| In recent years, as the complexity, disciplinary, Multiple extremum, hard to modeling and other problems in practical engineering becoming more and more prominent, traditional optimization methods have been unable to meet the complicated and polytropic actual needs, swarm intelligence optimization algorithm arises at the historic moment and get rapid development. Among them, Passino proposed the bacterial foraging optimization(BFO) in 2002 inspired by e. coli bacteria’s foraging behavior. It is a simple and effective stochastic global optimization algorithm, which attracts extensive attention because of its advantages such as good parallelism and strong local search ability, and it also pointed out new directions for many actual engineering problems that traditional optimization methods cannot satisfy. However, the time BFO proposed is relatively late, domestic researcher hadn’t began to study until 2007, the algorithm still have some defects like the weak optimization ability in global spatial, easy to fall into local optimum when solving high-dimensional multimodal problems, so BFO need to be further in-depth study.This paper studies the mechanisms of bacterial foraging optimization principle, and for the drawbacks of classic BFO, proposes an improved algorithm——Hybrid Bacterial Foraging Optimization Algorithm with Variable Probability(VPBFO). The randomly generated population is improved to the randomization good point set population which provide a more uniform, nonredundant, and diversified solution space. Inspired by the particle swarm algorithm, the chemotaxis direction strategy reflecting the bacteria individual cognitive and social learning with variable weight coefficient, combined with variable number of winding strategy is proposed so as to improve the precision and searching efficiency of the algorithm. The variable probability of migration operations is designed to helping flora quickly jump out of local extremum, and not to disturb current evolution mechanism and information that bacteria have a lready obtained, thus the convergence speed of BFO have been increased and the premature convergence have been avoided. Experiment result indicates that the algorithm outperforms the classic algorithm not only in terms of the solution accuracy, but also global convergence ability and convergence speed. In conclusion, VPBFO has a higher efficiency.The application of swarm intelligence algorithm in engineering optimization problem and image processing technology is been studied. A new image enhancement thought is proposed so that the problem of insufficient detail processed by traditional image enhancement technology and the offset of histogram distribution after the transformation is solved. In this thought, image enhancement problem has been converted into bacterial foraging optimization issue, and parameters of incomplete Beta function has been used as initial flora, the gray-scale image enhancement problem which dimensions are high and variational has turned into a fixed two-dimensional parameters optimization problems of incomplete Beta function. Simulation results demonstrate the effectiveness of the method. Comparing with other methods, enhanced image make the details be more natural, the histogram distribution more uniform, and the light and dark areas more reasonable.In view of VPBFO, how to select and adjust the control parameters properly is one vital point to affect the performance of the algorithm. Besides, when optimizing some high-dimensional functions, the improved algorithm still has space to improve in the aspect of precision performance. In subsequent studies, the law of parameter settings of VPBFO will be researched to further improve its performance. In addition, as a new algorithm, BFO still have many untapped areas in engineering optimization, such as image restoration, color image enhancement, support vector machine and so on, in follow-up work, VPBFO can be considered applying to other research fields. |
PDF Full Text Request