Modification And Application Of Bacterial Foraging Optimization Algorithm

Optimization is the problems that people encountered frequently in the scientific research and production practice, and a large number of optimization problems has been researched in-depth, then develop its been an important disciplines. Conventional optimization such as steepest descent method, linear programming, simplex method that method gradient-based have high computational efficiency when its objective function of the problem is such circumstances as convex, continuous and differentiable. But in the practical application of logistics distribution center location, the optimal allocation of resources, equipment, plant products, products in areas such as scheduling and layout, there were many large-scale, nonlinear, and more extreme, more constrained, non-convex and other phenomena, which makes it difficult for traditional optimization methods of mathematical modeling to give a group of intelligent bionic characterized by calculation to provide a broad application of the stage, and the birth of a number of simulated biological behavior of the "heuristics", This is one of the typical including genetic algorithms, ant optimization algorithms, particle swarm optimization and bacterial foraging algorithm and so on. Research shows that these group intelligent algorithms more or less exist some convergence defects and problems of"premature", "late" or even "familiar", so many scholars view aimed at the different groups intelligent algorithms in the between each other and determine various ways and means to improve the performance of the optimal algorithm.The main groups in the above intelligent optimization algorithm, GA, ACO, PSO is inspired by objects based on higher organisms, and its form an adaptive artificial intelligence computing technology characterized by "generation + test", but BFO is an algorithms starting from the mechanism of microbial behavior, simulating the bacteria perception to the environment changes, forming a new optimization method. Since the time was too short that micro-organisms intelligent bionic technology advents, there are many gaps in this new type of intelligent bionic algorithm in the international academic community, and it’s far from the adequate attention should receive from academic community. Therefore, this article attempts to analyze the microbial mechanisms and physiological characteristics for modeling and simulation, and to explore the improvement way of this new calculation method, for enriching the bionic optimization algorithm in the field of microbial intelligent computing, and then provided a sense of technology learn to other bionic optimization algorithm, provided us with new ways and new perspectives for the existing biological systems optimization approaches. This research based in normative analysis and experimental research methods, analyze the ideological basis, the main categories and algorithms program of the bionic optimization; and its also discussed the performance comparison test function, the merits of the algorithm performance comparison index and the diversity of the population algorithm iteration metrics of bionic optimization problem. and then we discussed the basic principles and achievement steps of BFO algorithm, analysis and discuss the main problems in the execution of a program showing in BFO algorithm existing operator chemokine breeding and migration, then the algorithm was experimentally proposed algorithm parameter settings useing the test function. Finally, the article seting some experiments to the BFO algorithm based on optimal foraging theory, discussed different feeding methods algorithm performance that the bacteria effects in the process of chemotaxis.Based on above theoretical analysis, the article attempts to implyment algorithm Theoretical integration in swarm intelligence algorithm, attempts to analyze the different features of existing swarm intelligence algorithm, based on group collaboration in higher organisms, the genetic evolution of biological populations and statistical study of the distribution estimated three levels, to improve the test performance of basic BFO from the micro-behavior, genetic improvement and macro guidance, and make it evolve with the coordination of multiple intelligence and learning to adapt, so as to improve the algorithm search speed and accuracy purposes. This paper attempts to build the BFO algorithm not only has some innovation in method, but also has some sense of the theoretical innovation with more positive thinking for the existing smart computing reference. The main thesis research in the following areas:(1) summarizes the basic principles and the main method of group intelligent computing in bionic optimization, analyze the no free lunch theorems (NFL) of existing algorithms to compare the performance, describes the standard test functions and the indicators that the algorithm used to compare the performance and to assessment population diversity, revealing the ideological basis of the group intelligent calculation, giving some evaluation system and theoretical reference for algorithm optimization and its improvement for the follow-up article. (2) study the basic principles in-depth of bacterial foraging optimization, analyze the algorithm limitations in algorithm optimization of the current trend, breeding and migration operator, getting some experience and learn from the analysis and discussion, compare the PSO and BFO in the foraging behavior of the two strategies, use the bacteria energy and fitness to simulate unconventional and conventional feeding strategies to verify the different feeding strategies on the performance of the algorithm impact, concludes with four improvement goals and improving behavioral strategies in bacterial foraging optimization.(3) analyzed the two main acts of competition and collaboration in biological food processes in the food, discussed the collaborative and learning idea of fish-based algorithm and particle swarm algorithm. Algorithm based on the idea of fish group, giving the bacteria the ability to perceive the state population, can tracking the excellent value and move closer to the community center, presented environmental awareness BFO algorithm to improve the problem solution accuracy. PSO-based self-learning and social learning ideas, put forward the collaboration BFO, makes the algorithm has a greater probability to obtain the global optimal solution.(4) analyzed the survival of the fittest, species selection and the genetics theory of biological evolution, discussed a broad evolutionary computation method,and analyzed the main ideas based on genetic algorithms and evolutionary computation. Bacteria based on differential evolution ideas to achieve bacterial population optimal through the difference between individuals within the group cooperation and competition, thus to amend the dimensions degradation of cycle, the differential operator improved significantly BFO algorithm accuracy, robustness and the ability to obtain the global optimum. Based on the idea of life immune system, this article set up clonal reproduction operator based on immune-body, and to clone elite bacteria, to variate high-frequency and to cross random after the period of chemotactic completion, so to guide the operator search, making the algorithm has good applicability on some test function, and can converge quickly to find the global optimum.(5) analyzeing the latest smart way of distribution of estimates, discussing the possibility of estimates algorithms of distribution introducing to intelligent computing, which can take full advantage of the practical problems of prior information, to complete the model from the guiding in the macro ideology. Based on the Gaussian distribution ideas, in the breeding cycle after the end of bacteria chemotaxis, we introducting the concept of Gaussian distribution breeding which to build statistical modeling from the macro based the best part of the bacteria, new algorithm improves significantly the BFO accuracy and robustness, and has great adaptability to some test function. Based on the real bacteria growth curve, breaking three nested frames of BFO algorithm, this article simulate the law of bacterial flora free distribution in the optimization process, established the system model of self-reproduction and self-demise of the bacteria, then privoded some evidence for the above BFO algorithm related to performance from another side. (6) use standard test functions to test the validation of the performance for bacteria foraging optimization algorithm improvement, and designed and developed a corresponding computer program in the MATLAB software platform that attached. For the actual optimization of continuous space and discrete space, we use neural networks prediction model and job shop scheduling problem to verify the performance of the new algorithm, extends the application space of the continuity BFO, and provides a new information processing and intelligent computing tools for weight optimization of neural network model and the job scheduling solution optimization.