基于模拟退火与Hopfield神经网络的输气管优化运行  

Optimal operation of gas pipeline based on simulated annealing and continuous Hopfield neural network optimization algorithm

作　　者：段善宁[1] 周昊[2] 汪玉春[3] 

机构地区：[1]中石化西南油气分公司川西采气厂 [2]常州大学江苏省油气储运技术重点实验室 [3]西南石油大学

出　　处：《内蒙古石油化工》2013年第8期69-72,共4页Inner Mongolia Petrochemical Industry

摘　　要：针对输气管优化运行属于组合优化的问题,建立了天然气气源输量变化下的输气管优化运行模型。同时,为了解决连续型Hopfield神经网络的鲁棒性较差和容易陷入局部最优解的问题,引入了模拟退火与神经网络相结合的混合优化算法求解输气管优化运行模型。优化算例表明:该方法改进了标准连续性神经网络算法的收敛过程,能有效防止搜索陷入局部最优解和避免对于初始迭代值的过度依赖,且优化结果优于标准连续性神经网络算法的计算结果,具有更高的优化效率和更强的鲁棒性,能够获得高性能的优化运行方案。Aiming at the fact that optimal operation of gas pipeline is essentially a combination optimization problem, the optimal operation model of gas pipeline is established under the circumstance that the throughput of gas source varies. In the same time the computing method combing simulated annealing algorithm with continuous Hopfield neural network algorithm is applied to solve the optimal operation model of gas pipeline in order to solve the problem that continuous Hopfield neural network has robustness instability and easily falls into the situation of locally optimal solution. The example indicates that this method improves the standard continuous Hopfield neural network convergence process, can eIfeetively avoid searching for the locally optimal solution and relying on the initial iteration value overly. Furthermore the optimization results from this method are better than those from standard continuous Hopfield neural network optimization algorithm; it possesses upper optimum efficiency and stronger

关 键 词：输气管 优化运行 模拟退火 神经网络 算法 

分 类 号：TE832[石油与天然气工程—油气储运工程]