自适应变异粒子群优化算法及在新冠肺炎疫情传播预测中的应用

为克服粒子群算法容易陷入局部最优和全局寻优精度不高的缺点,通过对算法的局部寻优和全局寻优的特点进行分析,首先使用正态分布衰减策略改进惯性权重;同时基于算法运行的时间自适应采用不同的基于高斯分布及柯西分布的变异优化策略,解决全局搜索和局部开发能力的不平衡问题,实现了局部寻优和全局寻优的双重优化,满足了提高寻优速度和寻优精...

Adaptive Mutation Particle Swarm Optimization and Its Application in Predicting the COVID-19 Epidemic Transmission

Aiming at the shortcomings that particle swarm optimization is easy to fall into local optimization and low precision.By analyzing the characteristics of algorithm’s local and global optimization, firstly,the Gaussian distribution decay strategy is used to improve the inertia weight.Based on the running time,Gaussian distribution and Cauchy distribution are adopted on mutation strategies adaptively.The algorithmsolves the imbalance between of global search and local search,realizes the dual optimization of local optimization and global optimization,and improvs the convergence speed and accuracy.COVID-19 epidemic situation is predicted by the improved algorithm in order to verify the effectiveness and practicability.The 100 days’ data of COVID-19 daily newly diagnosed patients are collected and the algorithm is used to optimize the parameters of feedforward neural network, so that the trained neural network model can improve the prediction performance.The experiment first compares the improved algorithm with the other five algorithms,and shows the good performance of the improved algorithm.Finally,the algorithm is applied to the prediction of the COVID-19.