摘要
设计了一种新的混合蚁群算法,该算法以一种新的二进制蚁群算法为基础,混合PBIL算法及遗传算法的交叉操作和变异操作,从而大大提高了种群的多样性及算法的收敛速度,改善了全局最优解的搜索能力。通过函数优化测试表明该算法具有良好的收敛速度和稳定性,同时将该算法应用到裂解炉裂解深度的神经网络软测量建模中,取得了很好的应用效果。
A kind of new hybrid ant colony algorithm was designed. It used a new binary ant colony algorithm as foundation. Combining with PBIL and the crossover operation and the mutation operation of GA,it had a better ability of breaking away from the local minima. Its population polymorphism and speedy convergence rate were improved remarkably. Optimization simulation results based on typical functions show that the hybrid algorithm has the speedy convergence rate and stability. The availability of algorithm in optimizing neural network is proved by applying neural network in the modeling of depth of fragmentation for fragmental furnace.
出处
《化工自动化及仪表》
CAS
2008年第6期9-13,共5页
Control and Instruments in Chemical Industry
基金
国家"863"计划资助项目(2007AA04Z171)
上海市重点学科建设项目资助(B504)
上海市自然科学基金资助项目(06ZR14027)
关键词
蚁群算法
PBIL
神经网络
遗传算法
ant colony algorithm (ACA)
population based incremental learning (PBIL)
neural network
genetic algorithm