| Artificial Intelligence and Parallel Computation is developing rapidly during21st Century. Intelligent computation, one of the most important branches of artificial intelligence, is becoming a hot research topic. Famous intelligent algorithms include Genetic Algorithm, Particle Swarm Optimization, Artificial Fish Swarm Algorithm and Artificial Bee Colony Algorithm. Because of their swarm characters, they are also called swarm intelligent algorithm. Parallel computation, compared with serial computation, generally executes multiple commands simultaneously on a workstation. The main purpose of parallel computation is solving large-scale and time-consuming problems. Platforms of parallel computation are usually based on Linux, combining with related technologies such as MPI, CUDA and HADOOP.Intelligent computation has been widely used in our daily life. For instance, Artificial Bee Colony Algorithm can be used to improve the quality of image edge detection while Particle Swarm Optimization can be applied to route the shortest way of Wireless Sensor Network. In addition, Genetic Algorithm is an effective method for recognizing the structure of protein. Thus, intelligent algorithm demonstrates outstanding performances in the real world. On the other hand, due to the development of computer hardware and software, parallel computation has permeated our daily life. As we all know, in the area of meteorology, meteorologist need massive computation resource to make it. Many developed countries have built parallel computation platform in order to forecast the weather precisely. When it comes to the oil exploration technology, the big data technology is widely used, which demands complex algorithms and high-performance computation. Thus, parallel computation technology is becoming the key for improving the performance of oil exploration. Furthermore, biomedicine, especially signal processing and image processing, needs a high performance computation platform. For instance, scientists try to use these technologies to recognize the sequence of DNA.In this paper, we discuss the theories, the improving strategies and applications of Artificial Bee Colony Algorithm, while we also analyze its parallel strategy in details. As we all know, Artificial Bee Colony Algorithm is a novel swarm intelligence algorithm which imitates the bees’ foraging behavior. In nature, bees can find reliable food source effectively. Based on swarm characters, scientists designed Artificial Bee Colony Algorithm so that it can find the optimal solution in the real world. Furthermore, the Parallel Artificial Bee Colony Algorithm has been achieved by using Linux and MPI.The main contributions of this paper are showed as follows:Firstly, basic theories of Artificial Bee Colony Algorithm have been studied in detailed and its improving strategies are introduced in order to overcome its drawbacks. For instance, the strategy of adaptive mutation step size is introduced to boost the effectiveness, when we try to solve the job shop scheduling problem. In addition, the performance of improving algorithms has been compared with that of the initial Artificial Bee Colony Algorithm.As for the application areas, the optimal turning of robust PID controllers can be solved by Artificial Bee Colony Algorithm. In addition, the multi-objective optimization based on Artificial Bee Colony Algorithm is achieved, including Integral Square Error, Maximum Overshoot and Setting Time. The Minimax Method is also introduced in order to design a robust PID controller.According to the nature parallelism of Artificial Bee Colony Algorithm, a Parallel Artificial Bee Colony Algorithm is introduced in order to solve the Travel Salesman Problem. For the reason that achieving a massive Traveling Salesman Problem is really time-consuming, parallel strategy is applied to improving the performance. The parallel platform is established on the Linux operation system, while the data and tasks are allocated by using MPI. |
PDF Full Text Request