中文摘要：

针对混合蛙跳算法的寻优机制在寻优过程中易陷入局部最优和收敛效果不理想的问题,该文提出一种改进的混合蛙跳算法。该算法在更新群中最差个体时同步更新最优个体。更新最差个体步长时引入上一次的移动步长并赋予动态权值。改进算法舍弃了原算法中用随机值代替最差值的做法,引入高斯变异算子对最差个体进行高斯变异,使种群进化更趋合理。将改进的混合蛙跳算法运用到模糊C均值聚类算法的聚类中心优化中,得到最优的聚类中心。利用该聚类中心对样本进行模糊C均值聚类,并用高斯过程回归对各类样本子集分别建立对应的子模型,通过加权得到系统输出。以双酚A生产过程结晶单元为例进行仿真,对装置出口处的苯酚浓度进行软测量建模,获得了较好的实验结果。

英文摘要：

In order to solve the problem that the optimization mechanism of the shuffled frog leaping adgorithm （ SFLA） is easily falling into the local optimum during the optimization process and the convergence result is unsatisfactory, an improved shuffled frog leaping algorithm （ISFLA） is proposed here. The worst individual and the best individual among subgroups are updated simultaneously. The last moving step-length with the dynamic weight is applied to update the worst individual step-length and makes the population evolution more rational after the Gaussian mutation operator is used on the worst individual instead of the original random mutation operator. The optimal clustering result is cal-culated with the application of the ISFLA in the optimization of clustering centers by using the fuzzy C-means clustering algorithm. The clustering centers are optimized by the fuzzy C-means clustering algorithm. In addition, the final result is outputted by weighted Gaussian sub-models towards different categories. A sample of the crystallization unit of a bisphenol-A production is applied to make a sim-ulation, and the soft-sensor model of the phenol concentration is built at the exit device with a good experiment result.