| 基于蚁群算法的多层前馈神经网络 |
| |
| 引用本文： | 洪炳熔,金飞虎,高庆吉.基于蚁群算法的多层前馈神经网络[J].哈尔滨工业大学学报,2003,35(7):823-825. |
| |
| 作者姓名： | 洪炳熔 金飞虎 高庆吉 |
| |
| 作者单位： | 哈尔滨工业大学,计算机科学与技术学院,黑龙江,哈尔滨,150001 |
| |
| 基金项目： | 国家高技术研究发展计划资助项目(2001AA422270) |
| |
| 摘 要： | 反向传播算法是神经网络中应用广泛的一种多层前馈神经网络模型．但算法有求解精度低、搜索速度慢、易于陷入极小的缺点．蚁群算法是一种新型的模拟进化算法，有正反馈、分布式计算、启发性收敛等特性．这些特性使得解题过程加快，易于实现分布式计算．将蚁群算法和神经网络相结合起来，实现了非线性模型的辨识问题及倒立摆的控制．仿真实验表明：用蚁群算法训练神经网络，可兼有神经网络广泛映射能力和蚁群算法快速全局收敛的性能．
|
| 关 键 词： | 蚁群算法 多层前馈神经网络 反向传播算法 非线性系统 倒立摆 收敛性 映射能力 网络训练 |
| 文章编号： | 0367-6234(2003)07-0823-03 |
| 修稿时间： | 2002年9月17日 |
Multi-layer feedforward neural network based on ant colony system |
| |
| HONG Bing-rong,JIN Fei-hu,GAO Qing-ji.Multi-layer feedforward neural network based on ant colony system[J].Journal of Harbin Institute of Technology,2003,35(7):823-825. |
| |
| Authors: | HONG Bing-rong JIN Fei-hu GAO Qing-ji |
| |
| Abstract: | Back Propagation is a kind of feedforwad neural networks widely used in many areas, but it has some shortcomings, such as low-precision solutions, slow search speed and easy convergence to the local minimum points. Ant colony system is a novel simulated evolutionary algorithm. Ant system has positive feedback, distributed computation, and use of a constructive greedy heuristic. These characteristics account for rapid discovery of good solutions and easy to realize distributed computation. The combination of ant system with neural network is adopted so that a nonlinear model can be identified and an inverted pendulum controlled. Simulation results show that extensive mapping ability of neural network and rapid global convergence of ant system can be obtained by combining ant system and neural network. |
| |
| Keywords: | ant colony system neural network system identification inverted pendulum system |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
