An intelligent method to discover transition rules for cellular automata using bee colony optimisation
An intelligent method to discover transition rules for cellular automata using bee colony optimisation
复制标题
DOI：
10.1080/13658816.2013.823498
复制
发表时间：
2013-10
影响因子：
5.7
通讯作者：
Jianyi Yang;G. Tang;Min Cao;Rui Zhu
中科院分区：
文献类型：
--
作者：
Jianyi Yang;G. Tang;Min Cao;Rui Zhu
本文提出了一种新的，智能的方法来发现基于蜂群优化（BCO-CA）的地理元胞自动机（CA）的过渡规则，可以通过蜜蜂的合作和互动来执行复杂的任务。采用人工蜂群挖掘算法发现转换规则。在BCO-CA中，食物源位置由每个属性的上阈值和下阈值定义，并且每个蜜蜂在每个属性中搜索最佳上阈值和下阈值作为区域。当每个属性中的区域通过运算符“And”连接到另一个节点并链接到单元格状态值时，将组织转换规则。转换规则用逻辑结构语句“IF-Then”表示，该语句清晰明了，易于理解。蜂群优化算法通过迭代过程中的局部和全局搜索，较好地避免了易陷入局部最优的倾向，并且不需要对属性值进行离散化。最后，利用BCO-CA模型对西安咸阳城区的城市发展进行了模拟。初步结果表明，这种BCO方法是有效的捕捉空间变量和城市动态之间的复杂关系。实验结果表明，BCO-CA模型比NULL和ACO-CA模型具有更高的精度，证明了该模型在复杂城市动态变化模拟中的可行性和有效性。
This paper presents a new, intelligent approach to discover transition rules for geographical cellular automata (CA) based on bee colony optimisation (BCO–CA) that can perform complex tasks through the cooperation and interaction of bees. The artificial bee colony miner algorithm is used to discover transition rules. In BCO–CA, a food source position is defined by its upper and lower thresholds for each attribute, and each bee searches the best upper and lower thresholds in each attribute as a zone. A transition rule is organised when the zone in each attribute is connected to another node by the operator ‘And’ and is linked to a cell status value. The transition rules are expressed by the logical structure statement ‘IF-Then’, which is explicit and easy to understand. Bee colony optimisation could better avoid the tendency to be vulnerable to local optimisation through local and global searching in the iterative process, and it does not require the discretisation of attribute values. Finally, The BCO–CA model is employed to simulate urban development in the Xi’an-Xian Yang urban area in China. Preliminary results suggest that this BCO approach is effective in capturing complex relationships between spatial variables and urban dynamics. Experimental results indicate that the BCO–CA model achieves a higher accuracy than the NULL and ACO–CA models, which demonstrates the feasibility and availability of the model in the simulation of complex urban dynamic change.