In this paper we have discussed the improvedK-Means Clustering with ParticleSwarmOptimization. One of the major drawback of K-Means Clustering is the random selection of seed, the random selection of initial seed results in different cluster which are not good in quality. The K-Means Clustering algorithm needs the steps (1) declaration of k clusters (2) initial seed selection (3) similarity matrix (4) cluster generation. The PSO algorithm is applied in step (2) the PSO gives the optimal solution for seed selection. In this paper, the standard K-Means is applied with different PSO to produce results which are more accurate and efficient than K-Means algorithm. In this paper there is no need to given k number of clusters in advance only a threshold value is required.
optimization, nonlinear programming, and modern heuristic algorithms. Along with these methods, a large variety of optimization models, such as dynamic programming (DP) (Archibald et al. 1997; Faber and Stedinger 2001; Kumar and Baliarsingh 2003; Tejadaguibert et al. 1995), artificial neural networks (ANN) (Chaves and Chang 2008; Chaves and Kojiri 2007; Neelakantan and Pundarikanthan 2000), the genetic algorithm (GA) (Chen 2003; Jothiprakash and Shanthi 2006; Kim et al. 2006; Li and Wei 2008; Sharif and Wardlaw 2000), the ant colony optimization (ACO) algorithm (Jalali et al. 2006, 2007; Kumar and Reddy 2006), fuzzy programming (Russell and Campbell 1996), and the shuffled complex evolution (SCE) algorithm (Ngo et al. 2007), as well as hybrid models (Cai et al. 2001, 2004; Chandramouli and Raman 2001; Cheng et al. 2008; Chiu et al. 2007; Mousavi et al. 2004a, 2004b; Ponnambalam et al. 2003; Reis et al. 2005; Yuan et al. 2008), have been widely used. More extensive and detailed understanding of these applications can be found in Labadie (2004) and Yeh (1985). Although so many models have been applied to the optimization of multi-reservoir system operation, the gap has been widening between theoretical research and real-world implementation, as pointed out by researchers such as Labadie (2004). Because of the complexity of multi-reservoir systems, featuring stochastic, dynamic, multi-dimensional, and nonlinear characteristics, many handicaps are not avoidable in the practical implementation of multi-reservoir system operation, including the so-called curse of dimensionality, which results in failure of many optimization models such as fuzzy programming (Russell and Campbell 1996).
ParticleSwarmOptimization (PSO) incorporates swarming behaviours observed in flocks of birds, schools of fish, or swarms of bees, and even human social behaviour, from which the idea is emerged. PSO is a population-based optimization tool, which could be implemented and applied easily to solve various function optimization problems, or the problems that can be transformed to functional optimization problems. As an algorithm, the main strength of PSO is its fast convergence that compares favourably with many global optimization algorithms like Genetic Algorithms (GA), Simulated Annealing (SA) and other global optimization algorithms. Population-based heuristics are more costly because of their dependency directly upon function values rather than derivative information. They are however susceptible to premature convergence, which is especially the case when there are many decision variables or dimensions to be optimized.
After extracting the features, the image is further segmented by using the combination of two techniques such as, Threshold based ParticleSwarmOptimization (T-PSO), and Fuzzy K-Means (FKM). The main reason for using these techniques are, the common pixels that are segmented by the two techniques are considered for an efficient segmentation. Typically, image segmentation is defined as the process of splitting an image into sub-parts for further processing. The accuracy of image processing system is fully depend on the process of segmentation. Thresholding is one of the most widely used segmentation technique that separates the pixels of the image in varying classes based on the level of intensity. But, it cannot be applicable for processing the multichannel image.
The detailed flow chart helps to better understand the PSO algorithm. The study aims to improve the algorithm in the purpose of applying it to solve the ED of power system. With the always increasing need of power, the competitive market and the growing of the commodity material, added to the fact that power cannot be stored at a high level, it is essential that the power produced must be used in its optimal capability: thus the importance of economic dispatch in power system. The PSO code modified and adapted to address this issue is proposed in . In thermal power generating firm which is the case study, ED means to minimize the fuel cost for power generation through proper dispatch schedule. That is the firm need to reduce fuel cost, for better profit at the same time it should satisfy system load demand, reactive power limit, voltage limit, power transmission limit and other limitations . To achieve that, the firm has to spend money which affects the power generating profit [28, 29]. Therefore the two opposite objectives should be compromised to find the firm’s optimal profit [30, 31]. So the objective of an economic load dispatch here is to find the optimal combination of power generation that minimizes the total generation cost while satisfying quality and inequality constraints [29, 30]. Furthermore, DED involves more parameters than the static one. This is a complex optimization situation. Its importance may increase with the growth in power generation competition [32- 35].
Abstract — Single Nucleotide Polymorphisms (SNPs) are confirmed as a major factor in human genome polymorphisms, and are found to be suitable as a genetic marker for disease characteristics. Determining the relationship between disease complexity and SNPs requires complex genotyping for large SNP data sets, and is thus very expensive and labor-intensive. Tag-SNP selection is a useful technique for selecting the SNP subset from the original dataset with minimal errors, thus reducing the cost of genotyping. We propose a new method that combining strategy particleswarmoptimization (SPSO) and Linkage Disequilibrium (LD) to select highly correlated SNPs. We use this strategy in PSO to select more optimal solutions and replace less accurate particles. The method is demonstrated using the HapMap dataset through evaluating the leave-one-out-cross-validation (LOOCV) and K-nearest neighbor (KNN) method. Experimental results show that our proposed method results in a smaller subset of tag SNPs and provides improved accuracy over PSO and other tag SNP selection methods.
There are many researches based on using PSO as an optimization step to improve clustering algorithms like K-Means and Fuzzy C-Means. The clustering algorithms like FCM are very sensitive to initial parameters. The algorithm may lead to fall into the local optima, if the initial values are not selected properly. To overcome these kind of weaknesses, which results that the FCM algorithm cannot reach the global optimum solution, the using of ParticleSwarmOptimization (PSO) as an optimization method has been introduced.
Investigasi kecelakaan penerbangan di Indonesia pada tahun 2010 sampai 2016 sebesar 212 investigasi. Hal tersebut dapat dihindari apabila ada sistem penerbangan yang dapat memastikan penerbangan berjalan aman, seperti sistem lalu lintas udara yang dapat mendeteksi apabila pesawat bergerak menuju ke arah yang salah. Penelitian ini bertujuan untuk mengelompokkan rute penerbangan pada data Automatic Dependent Surveillance-Broadcast menggunakan metode clustering untuk mendapatkan similaritas rute penerbangan. Penulis mengusulkan metode particleswarmoptimization untuk mengoptimalkan metode k-means, yang berguna untuk menentukan titik centroid awal dengan silhouette coefficient sebagai fitness function. Hasil dari penelitian ini menghasilkan zona terbang berdasarkan kebiasaan sehingga dapat digunakan sebagai panduan penerbangan. Pengujian dilakukan dengan membandingkan nilai Davies-Bouldin index dengan metode k-means, k-medoids dan fuzzy c-means. Pada uji coba yang dilakukan, metode yang diusulkan menjadi kelompok metode terbaik pada lima dari enam segmen yang ada serta menghasilkan nilai Davies-Bouldin index lebih baik pada satu segmen sebesar 0,779.
The global best position of Se-PSO showed a very improved result in short time when compared to PSO in Griewank function depicted in Table XIII. Moreover, as shown in Tables XI and XII the convergence speed of the Se-PSO towards the optimum values was faster when compared to PSO. It’s however, noticeable that the convergence of Se-PSO was quick in the all function but slowed down searching large space for the global best position before to be decided by Se- PSO algorithm as it approaches the optimal. The Se-PSO it took 0.037s to reach the best global position but the PSO reached the global best position in 0.014s. Where, Se-PSO takes 0.002s to decide the global best position while PSO takes 0.013s are depicted in Tables XI and XII in the self –time column. Moreover, Se-PSO searching large space almost 7 times of PSO as it described in the Calls column (4920 and 650), respectively.
energy has been used in transmitting and receiving data. Besides that, in order to conserve energy, the ways on placing the sensor is crucial. The issue on coverage area is also important and it is closely related to the energy efficiency in wireless sensor network. Therefore, researcher has researched many ways to work out the problems in wireless sensor network, for example using Genetic Algorithm, ParticleSwarmOptimization and many more other algorithm. In this project, ParticleSwarmOptimization algorithm is used. Besides that, clustering is also important in conserve energy in wireless sensor network. Clustering is the process of grouping data together based on the relationship among them. In this project, K-means clustering protocol is used. Further explanation on ParticleSwarmOptimization algorithm and K-means clustering protocol will be explained in Chapter 3.
This paper presents an improved extended Markowitz mean-variance portfolio model with the introduction of new constraint known as expert opinion. This work would the first in literature pioneering this new innovation to portfolio selection model since advent of portfolio selection problem. This new extended model consists of four constraints namely: bounds on holdings, cardinality, minimum transaction lots, and expert opinion. An efficient heuristic method of particleswarmoptimization algorithm was engaged and genetic algorithm to make a comparison of the results obtained. Five performance evaluation criteria already used in similar works by other researchers in literature was used tocompare performance between particleswarmoptimization algorithm and genetic algorithmon both small and large data set for the improved extended portfolio model developed. In all the test cases PSO achieve better solution than GA, however with higher computational mean execution time. Further studies are to engage comparative study of other swarm intelligence techniques to the new extended portfolio model developed in this paper.
Particleswarm algorithms have been shown to success- fully optimize a wide range of problems [1, 2, 3, 4, 5]. The development of the PSO and its applications was reviewed in Ref. . Ref.  proposed a PSO approach to identify the autoregressive moving average with an exogenous vari- able model for short-term power load forecasting. Ref.  presented a modified PSO algorithm for engineering opti- mization with constraints that algorithm was tested on sev- eral engineering design optimization problems. Ref.[4, 5] are two recent books which reviewed particleswarm opti- mization algorithms and their engineering applications. The PSO algorithm is based on a metaphor of social interaction to search a space by adjusting the trajectories of individual vectors, called “particles” conceptualized as moving points in a multidimensional space. The random weight of the control parameters is used to give a kind of explosion as particle velocities and positional coordinates careen toward infinity . Most algorithm versions have some undesirable dynamical properties [7, 8, 9]. Notably, the particle veloc- ities need to be limited in order to control their trajecto- ries. Clerc and Kennedy  have significantly improved the convergence tendencies of particleswarm systems by intro- ducing a constriction coefficient and by resorting to random weights to control the search space of the particle trajec- tories. PSO was also combined with other algorithms to extend the search space. Typically, Ref.  used chaos and PSO in an alternative fashion to avoid getting trapped in local minima.
It can be observed from Fig. 1 that while subsequently modifying the inertia weight, cognitive and the social component in the control equation of the conventional PSO, the convergence characteristics are progressively improved by avoiding more and more local trappings. It can also be observed from the figure that except IPSO, the initial shape of convergence is more or less same. Therefore, the introduction of the constriction factor in the social component of particle’s velocity is found to be one of the key factors that is responsible for better convergence in IPSO. A similar conclusion can be drawn from Fig. 2. It can be observed from the figure that in IPSO alone, the particles are not identifying the promising region during early iterations. Therefore, it offers advantage to exploit this region meticulously and thus produces better quality solutions.
Feature selection is the process of choosing a subset of features from the original feature set and thus can be viewed as a principal pre-processing tool prior to solving classification problems . Feature selection is a NP-hard problem. The goal is to select a subset of d features from a set of D features (d < D) in a given data set . D is comprised of all features in a given data set; it may include noisy, redundant, and misleading features. Therefore, an exhaustive search performed in the solution space, which usually takes a long time, often does not work in practice . To resolve these feature selection problems, we aimed at retaining only d relevant features. Irrelevant features are not only useless for classification, but could also potentially reduce the classification performance. By deleting irrelevant features computational efficiency can be improved and classification accuracy increased.
Extending from the existing parameter setting techniques on inertia weight – and acceleration coefficients –, this paper develops a systematic adaptation scheme. The PSO parameters are not only controlled by ESE but also taking the different effects of these parameters in different states into account. In addition, departing from mutation , reset , , or reinitialization ,  operations, the ELS is pro- posed in this paper to perform only on the globally best particle and only in a convergence state. This is not only because the convergence state needs the ELS most but also because of a very low computational overhead. Further, the adaptive ELS will maintain population diversity for jumping out of the potential local optima. Moreover, tests are to be carried out on various topological structures in the PSO paradigm to verify the effectiveness of the APSO and to more comprehensively compare with other improved PSO algorithms.
Druhým používaným sousedstvím a pro sociální sousedství relevantnější je lokální sousedství (Local Best topologie). V tomto sousedství se většinou používá jako proměnná pro nejlepší nalezené řešení v sousedství v rovnici (3.2) lBest místo gBest. Nejčastější topologie je kruhová mřížka, kde každý jedinec je připojen k oběma sousedním členům populační řady (kruhu, K = 2). Obecně K > 2, kde K je počet sousedů. Výměna informace uvnitř sousedství jedinců odráží lokální znalost prostředí. Jednotlivá sousedství nejsou oddělenými jednotkami, ale jsou navzájem propojená a mohou se navzájem ovlivňovat. Výhoda této topologie je v možnosti paralelního prohledávání, protože jednotlivé populace mohou konvergovat v odlišných oblastech prohledávacího prostoru. Je tedy stabilní pro více stejně dobrých optim. Ačkoli toto sousedství konverguje pomaleji než globální sousedství, je méně zranitelné na uváznutí v lokálním optimu .
The original research by Kennedy and Eberhart started out to acquire local agent rules for simulating the social be- havior of a flock of birds or a school of fish. Inspired by the works such as  by Reynolds, who programmed a flock- simulating group of agents (boids) by sole description of agent interaction rules, they started to try various interaction rules among the agents. Soon, they found that some type of interaction in the flock can benefit the e ﬃ ciency of finding an important location, such as food in the flock’s domain of activity. They saw this nature e ﬀ ective as a novel algorithm for general nonlinear programming, so they improved and simplified the rule for updating the agent’s position, making use of some local memory and information sharing. Thus the ParticleSwarmOptimization was born.
Vehicle Routing Problem (VRP) is addressed to a class of problems for determining a set of vehicle routes, in which each vehicle departs from a given depot, serves a given set of customers, and returns back to the same depot. On the other hand, simultaneous delivery and pickup problems have drawn much attention in the past few years due to its high usage in real world cases. This study, therefore, considered a Vehicle Routing Problem with Time Windows and Simultaneous Delivery and Pickup (VRPTWSDP) and formulated it into a mixed binary integer programming. Due to the NP-hard nature of this problem, we proposed a variant of ParticleSwarmOptimization (PSO) to solve VRPTWSDP. Moreover, in this paper we improve the basic PSO approach to solve the several variants of VRP including Vehicle Routing Problem with Time Windows and Simultaneous Delivery and Pickup (VRPTWSDP), Vehicle Routing Problem with Time Windows (VRPTW), Capacitated Vehicle Routing Problem (CVRP) as well as Open Vehicle Routing Problem (OVRP). In proposed algorithm, called ImprovedParticleSwarmOptimization (IPSO), we use some removal and insertion techniques and also combine PSO with Simulated Annealing (SA) to improve the searching ability of PSO and maintain the diversity of solutions. It is worth mentioning that these algorithms help to achieve a trade- off between exploration and exploitation abilities and converge to the global solution. Finally, for evaluating and analyzing the proposed solution algorithm, extensive computational tests on a class of popular benchmark instances, clearly show the high effectiveness of the proposed solution algorithm.
In this paper an ImprovedParticleSwarmOptimization (IPSO) algorithm is proposed to solve the optimal reactive power Problem. In order to overcome the drawbacks of standard genetic algorithm (GA) and particleswarmoptimization (PSO) , some improved mechanisms based on non-linear ranking selection, competition and selection among several crossover offspring and adaptive change of mutation scaling are adopted in the genetic algorithm, & dynamical parameters are adopted in PSO. The new population is produced through three approaches to improve the global optimization performance, which are elitist strategy, PSO strategy and enhanced genetic algorithm strategy. The effectiveness of the proposed algorithm has been compared with Gas and PSO, synthesizing a circular array, a linear array and a base station array. In order to evaluate the efficiency of the proposed algorithm, it has been tested in standard IEEE 118 & practical 191 bus test systems and compared other algorithms. Simulation results show that real power loss considerably reduced and control variables are within the limits.
In this paper, eight dierent improvements in the PSO method are proposed to solve the CHPED problem with valve point loading eects on pure power units and power losses in transmission lines. The proposed improved PSO (IPSO) methods are tested on six systems and the obtained results are compared with those from other methods such as LR , GA , IACSA , EP , IGA-MU , LR-SQP , SARGA , BCO , HSA [1,12-13], MADS-LHS , MADS-PSO , MADS-DACE , NDS , AIS , LR-SSMU-CSS , LR-SSMU-SSBS , TVAC-PSO , CGSO , and IGSO  in terms of cost and execution time.