免疫文化算法及其在乙烯裂解炉故障诊断中的应用

Immune cultural algorithm and its applications in fault diagnosis of ethylene cracking furnace

引言迄今为止,科研人员已经根据生物进化的机理提出很多用以解决复杂优化问题的方法,如遗传算法、蚁群优化算法、粒子群优化算法等。然而这些传统的进化算法只提供有限的或者隐性关于种群个体经验的知识表示和保存机制,这就让研究人员开始寻找一种利用显性机制来获取并保存种群进化求解知识和经验。在人类学的角度上,

A new immune cultural algorithm（ICA）based on immune clone selection was proposed.In ICA,immune clone machine was used for training and testing sampling data from SRT-Ⅲ furnace.One selection was taken as population space of cultural algorithm.In belief space,the knowledge extraction,expression,storage,update methods were proposed according to their evolutionary characteristics.Communication function was improved at the same time which in turn improved the capacity of algorithm evolution.The test results showed that compared with genetic algorithm（GA）and chemotactic differential evolution algorithm（CDEA）,immune cultural algorithm had much improvement in search precision and convergence speed.The algorithm was applied to the support vector machine parameter optimization for solving fault diagnosis of ethylene cracking furnace.Multi-class classifier was made by support vector machine.Compared with fault classification using the parameters optimized by genetic algorithm,the simulation results showed that the proposed algorithm achieved good result in classification accuracy,20 percentage points higher than the method without using immune cultural algorithms.