基于AEA与加权正态分布的改进算法NAEA及其在重油热解模型中的应用

Alopex-Based Evolutionary Algorithm Combining Weighted Normal Distribution—NAEA and its Application on Parameter Estimation of Heavy Oil Thermal Cracking Three Lumps Model

提出了一种AEA(Alopex-based evolutionary algorithm)的改进算法NAEA。该算法从AEA迭代过程的每一代种群中选择若干适应度值较高的个体,并假设被选中的个体服从正态分布。根据该分布重新采样产生新个体,选择最好的若干新个体替换原种群同等数量的最差个体。在计算正态分布的均值时,采用个体适应度值的大小来分配权值,从而使得算法可以更快更准地收敛到函数最优值。22个标准测试函数的优化结果表明:改进后的算法与AEA相比,不仅在寻优能力方面有了很大的提高,并且收敛速度也有了显著的提升。最后,将NAEA应用在重油热解模型的参数估计中,良好的结果也反映了NAEA算法的良好性能和实际应用价值。

An improved AEA—NAEA is proposed. In the NAEA, several individuals with high fitness value chosen from the population were assumed to follow a normal distribution. The worst individuals in original population were replaced with the good individuals sampled from the normal distribution. In this paper, individual fitness value was used to calculate the mean of the normal distribution, so as to make the algorithm faster and more accurately converge to the optimal value. The optimization results tested by 22 benchmark functions show that compared with AEA the NAEA has achieved significant improvement in both optimizing capability and convergence rate. Finally, the NAEA was applied to estimate the parameters of heavy oil thermal cracking three lumps model, which also demonstrates that the NAEA is an algorithm with good performance and practical application value.