基于粒子群改进算法的生物网络建模与优化研究

发布时间：2018-05-15 19:28

本文选题：随机漂移粒子群算法 + 生物网络　；参考：《浙江大学》2017年硕士论文

【摘要】：作为生物学与工程学的交叉学科,合成生物学结合了系统辨识和控制理论诸多特征,包括反馈和振荡行为等。考虑到生物系统的复杂性,仅凭借经验知识不足以准确捕获其主要动态特征并开展深入的分析研究,因此有必要构建生物系统的有效数学模型,从而获得系统的机理特性。当前,以粒子群优化为代表的多种智能优化算法在系统建模领域取得了显著的进展,但依然存在算法易陷入局部最优的问题。为此,本文通过对粒子群算法的拓扑结构进行改进来改善其全局搜索能力,并开展了生物网络的参数估计及结构设计的应用研究。本文的主要工作包括:1.针对粒子群优化算法和随机漂移粒子群优化算法存在陷入局部最优和对参数敏感的问题,结合局部拓扑结构具有增强算法全局搜索能力的特性,提出了冯·诺依曼拓扑结构的随机漂移粒子群优化算法,并通过对经典测试函数的寻优仿真验证了算法的有效性。2.针对生物网络参数估计时存在的非线性难题,应用所提出的具有较强全局搜索能力的算法进行仿真求解,并与其他四种算法的结果进行了比较,结果表明所提出的改进算法能有效提高实例系统的参数估计性能。3.针对合成基因振荡网络的鲁棒性设计问题,提出了一种改进算法的离散型优化方法,结合两步优化思想,在优化确定网络结构的基础上进行了鲁棒性能的优化设计,仿真结果表明,经过鲁棒性能的优化后,设计得到的网络在结构和参数方面的鲁棒性能得到了很大的提高。4.同时考虑结构和参数的影响时,具有同步行为的振荡器耦合网络的优化设计成为一个混合整数优化问题。应用所提出的改进优化算法进行了典型实例的仿真研究,结果表明,经过两步优化设计能够得到具有较强同步特性的耦合网络。
[Abstract]:As an interdiscipline between biology and engineering, synthetic biology combines many characteristics of system identification and control theory, including feedback and oscillatory behavior. Considering the complexity of biological system, it is not enough to capture its main dynamic characteristics and carry out in-depth analysis and research by relying on the knowledge of experience. Therefore, it is necessary to construct an effective mathematical model of biological system to obtain the mechanism characteristics of the system. At present, many intelligent optimization algorithms, represented by particle swarm optimization, have made remarkable progress in the field of system modeling, but there is still the problem that the algorithm is prone to fall into local optimization. Therefore, this paper improves the global searching ability by improving the topology of particle swarm optimization, and studies the parameter estimation and structure design of biological network. The main work of this paper includes: 1. In view of the problem that particle swarm optimization and random drift particle swarm optimization are trapped in local optimum and sensitive to parameters, the local topology has the characteristics of enhancing the global search ability of the algorithm. A random drift particle swarm optimization algorithm for von Neumann topology is proposed. The effectiveness of the algorithm is verified by the optimization simulation of classical test functions. Aiming at the nonlinear problem in the estimation of biological network parameters, the proposed algorithm with strong global searching ability is used to simulate and solve the problem, and the results are compared with the results of the other four algorithms. The results show that the proposed improved algorithm can effectively improve the performance of parameter estimation. In order to solve the problem of robust design of synthetic gene oscillation network, a discrete optimization method based on improved algorithm is proposed. Based on the optimization of network structure, a robust performance optimization design is carried out based on the idea of two-step optimization. The simulation results show that the robust performance of the designed network is greatly improved in terms of structure and parameters after the robust performance optimization. Considering the influence of structure and parameters, the optimal design of oscillator coupling network with synchronous behavior becomes a mixed integer optimization problem. The simulation results show that the coupling network with strong synchronization characteristics can be obtained by the two-step optimization design.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：Q811.4;TP18

【参考文献】