International Journal of Swarm Intelligence (13 papers in press)
Concurrent parametric optimization of single pass end milling through GRA coupled with PSO for calmax-635 die steel
by Bikash Bepari, Ankit
Abstract: The present experimental investigation is to obtain optimized parametric combination for enhanced surface quality during finishing of Calmax-635 die steel through Grey Relational Analysis (GRA) coupled with Particle Swarm Optimization (PSO). GRA converts multiple objectives into a single objective by defining the Grey Relational Grade (GRG). However, it yields discrete parametric combination within the problem space and fetches quasi-optimal solution. Whereas, PSO being a swarm intelligent technique can obtain optimal solution if the fitness function is available. Due to absence of the fitness function for Calmax-635 die steel in the literature, a full factorial Design of Experiment (DOE) was conducted for milling parameters like, spindle speed, feed rate and depth of cut all at three levels. With the help of regression analysis and ANOVA, a relationship between the process parameters and GRG was established which acted as a fitness function for PSO within the problem space. Thus, when PSO was coupled with GRA, the optimized process parameters became 5660.6 rpm, 579.4 mm/min, 0.105 mm respectively and the roughness values obtained were 0.862 m, 6.591 m and 4.638 m for Ra, Rmax and Rz respectively which showed significant improvement pertinent to surface quality. Therefore the proposed methodology reveals an alternative way for optimization in absentia of fitness function.
Keywords: Calmax-635; Grey relational analysis; Principal component analysis; Grey relation grade; Particle Swarm Optimization; Surface quality.
Improving The Cooperation of Fuzzy Simplified Memory A* Search and Particle Swarm Optimization for path planning
by Mehdi Neshat
Abstract: Problem solving is a very important subject in the world of AI. In fact, a problem can be considered one or more goals along with a set of available interactions for reaching those goals. One of the best ways of solving AI problems is to use search methods. The SMA* (Simplified Memory Bounded A*) is one of the best methods of informed search. In this research, a hybrid method was proposed to increase the performance of SMA* search. The combining fuzzy logic with this search method and improving it with PSO algorithm brought satisfactory results. The use of fuzzy logic leads to increase the search flexibility especially when a robot dealing with lots of barriers and path changes. Furthermore, combining PSO saves the search from being trapped into local optimums and provides for search some correct and accurate suggestions. In the proposed algorithm, the results indicate that the cost of search and branching factor are decreased in comparison with other methods.rnrn
Keywords: : Informed Search; Fuzzy Logic; Particle Swarm Optimization;Simplified Memory Bounded A*; Robot Navigation.
Two swarm intelligence based approaches for the p-center problem
by B. Jayalakshmi, Alok Singh
Abstract: The p-center problem is an important facility location problem. In this problem, the objective is to find a set Y of p vertices on an undirected weighted graph G = (V, E) in such a way that Y is a subset of V and the maximum distance over all the distances from vertices to their closest vertices in Y is minimized. The vertices in set Y are called centers. In this paper, we have proposed two swarm intelligence based approaches for the p-center problem. The first approach is based on artificial bee colony (ABC) algorithm, whereas the latter approach is based on invasive weed optimization (IWO) algorithm. The ABC algorithm and IWO algorithm are relatively new metaheuristic techniques inspired respectively from collective intelligent behavior shown by honey bees while foraging and the sturdy process of weed colonization & dispersion in an ecosystem. Computational results on the well-known benchmark instances of p-center problem show the effectiveness of our approaches in finding high quality solutions.
Keywords: artificial bee colony algorithm; facility location problem; invasive weed optimization algorithm; p-center problem; swarm intelligence.
A self-tuning Firefly Algorithm to tune the parameters of Ant Colony System (ACSFA)
by Anuradha Ariyaratne, T.G.I. Fernando, Sunethra Weerakoon
Abstract: Ant colony system (ACS) is a promising approach which has been widely used in problems such as Travelling Salesman Problems (TSP), Job shop scheduling problems (JSP) and Quadratic Assignment problems (QAP). In its original implementation, parameters of the algorithm were selected by trial and error approach. Over the last few years, novel approaches have been proposed on adapting the parameters of ACS in improving its performance. The aim of this paper is to use a framework introduced for self-tuning optimization algorithms combined with the firefly algorithm (FA) to tune the parameters of theACSsolving symmetric TSP problems. The FA optimizes the problem specific parameters of ACS while the parameters of the FA are tuned by the selected framework itself. With this approach, the user neither has to work with the parameters of ACS nor the parameters of FA. Using common symmetric TSP problems we demonstrate that the framework fits well for the ACS. A detailed statistical analysis further verifies the goodness of the new ACS over the existing ACS and also of the other techniques used to tune the parameters of ACS.
Keywords: Self tuning framework; Ant Colony System; Travelling Salesman problem; Firefly Algorithm.
Special Issue on: Computational Intelligence Theory, Methods and Applications
Performance Evaluation of Pitch Values for Finding Emotions in Tamil Speech
Abstract: All living creatures in this world have emotions naturally like humans, birds, animals, fishes, plants and so on. Sometimes, non-living things also can express sentimental emotions. Lot of researches and epic writers have written about emotions in different angles. The humans emotions may differ or expose based on culture, life style, living habits, food habits, behavioural habits, experience and knowledge. Emotional expression may be differing based on experience and interpretations. Pitch of vocal speech is a significant parameter for categorising emotions in speech. This paper is special for finding the pitch values in Tamil speech(voice modulation), frequency of speech, length of frequency and for the Tamil speechs recorded voice to text conversion.
Keywords: Tamil; Emotions; Pitch; Recognition; Text; Analysis.
A classed approach towards rainfall forecasting: Machine learning method
by Shanu Khan, Vikram Kumar, Sandeep Chaurasia
Abstract: The interest for precipitation anticipating has turned into a huge element in the outline of rainfall runoff and other hydrological models. As of now, the artificial neural system (ANN) is the most well-known model that is utilized to evaluate rainfall using different climatic parameters. However, classed approach, called the extreme learning machine (ELM) algorithm, has been introduced in this present paper and ELM-based learning framework is used to predict rainfall-runoff forecasting. Extreme learning machine algorithm is much faster as compared to the artificial neural system, and outcomes in a high generalization competence. In view of these outcomes we assert that out of the machine learning calculations tried, the ELM was the more expeditious tool for the forecast of rainfall and its related properties.
Keywords: data mining; artificial neural network; extreme learning machine; rainfall-runoff prediction.
Special Issue on: Nature-inspired Optimisation Algorithms
Total Memory Optimiser: Proof of concept and compromises
by Clerc Maurice
Abstract: For most usual optimisation problems, the Nearer is Better assumption is true (in probability). Classical iterative algorithms take this property into account, either explicitly or implicitly, by forgetting some information collected during the process, assuming it is not useful any more. However, when the property is not globally true, i.e. for deceptive problems, it may be necessary to keep all the sampled points and their values, and to exploit this increasing amount of information. Such a basic Total Memory Optimiser is presented here. We show on an example that this technique can outperform classical methods on deceptive problems. As it gets very computing time expensive when the dimension of the problem increases, a few compromises are suggested to speed it up.
Keywords: stochastic optimisation; nearer is better; deceptive problems; signatures.
Fibonacci series based local search in spider monkey optimisation for transmission expansion planning
by Ajay Sharma, Harish Sharma, Annapurna Bhargava, Nirmala Sharma
Abstract: The power system is a complex interconnected network which consists of four components: generation, distribution, transmission, and load. The loads may be varying in nature. For supplying these loads, with an aim of minimum losses in transmission and distribution, the additional transmission lines are required to be added for expansion. To find out the optimum locations of these additional lines are a complex and challenging task. Swarm intelligence motivated algorithms have been proved to be efficient to deal this type of optimisation problem. Therefore this work applies, a recent swarm intelligence motivated algorithm namely, spider monkey optimisation (SMO) to identify the optimum locations for the additional lines in the system. Further, to augment the solution search capacity of SMO algorithm, a Fibonacci series inspired local search strategy (FLS) is proposed and incorporated with SMO. The modified SMO is named as Fibonacci inspired spider monkey optimisation algorithm (FSMO). The authenticity of the suggested FSMO is analyzed through statistical analysis over 20 benchmark functions. Further, both the algorithms, FSMO and SMO are applied to solve the transmission expansion planning (TEP) problem for IEEE-24 bus system. The reported results are judged against other state-of-art algorithms for solving TEP issues.
Keywords: Swarm intelligence; Fibonacci inspired local search; Transmission expansion planning; Nature inspired algorithm; Loss minimization.
Enhanced Grey Wolf Optimization Algorithm for Constrained Optimization Problems
by Sankalap Arora, Himani Joshi
Abstract: Grey Wolf Optimizer (GWO) is a recent, fast and easy-to-implement, nature inspired meta-heuristic optimization algorithm that focuses on social behaviour of grey wolves. GWO algorithm is prominent in terms of finding global optima without getting trapped in premature convergence. In order to find a fast convergent behaviour of GWO, an Enhanced Grey Wolf Optimization (EGWO) algorithm is proposed in this paper. Basically, GWO is modified in two ways in this study, first, to improve exploitation capability of GWO, the hunting mechanism makes the best use of the global best solution i.e. alpha and secondly, a random parameter of existing GWO algorithm is emended in order to produce promising results compared to state-of-the-art algorithms. To validate the effectiveness of proposed EGWO algorithm, penalty function is consolidated and diverse experiments are executed on different constrained benchmark functions of different complexities and characteristics. Further, a classical engineering design problem (pressure vessel) is solved using the proposed algorithm. The performance evaluation of proposed EGWO algorithm along with other standard meta-heuristic optimization algorithms proved that the proposed EGWO algorithm to be a competitive algorithm in the field of nature inspired meta-heuristic optimization algorithms.
Keywords: Constrained optimization problem; Penalty function; Optimization algorithm; Grey wolf optimizer.
A Hybrid Optimization Algorithm Based on Butterfly Optimization Algorithm and Differential Evolution
by Sankalap Arora, Satvir Singh
Abstract: Butterfly Optimization Algorithm (BOA) is a newcomer in the family of nature inspired optimization algorithms. Although, it is an effective algorithm, still, like other population based optimization algorithms, it encounters two probable problems; (1) entrapment in local optima and (2) slow convergence speed. In order to increase the potential of the algorithm, it is hybridized with an efficient algorithm, Differential Evolution (DE), which accelerates the global convergence speed to the true global optimum while preserving the main feature of the basic BOA. In this paper, a novel hybrid algorithm based on BOA and DE, namely BOA/DE is proposed to solve numerical optimization problems. The proposed algorithm has advantages of both BOA and DE which enable the algorithm to balance the tradeoff between exploration and exploitation which produces efficient results. Engineering design problem and standard benchmark functions are employed to validate the proposed algorithm and according to the simulation results, the performance of the hybrid algorithm is superior to or at least highly competitive with the standard BOA and DE.
Keywords: Butterfly Optimization Algorithm; Differential Algorithm; Numerical Optimization; Benchmark Functions; Engineering Design.
Particle Swarm Optimization based Contextual Recommender Systems
by Mohammed Wasid, Rashid Ali, Vibhor Kant
Abstract: Collaborative Filtering (CF), the widely used and implemented technique in the area of Recommender Systems (RS), provides useful recommendations to users based on their similar users. However, CF has been investigated and improved extensively over the past years, but it fails in many cases and cant handle multiple issues like cold-start and sparsity problems due to the absence of user-item rating information. Further, it has been seen that the contextual information plays a significant role for generating user relevant situational recommendations but the incorporation of contextual information into CF directly is the major problem in RS. This paper is an effort toward developing recommendation strategy based on contextual fuzzy CF by utilizing Particle Swarm Optimization (PSO) algorithm. This work has been completed in two fold. First, we incorporate contextual information into fuzzy CF algorithm through context modeling approach. Second, we extend the previous method by employing PSO algorithm in order to learn user weights on various hybrid fuzzy features for enhancing the performance of CF technique. Effectiveness of our proposed recommendation strategy is established through experimental results in terms of mean absolute error and coverage performance measures using the LDOS-CoMoDa dataset.
Keywords: Collaborative Filtering; Context-Awareness; Cold start; Particle Swarm Optimization; Sparsity; Recommender Systems.
Optimal QFT Controller and Pre-Filter for Buck Converter using Multi-objective Genetic Algorithm
by Nitish Katal, Shiv Narayan
Abstract: Buck converters are one of the most widely used convertors in applications that are dependent upon constant load voltage supply like power electronics, drive, UPS system etc. But these convertors inherent non-linarites because of the switching operation and continuous operation leads to the introduction of parametric uncertainties, making it difficult to assure quality output overtime. In order to mitigate such effects, in this paper QFT controller has been designed. The paper explores a templates and bounds free approach for the designing the QFT controller. The design problem has been posed as a multi-objective optimization problem and solved using genetic algorithm (nsGA-II). The designed controller has been implemented for buck converter for variable input voltage and load variations. As a Pareto optimal set (POS) of solutions are obtained at the end of optimization. The use of level diagrams has also been explored for choosing the ideal solution from POS. The results obtained have been compared with the PID controller tuned using classical method of Ziegler Nichols. As per the simulation results obtained, the designed controller offers a robust response to parametric uncertainties and also has very less current and voltage ripples; whereas the Ziegler Nichols controller fails to offer a steady output.
Keywords: Quantitative feedback theory; robust stability; buck converter; level diagrams; multi-objective genetic algorithm.
Quality Improvement of Electrochemical Discharge Machining Process using Firefly Algorithm: A Case Study
by Debanjan Maity, Bappa Acherjee, Arunanshu S. Kuar
Abstract: In this paper, the firefly algorithm is employed with the RSM (response surface methodology) to investigate the ECDM (electro chemical discharge machining) process in terms of effects of the process parameters on the output quality characteristics, as well as, to set the optimal process parameters for achieving the better product quality. Firefly algorithm is a powerful nature-inspired metaheuristic algorithm that can solve complex optimization problems, both, single objective and multi objective optimization problems. It is noticed that firefly algorithm can find the optimal solution, efficiently, within very less timescale. It is also very effective in predicting the parametric trends of responses. The results obtained by using firefly algorithm for parametric optimization of ECDM are compared with those derived by the past researchers, which prove the applicability and effectiveness of this algorithm in enhancing the performance of the ECDM process.
Keywords: Electrochemical discharge machining; metaheuristic algorithm; firefly algorithm; RSM; optimization.