基于规则学习的林火蔓延元胞自动机模型

由于元胞自动机模型本质上的自组织特性与林火行为很接近,近年来基于元胞自动机的林火蔓延模型引起了广泛的关注.典型的元胞自动机是按照确定的映射规则自主地随时间演进的.当环境发生变化的时候规则需要由人工静态地进行调整、修正,因此效率低、对环境变化的适应能力差,模型的可移植性也不令人满意.这是基于CA的林火蔓延模型难以推广的主要原因之一.本文提出了一种应用人工免疫机制对元胞自动机规则进行自适应调节的方法,并在此基础上建立了一个林火蔓延实验模型.仿真实验表明经过规则学习,该模型具有了一定的火线形状的自纠正能力.

A forestry fire spread model based on evolutional cellular automata

The self-organization nature of the cellular automata is quite close to the behavior of forestry fire.So,cellular automata model is commonly used to simulate forestry fire spreading.However,typical cellular automata run with a set of deterministic rules.Thus in case of the environment change,which demand rule adjustment and is typically done manually and statically,the model tends to be less efficient and inability to fit the environment change.This is one of the main difficulties when using cellular automata model to simulate the forestry fire spreading.The paper presents an immunity-based rule learning algorithm for cellular automata aiming to promote the agility of the model.The experiment shows that with rule learning the model successfully revises the shape of front line whenever obstacle meet.