigation controller are proved. A numerical simulation is provided to verify the correctness of the conclusion and the effectiveness of the proposed algorithm.

Distributed sparse diffusion estimation based on set membership and affine projection algorithm

The wireless sensor networks (WSNs) is the result of evolution of wireless and networking technologies, micro-electromechanical systems, and micro-services. One important task which can be performed by nodes in WSNs is to find a common solution to a problem by using distributed processing. In this paper, we study the problem of distributed estimation, where a group of nodes are required to collectively estimate a sparse parameter vector of interest, and we solve it by an estimation algorithm based on the set membership (SM) and affine projection (AP) methods. At each iteration of the algorithm, in addition to the current data, the proposed algorithm uses data obtained from previous measurements to attain faster convergence rate. A method is also proposed to select the constraint bound for set membership such that the computational load is uniformly distributed among the nodes of network. Then the distributed estimation algorithm is analyzed and a closed form expression for the steady state mean square deviation (MSD) is developed. The performance of proposed method is assessed via computer simulations. The simulation results show that the proposed algorithm provides faster convergence rate and smaller steady state MSD value when compared to conventional methods such as diffusion least mean squares (LMS), distributed set membership normalized LMS (DSM-NLMS), and distributed APA. Moreover, it achieves lower computational load compared to the AP method. These advantages make the proposed method useful in sparse parameter vector estimation whenever the nodes have sufficient memory size.

Improved Alopex-based evolutionary algorithm by Gaussian copula estimation of distribution algorithm and its application to the Butterworth filter design

ABSTRACTThe application of evolutionary algorithms (EAs) is becoming widespread in engineering optimisation problems because of their simplicity and effectiveness. The Alopex-based evolutionary algorithm (AEA) possesses the basic characteristics of heuristic search algorithms but is lacking in adequate information about the fitness landscape of the input domain, reducing the convergence speed. To improve the performance of AEA, the Gaussian copula estimation of distribution algorithm (EDA) is embedded into the original AEA in this paper. With the help of Gaussian copula EDA, precise probability models are built utilising the best solutions, which can increase the convergence speed, and at the same time, keep the population diversity as much as possible. The simulation results on the benchmark functions and the application to the Butterworth filter design demonstrate the efficiency and effectiveness of the proposed algorithm, compared with several other EAs.

An investigation of the selection strategies impact on MOEDAs: CMA-ES and UMDA

Researchers have acknowledged the importance of the selection, replacement, and archiving strategies in the behavior of evolutionary algorithms (EAs). However, these important relationships have not been deeply investigated in terms of Estimation of Distribution algorithms (EDAs). In a preliminary research, we focused on UMDA, a simple EDA that uses univariate Gaussian factorizations. Experimental results confirm that the choice of the selection method can provoke probabilistic modeling to be more effective for some classes of functions. Motivated by these results, this study is extended by evaluating several variants of selection strategies and probabilistic modeling approaches. The aim is to detect possible interactions between these two important components of evolutionary algorithms. Specifically, we use the selection strategies as defined for NSGA2, SPEA2, and IBEA algorithms, and the probabilistic models implemented as part of UMDA and CMA-ES, and a simple crossover operator (SBX). The recently introduced COCO framework comprising 55 bi-objective functions is used as the benchmark for the analysis. The results show that probabilistic modeling has an advantage over the classical genetic operator regardless of the selection method applied. Nevertheless, the results also show that some selection methods have a better performance when applied together with EDAs.

A cloud-based smart metering infrastructure for distribution grid services and automation

Abstract The evolution of the power systems towards the smart grid paradigm is strictly dependent on the modernization of distribution grids. To achieve this target, new infrastructures, technologies and applications are increasingly required. This paper presents a smart metering infrastructure that unlocks a large set of possible services aimed at the automation and management of distribution grids. The proposed architecture is based on a cloud solution, which allows the communication with the smart meters from one side and provides the needed interfaces to the distribution grid services on the other one. While a large number of applications can be designed on top of the cloud, in this paper the focus will be on a real-time distributed state estimation algorithm that enables the automatic reconfiguration of the grid. The paper will present the key role of the cloud solution for obtaining scalability, interoperability and flexibility, and for enabling the integration of different services for the automation of the distribution system. The distributed state estimation algorithm and the automatic network reconfiguration will be presented as an example of coordinated operation of different distribution grid services through the cloud.

Estimation of Travel Time Distributions in Urban Road Networks Using Low-Frequency Floating Car Data

Travel times in urban road networks are highly stochastic. However, most existing travel time estimation methods only estimate the mean travel times, while ignoring travel time variances. To this end, this paper proposes a robust travel time distribution estimation method to estimate both the mean and variance of travel times by using emerging low-frequency floating car data. Different from the existing studies, the path travel time distribution in this study is formulated as the sum of the deterministic link travel times and stochastic turning delays at intersections. Using this formulation, distinct travel time delays for different turning movements at the same intersection can be well captured. In this study, a speed estimation algorithm is developed to estimate the deterministic link travel times, and a distribution estimation algorithm is proposed to estimate the stochastic turning delays. Considering the low sampling rate of the floating car data, a weighted moving average algorithm is further developed for a robust estimation of the path travel time distribution. A real-world case study in Wuhan, China is carried out to validate the applicability of the proposed method. The results of the case study show that the proposed method can obtain a reliable and accurate estimation of path travel time distribution in congested urban road networks.

Fast reliability evaluation method for composite power system based on the improved EDA and double cross linked list

To improve the computational efficiency of Monte Carlo simulation in composite power systems reliability evaluation, this study presents a method based on the improved estimation of distribution algorithm (EDA) and double cross linked list. Compared to traditional techniques, this method is comprehensively improved in the stage of both sampling and state evaluation. In the sampling stage, the population-based incremental learning algorithm is presented, where the probability vector is updated based on the distribution characteristics of excellent samples in population of previous generations. Meanwhile, setting a limit to the probabilities of elements in normal state and mutation strategy are introduced, which improves the excellent characteristics of the population. In the state evaluation stage, the state search and match process is speed up by utilising the intelligent storage technology based on the double across linked list. It avoids calling the optimal power flow for the same state repeatedly. Finally, the proposed method is tested in IEEE RTS 79. As the result shows, compared with other methods ever used in reliability evaluation, this method is not only more efficient in computation but also more accurate. Thus, the proposed method is proved to be reliable and effective.

Determining the inspection intervals for one-shot systems with support equipment

This paper considers systems that comprise one-shot devices and support equipment. One-shot devices are stored for long periods of time, and failures are detected only upon inspection. The support equipment needed to operate one-shot devices is maintained immediately upon failure. This paper addresses the inspection schedule problem for such systems with limited maintenance resources. The interval availability and life cycle cost are used as optimization criteria. The aim is to determine near-optimal inspection intervals for one-shot systems to minimize the expected life cycle cost and satisfy the target interval availability between inspection periods. An estimation of distribution algorithm (EDA) and a heuristic method are proposed to find the near-optimal solutions, and numerical examples are given to demonstrate the effects of the various model parameters to the near-optimal inspection intervals.