遗传算法在甲流SIR模型参数求解中的应用

Application of Genetic Algorithm to SIR Model Parameters

论文以全国甲型H1N1流行性感冒(下简称甲流)疫情数据为实例,讨论了采用SIR模型对甲流的传播过程进行模拟时相关参数的求解问题。分别通过优化的遗传算法(Genetic Algorithm,GA)和模拟退火算法(Simula-ted Annealing Algorithm,SA)求得该非线性模型中的重要参数阈值(日治愈率与日传染率的比值),并由该参数阈值计算出各月患病人数。论文比较分析了两种算法在精度和效率上的优劣,发现遗传算法优于模拟退火。同时模拟结果验证了SIR模型适合甲流疫情的分析模拟。

Genetic Algorithms mimic the process of natural revolution with of inheritance,mutation,selection and crossover techniques in order to solve optimization and search problems.This paper focuses on the validity of applying a Genetic Algorithm（GA） to estimate the parameters for a Susceptible-Infectious-Recovered（SIR） model,based on monthly collected data of New Influenza A（H1N1） infections in China on a provincial scale.By estimating an important threshold value in the non-linear SIR model,both Genetic Algorithm and Simulated Annealing（SA） are adopted to calculate the number of infections.The result shows that GA is a more effective method in SIR model calculation in the case of H1N1,considering both precision and efficiency.Our findings indicate that applying GA to the SIR model is appropriate for data fitting,process simulation and trend prediction for the H1N1 pandemic.