15
RHINE WAAL UNIVERISTY OF APPLIED SCIENCES Faculty of Communication and Environment Advanced Logistics Concepts for Production and Supply Winter Semester 2014/2015 Prof. Dr. Andreas Schürholz Title of Paper: Decentralized Control in Nature for Production and Logistics Syed Shahzaib Raza 16277 January 05, 2015

Decentralized control in nature for production and logistics

Embed Size (px)

Citation preview

Page 1: Decentralized control in nature for production and logistics

RHINE WAAL UNIVERISTY OF APPLIED SCIENCES

Faculty of Communication and Environment

Advanced Logistics Concepts for Production and Supply

Winter Semester 2014/2015

Prof. Dr. Andreas Schürholz

Title of Paper:

Decentralized Control in Nature for Production and Logistics

Syed Shahzaib Raza

16277

January 05, 2015

Page 2: Decentralized control in nature for production and logistics

List of Contents

1. Introduction: ................................................................................................................ 1

2. Swarm Intelligence ...................................................................................................... 1

2.1 Ant Colony Optimization (ACO) ............................................................................................. 2

2.1.1 Procedure ................................................................................................................ 3

2.1.2 Applications ............................................................................................................. 3

2.2 Particle Swarm Optimization (PSO) ....................................................................................... 3

2.2.1 Procedure ................................................................................................................ 4

2.2.2 Applications ............................................................................................................. 5

2.3 Artificial Bee Colony (ABC) .................................................................................................... 5

2.3.1 Procedure ................................................................................................................ 6

2.3.2 Applications ............................................................................................................. 6

2.4 Firefly Algorithm (FA) ............................................................................................................ 6

2.4.1 Procedure ................................................................................................................ 7

2.4.2 Applications ............................................................................................................. 7

2.5 Cuckoo Search (CS) ................................................................................................................ 7

2.5.1 Procedure ................................................................................................................ 8

2.5.2 Applications ............................................................................................................. 8

3. Conclusion ................................................................................................................... 9

4. Recommendation ......................................................................................................... 9

References ......................................................................................................................... 10

Page 3: Decentralized control in nature for production and logistics

List of Abbreviations

ABC Artificial Bee Colony

ACO Ant Colony Optimization

CS Cuckoo Search

FA Firefly Algorithm

JSSP Job Shop Scheduling Problem

NP Non-deterministic Polynomial

PSO Particle Swarm Optimization

SI Swarm Intelligence

Page 4: Decentralized control in nature for production and logistics

List of Figures

Figure 1 Generalized procedure of ACO ............................................................................ 3

Figure 2 Generalized procedure of PSO ............................................................................. 4

Figure 3 Generalized procedure of ABC ............................................................................ 6

Figure 4 Generalized procedure of FA ............................................................................... 7

Figure 5 Generalized procedure of CS ................................................................................ 8

Page 5: Decentralized control in nature for production and logistics

1

1. Introduction: Nature has inspired scientists in different ways. Biologists and computer scientists found

ways of inspiration from the behaviour of living organisms in their groups. The learning

outcomes from this behaviour of living organisms developed new concepts and

algorithms. They can be based on swarm intelligence (SI), biological, chemical or

physical systems. These algorithms are referred as meta-heuristic optimization techniques

which have significant role in production and logistics industry. Many of these

algorithms are efficient and helpful in solving real-world problems. The nature-inspired

meta-heuristic optimization techniques which are based on swarm intelligence have

received great attention in recent years. Swarm intelligence emerged from the collective

behavior of the social insects like ants, bees, wasps and other animals like flocks of birds

or fish. SI based algorithms have decentralized control, self-organization and information

sharing between the agents and learning which provides the high efficiency to the

complex systems. The algorithms have iterative procedure, which after number of

iterations converges to the optimal solution for the problem. SI based algorithms such as

particle swarm optimization, ant and bee colony, cuckoo search and firefly algorithms

have advantages over other algorithms. These SI-based algorithms can be classified on

different levels depending upon the complexity of the problem. This paper mainly

focuses on the literature review of SI-based algorithms. After the literature review there is

a short summary of SI-based algorithms which are discussed in the report. are divided

into groups of general problems and their application procedures. Further the paper

describes the comparison of SI-based algorithms.

2. Swarm Intelligence Swarm Intelligence (SI) is a discipline of artificial intelligence (AI) which consists of

intelligent multi agents emerged by studying the collective behaviour of social insects

and animals. The agents work in colonies with cooperation which are non-sophisticated

individuals and achieve the optimal solution of complex tasks. The individuals interact

with the members and perform actions to solve the problem. The common features of the

cooperative activities are decentralized control, self-organization and information

sharing. The examples can be taken from the potential of wasps to build their nests, the

Page 6: Decentralized control in nature for production and logistics

2

formation of ants to search food, and organization of birds in a flock. A bird in a flock is

can be considered as autonomous as it has no commands to follow. It involves itself in a

flock and interacts with it mates and perform the task for searching of food avoiding any

collisions with neighbours. In the same way swarm intelligence has autonomous agents

which are subsystems. They communicate with the environment which typically consists

of its neighbours and performs the tasks itself. The advanced view of swarm proposes the

N number of agents that are supportive to each other in order to achieve the optimal

solution of a problem. These optimization techniques have great importance in world of

science and industry. Some of the application areas of these optimization techniques are

manufacturing, logistics, communication, networks, transportation and informatics. There

are various optimization techniques which are being implemented in industries. Some

delivered optimal solutions to the real problems efficiently. Table 1 represents the

popular optimization techniques of swarm intelligence whereas few of them are discussed

in the report.

Algorithm Author Reference

Ant Colony Optimization Dorigo 4.

Particle Swarm Optimization Kennedy and Eberhart 7.

Artificial Bee Colony Karaboga and Basturk 8.

Firefly Algorithm Yang 11

Cuckoo Search Yang and Deb 14

Table 1 Algorithms of Swarm Intelligence

2.1 Ant Colony Optimization (ACO)

The first ant colony optimization algorithms were introduced by Marco Dorigo and his

colleagues in early 90s by the observing ant colonies. Ants are social insects and live in

colonies. The focus of ants is on the survival of the colony rather than the individuals.

The foraging behaviour of ants inspired the researchers for developing the concept of ant

colony optimization which was initiated for the application of discrete optimization

problems. The basis of this behaviour is the communication between the ants by using

pheromone, a chemical substance which allows indirect communication to find the

shortest path between the food and their nest. ACO was implemented in the field of

Page 7: Decentralized control in nature for production and logistics

3

communications to solve the problems in networking and later on also used for

continuous optimization problems. This optimization technique solved many problems in

routing and load balancing. ACO algorithms are meta-heuristic as they solve the complex

problems at upper level. The ACO algorithm consists of the initial condition which is the

nest and food is represented as terminal condition. The considered as agent which moves

in a network so that it could drop the pheromones over the interface and selects the node

of the network for further steps. On the other hand the application of ACO also used in

manufacturing to minimize the time and number of work stations.

2.1.1 Procedure

2.1.2 Applications

ACO can be applied to various real-world problems. This algorithm focuses on the

shortest path finding which is found to be useful in solving complex problems. The

popular application areas of ACO are travelling salesman problem, scheduling, graph

colouring, constraint satisfaction, and routing in telecommunication networks.

2.2 Particle Swarm Optimization (PSO)

A population based stochastic optimization technique which is modelled on the behaviour

of social animals like flock of birds or school of fish is described as particle swarm

Set current position

Find the best point until the best path is discovered

Evaporate pheromone

Update path pheromones until the maximum iterations reached

Figure 1 Generalized procedure of ACO

Page 8: Decentralized control in nature for production and logistics

4

optimization. PSO has been found to be very interesting for scientists and researchers as

it is efficient in solving optimization problems. James Kennedy and Russell Eberhart

proposed PSO in 1995. The PSO emerged from the inspiration from fish and birds in

flocks where particles behave as individuals of a swarm to seek optimal solution of the

problem. The particles position themselves according to their neighbours and keep

updating their velocities with respect to the neighbours regardless of the space. The focus

of the swarm is on the area of search which contains high-quality solutions. The three

main characteristics of PSO are updating individuals in parallel, the value of each cell

depends on the old values and its neighbour, and all the same rules implies on all the cells

for updating. PSO delivers the quality solution of the problem and the particles search

considering the position and velocity that is advised. The principle is based on the flock

of birds which is in search of food. The members of the flock adjust their positions and

velocities according to their neighbours while the space is not taken into account.

2.2.1 Procedure

Initialize the particles

Calculate the fitness values for each particle

If the current fitness value of particle is better than previous then assign the new value otherwise keep

the previous value

Compare the best position of a particle with the position of the swarm, if the position of the particle is

better than assign the position of the particle as best position of swarm

Calculate the velocity of each particle

Use the velocity of each particle to update its data values

Repeat the process until the target

Figure 2 Generalized procedure of PSO

Page 9: Decentralized control in nature for production and logistics

5

2.2.2 Applications

PSO can be applied to various industries and can be very efficient. The general

applications of PSO are transportation, power systems, data mining, network design,

informatics and location finding.

2.3 Artificial Bee Colony (ABC)

Artificial bee colony (ABC) also referred as bee colony optimization was introduced by

Karaboga in 2005. Just like ACO and PSO are inspired by social life of ants and birds,

ABC is inspired by the social behaviour of bees in colonies. This meta-heuristic

algorithm initiated a new way of optimization to solve complex problems. The function

of ABC is described considering the bee’s behaviour in colony. First the bee surveys the

areas to find the food. When the food is discovered, the bee starts dancing in order to

inform the bees in the colony. The bees of the respective colony gather the food and get it

into the hive. There are three actions which they perform like searching for other food

location while abandoning the previous source of food, inform the nest mates before

returning to the food source, or continue the hunting of food. This foraging behaviour of

bees can be applied technically to solve many complicated problems of engineering,

transport, optimization and computational sciences. ABC is used by the researchers to

solve many complex problems including logistics and found to be efficient as compared

to other algorithms like PSO.

Page 10: Decentralized control in nature for production and logistics

6

2.3.1 Procedure

2.3.2 Applications

ABC is found to be very effective as compared to other algorithms especially in the areas

of bioinformatics, image processing, clustering, economic dispatch problem,

optimization, scheduling, vehicle routing and engineering design problems.

2.4 Firefly Algorithm (FA)

The firefly algorithm is a nature inspired algorithm which was introduced by Yang in

2007 based on the behaviour of fireflies and their flashing light pattern which helps in

finding mates and catching the attention of their prey. The swarm contains individuals

which move towards brighter locations with an objective to achieve the optimal solution

efficiently. The firefly algorithm is based on the attributes that artificial fireflies are

unisex, the more bright the flash light is the more attractive the firefly is which decreases

the distance as the most bright firefly convinces its neighbours to move towards it, and

the goal that is to be optimized can be the brightness of the flashing light. The process of

Initialize food positions

Select the food sources by determined by the neighbours of the employed bees

Store the location of food when all onlookers are distributed

Look for the locations of abandoned food sources

Create new positions for the abandoned food sources

Repeat the process until the termination criteria is satisfied

Figure 3 Generalized procedure of ABC

Page 11: Decentralized control in nature for production and logistics

7

iterations start from the lower intensity of light towards the higher intensity while the

distance of movement depends on the attractiveness of the firefly and the new firefly is

updated iteratively after evaluation. This process is repeated till the optimal solution is

obtained.

2.4.1 Procedure

2.4.2 Applications

FA has proved to be very efficient and the advancement in the literature shows the

potential of FA algorithm to deliver optimal solutions. The general application areas of

FA is JSSP, NP-hard, digital image compression and processing, antenna design and

multi-objective load dispatch problems.

2.5 Cuckoo Search (CS)

Cuckoo search (CS) is one of the latest optimization techniques which was developed

by Xin-She Yang and Suash Deb in 2009. This algorithm is natured inspired based on

the cuckoo species of birds and has been proved to be more efficient than PSO in

recent studies. Some species of cuckoo lay their eggs in the nests of other birds and

Create initial population of fireflies

Evaluate the fitness level of all the fireflies

Update the fitness value of fireflies

Rank the fireflies and update the position

Repeat the process until the maximum iterations number is reached

Figure 4 Generalized procedure of FA

Page 12: Decentralized control in nature for production and logistics

8

may remove the eggs of other birds to increase the probability of hatching of their

own eggs. The algorithm was developed which is based on the reproduction strategy

of cuckoo. This algorithm has three main rules i.e. a cuckoo lays one egg at a time

and selects the nest randomly to dump its egg, the nests which are best and contain

eggs of high quality will be carried over to the next generations, the host nests are

fixed in numbers and the cuckoo discovers its laid egg by the host bird. There can the

chance that the host bird gets rid of the egg or may abandon the nest to build its new

nest. These rules can be technically applied considering the egg in a nest as a solution

to the problem, while one cuckoo can lay one egg so it proposes one solution to reach

the goal of new and better solution replacing the old ones suggested. This can solve

many complex problems and may contain set of solutions to the problem.

2.5.1 Procedure

2.5.2 Applications

The CS algorithm is one of the latest algorithms and has been successful in solving real

world problems efficiently. The algorithm can be applied in the areas of pattern

recognition, job scheduling, networking, software testing, optimal path finding. This

algorithm is also applied in the business and health sector to deliver optimal solutions.

Start cuckoos with eggs

Find the nests and lay eggs in different nests

Repeat unless maximum population value is reached while some of the eggs are destroyed

Check the surviving eggs in the nests

Repeat the process until the condition is satisfied

Figure 5 Generalized procedure of CS

Page 13: Decentralized control in nature for production and logistics

9

3. Conclusion The SI-based algorithms have found to be very helpful in solving the real problems

efficiently. These algorithms are applied in different areas with different strategies to

deliver the optimal solutions. After the literature review of the algorithms, it can be

concluded that these algorithms have many advantage over each other. Considering the

algorithms for shortest path finding, ACO and CS can be implemented whereas the

output of these algorithms may vary from each other. The algorithms for job scheduling

are ABC and CS. For network designing problems, PSO and CS can be very helpful. FA

can be helpful for image processing whereas ACO and PSO are efficient in data

clustering. Thus there are various implementation areas where these algorithms can be

categorized and compared to solve the problems. These optimization techniques have

advantages over each other depending upon the area of application of the problem. The

complexities of the algorithms also vary from each other. These complexities can be

reduced while applying different techniques to achieve the best optimal solution for the

complex problems.

4. Recommendation

While considering the SI-based algorithms, there are many solutions that can be obtained.

These algorithms are being applied in different industries at different levels. It can be

recommended that studies of the applications of different algorithms in one industry can

be done. This can help the industries in evaluating the algorithms and their efficiency

while applying in the same industry. On the other hand the complexities of the algorithms

to a specific problem can also be determined. This would the industries to get optimized

solutions to a specific problem. On the other hand, the focus of the report was on SI-

based algorithms, whereas the literature review on other algorithms can also be carried

out. This would be helpful in evaluating the algorithms of other than nature inspired, and

the application areas of those algorithms can also give broader view of the advantages of

algorithms. Furthermore, there can be new sectors where these algorithms can be

implemented and could be helpful. The research on the applications of these algorithms

would be beneficial as well.

Page 14: Decentralized control in nature for production and logistics

10

References 1. A. Sadrnia et al., “A Review of Nature-Based Algorithms Applications in Green

Supply Chain Problems”, LASCIT International Journal of Engineering and

Technology, vol. 6, No. 3, June 2014.

2. G. Beni, “The concept of cellular robotic systems”, In Proceedings of the IEEE

International Symposium on Intelligent Systems, pg 57-62, IEEE Press,

Piscataway, NJ, 1988.

3. Christian Blum and Xiaodong Li, “Swarm Intelligence in Optimization”

4. Marco Dorigo, “Optimization, learning and natural algorithms”, Ph. D. Thesis,

Politecnico di Milano, Italy, 1992.

5. Yang Liu and Kevin M. Passino, “Swarm Intelligence: Literature Overview”,

Dept. of Electrical Engineering, The Ohio State University, March 30, 2000.

6. Pontus Svenson et al., “Swarm Intelligence for logistics: Background”, Swedish

Defence Research Agency, February 2004.

7. James Kennedy and Russel Eberhart, “Particle swarm optimization”, In Neural

Networks Proceedings of IEEE International Conference, vol. 4, pages 1942-

1948, IEEE, 1995.

8. Dervis Karaboga and Bahriye Basturk, “A powerful and efficient algorithm for

numerical function optimization: artificial bee colony (abc) algorithm”, Journal of

global optimization, vol. 39(3), pages 459-471, 2007.

9. Sagar Tiwari et al., “Algorithms of Swarm Intelligence Using Data Clustering”,

International Journal of Computer Science and Information Technologies, vol.

4(4), pages 549-552, 2013.

10. X. S. Yang, “Swarm intelligence based algorithms: a critical analysis”,

Evolutionary Intelligence, vol. 7, no. 1, pp. 17-28 (2014).

11. Xin-She Yang, “Firefly algorithm, stochastics test functions and design

optimization”, International Journal of Bio-Inspired Computation, vol. 2(2),

pages 78-84, 2010.

12. Ajith Abraham et al., “Swarm Intelligence Algorithms for Data Clustering”

13. Hemant Nagpure et al., “The Applications Survey on Bee Colony Optimization”,

International Journal of Computer Science and Information Technologies, vol. 3,

pages 5137-5140, 2012.

14. Xin-She Yang and Suash Deb, “Cuckoo search via L´evy flights”, In Nature &

Biologically Inspired Computing 2009, NaBIC 2009, World Congress on, pages

210-214 IEEE, 2009.

Page 15: Decentralized control in nature for production and logistics

11

15. Asaju La’Aro Bolaji et al., “Artificial Bees Colony Algorithm, Its Variants and

Applications: A Survey”, Journal of Theoretical and Applied Information

Technology, vol. 42, no. 2, January 20, 2013.

16. Micheal Brand et al., “Ant Colony Optimization Algorithm for Robot Path

Planning”, International Conference on Computer Design And Applications,

2010.

17. Ling Ai Wong, “Optimal Battery Sizing in Photovoltaic Based Distributed

Generation Using Enhanced Opposition-Based Firefly Algorithm for Voltage Rise

Mitigation”, The Scientific World Journal, vol. 2014, article752096.

18. Saeed Balochian and Eshagh Ebrahimi, “Parameter Optimization via Cuckoo

Optimization Algorithm of Fuzzy Controller for Liquid Level Control”, Journal

of Engineering, vol. 2013, article 982354

19. S. Talatahari et al., “Optimum Design of Gravity Retaining Walls Using Charged

System Search Algorithm”, Mathematical Problems in Engineering, vol 2012,

article 301628.

20. J. Kwiecień and B. Filipowicz, “Firefly algorithm in optimization of queueing

systems”, Bulletin of the Polish Academy of Sciences, Technical Sciences vol. 60,

No. 2, 2012.

21. Aphirak Khadwilard et al., “Application of Firefly Algorithm and Its Parameter

Setting for Job Scheduling”, The Journal of Industrial Technology, vol. 8, no. 1,

January – April 2012.

22. Theofanis Apostolopoulos and Aristidis Vlachos, “Application of the Firefly

Algorithm for Solving the Economic Emissions Load Dispatch Problem”,

International Journal of Combinatorics, vol. 2011, article 523806.

23. Ahmed S. Tawfik et al., “One Rank Cuckoo Search Algorithm with Application

to Algorithm Trading Systems Optimization”, International Journal of Computer

Applications, vol. 64, no. 6, February 2013.

24. Iztok Fister Jr. et al., “A Brief Review of Nature-Inspired Algorithms for

Optimization”, Elektrotehniˇski Vestnik 80(3): 1–7, English Edition, 2013.