Font Size: a A A

Research And Application Of Artificial Bee Colony Algorithms Guided By Their Evolving Information

Posted on:2019-04-07Degree:DoctorType:Dissertation
Country:ChinaCandidate:J X NingFull Text:PDF
GTID:1488306344959419Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
With the rapid development of science and technology,the optimization problem has become very important in the field of artificial intelligence and some other related fields.In recent years,it has become more and more complex,needing to consider multiple conflicting objectives and satisfy multiple constraint conditions simultaneously.With the increasing of the objective numbers,the difficulty of solving it is also increasing.This makes the solving methods of the many-objective optimization problem become a recent research hotspot.In addition,how to model and solve complex problems in other fields based on optimization theory is also worth studying.Nowdays,the swarm intelligence algorithm has become a main way for various optimization problems.It can solve some kinds of complex optimization problems quickly and approximately by simulating the behavior of some biological groups.Its most important feature is that it doesn't need to establish an accurate model of the problem itself,so it is suitable for solving those problems which are difficult to establish effective formal models and difficult to solve effectively or even cannot be solved by traditional artificial intelligence techniques.A variety of swarm intelligence algorithms have been proposed by researchers currently,such as particle swarm optimization,ant colony optimization,fish swarm algorithm and artificial bee colony algorithm.The artificial bee colony algorithm has been widely applied to solve optimization problems in many fields and shows strong competitiveness.Therefore,the research on improving its performance has a certain practical significance.Lots of information will be produced during the running of artificial bee colony algorithm,by which the search space problem related features can be timely accessed.If this kind of information can be effectively used in the solving process,the performance of the algorithm can be further improved.For this reason,aiming at the deficiencies about wasting of search resources due to completely randomness of the scout bee foraging and selecting neighbors of onlookers,and the problem of loss of selection pressure caused by neglecting some high quality solutions in multi-objective optimization,the process information that can be obtained during the algorithm running will be used to further improve its performance and applied to solve the problem of protein molecular docking and primer design in the field of bioinformatics.(1)Aiming at the problem of wasting searching resources in the existing artificial bee colony algorithm for solving the single objective optimization problem,caused by the scout bee searching for food sources blindly and the way the scout and the employed bee select an other food source to produce a candidate food source randomly,a single objective artificial bee colony algorithm based on evolving direction guidance is proposed to further improve its performance by increasing the utilization of search resources.According to the food source updating direction,the foraging strategies of onlookers,employed bees and scouts are studied.Then,the proposed algorithm is applied to solve the protein molecular docking problem.(2)Aiming at the problem of poor diversity of the non-dominated solutions caused by ignoring some high-quality food sources and the repeated foraging in some unpromising search areas in solving multi-objective optimization problems.From the perspective of how to effectively use the abandoned food source location information in the optimization process,a multi-objective artificial bee colony algorithm based on the exploration density is proposed in this thesis.A food source formation mechanism for global selection is designed,and the construction method of the exploration density tree based on the abandonment of food sources is a focal point.Based on it,a searching strategy of the scout bee based on search density is studied.(3)Aiming at the problem of performance degradation caused by the failure of food source selection strategy in the existing multi-objective artificial bee colony algorithm due to the explosive growth of non-dominant solution for solving the many-objective optimization problem,an artificial bee colony algorithm based on decomposition is proposed in this thesis from the perspective of how to effectively use the optimal solution of the sub-problems updating information.A food source evaluation method based on the amplitude of food source quality improvement and the generating model of sub-problems based on adaptive reference vector adjustment strategy are studied.Then,a scout bee foraging strategy for many objective optimization problem is implemented.(4)Aiming at primer design in molecular biology,the optimization model for it is studied and the corresponding discrete search space representation model is established according to the principle of primer design.Then,inspired by the basic idea of ant colony optimization,a hybrid artificial bee colony optimization model is established to solve the problem of primer design and the candidate food source construction method is designed.An onlooker foraging strategy based on the guidance of the global optimal path updating information is studied.This thesis focuses on establishing a representation model of path use frequency to find a promising search area.Based on it,a scout bee foraging strategy based on the search frequency information is designed.Based on the swarm intelligence algorithm,how to improve the performance of single objective,multi-objective and many-objective artificial bee colony algorithm by using process information from the perspective of artificial bee colony algorithm optimization process is studied.In addition,how to apply artificial bee colony algorithm to solve the optimization problem in other fields is also worth studying.Therefore,the proposed algorithm is applied to solve the protein molecular docking and primer design in the biological field.It can be seen that the research in this thesis can not only promote the development of swarm intelligence optimization methods,but also make the artificial bee colony algorithm improved constantly.
Keywords/Search Tags:swarm intelligence, muti-objective optimization, artificial bee colony algorithm, protein molecular docking, many-objective optimization, primer design
PDF Full Text Request
Related items