同步优化失效概率与建设成本的天然气管网布局流程

目前,天然气长输管网布局优化基本上都是首先以最短路径或最低建设成本为优化目标,然后在已确定的优化布局基础上进行失效概率分析,而在管网运行期间由各种失效导致的风险损失却变得越来越不容忽视,因而在管网布局规划阶段就应该考虑失效造成的风险损失。为了在布局规划阶段实现失效概率和建设成本的同步优化,提出了一种新的布局优化流程:①基于已建管网或相似管网的失效概率和土壤成分数据进行Back-Propagation Network(BP)神经网络预测模型的训练和验证;②基于待优化管网初始敷设路径对应的土壤成分数据预测其失效概率,并与待优化管网的建设成本同时归一化处理;③将归一化后的失效概率与建设成本的乘积定义为得分;④利用最小生成树算法得出所需的最优布局。案例分析结果表明:①路径最短的优化布局通常仅能保证建设成本为最低,难以同时实现失效概率和建设成本同步最小;②得分最小的优化布局能实现失效概率最小,同时其建设成本最接近最小建设成本,以此实现失效概率和建设成本的同步优化;③管网的优化布局方案不应简单只考虑单一因素,应根据失效概率和建设成本的权重关系确定最优的布局方案,以实现经济效益最大化。结论认为,该研究成果可同步优化失效概率和建设成本,具有较大的推广应用价值。

A layout process of natural gas pipeline network with the simultaneous optimization of failure probability and construction cost

At present, the layout optimization of natural gas pipeline network is firstly to reach the objective of the shortest path or the minimum construction cost and then to analyze the failure probability based on the determined optimized layout. However, the risk losses caused by various failures during the pipeline operation are more and more important, so it is necessary to take the risk losses caused by failures into account in the stage of pipeline layout planning. In this paper, a new gas pipeline layout optimization process was proposed in order to realize the simultaneous optimization of failure probability and construction cost in the stage of layout planning. First, the Back-Propagation Network (BP) neural network prediction model is trained and verified based on the failure probability and soil composition data of the constructed pipe networks or similar pipeline networks. Second, the failure probability is predicted based on the soil composition data corresponding to the initial laying path of the pipeline network to be optimized, and then the failure probability and the construction cost of the pipeline network to be optimized are normalized simultaneously. Third, the product of the normalized failure probability and construction cost is defined as the score. Finally, the optimal layout is obtained by means of the minimum spanning tree algorithm. The following results were obtained based on case studies. First, in the optimized layout with the shortest path, only the construction cost can be minimized, and the simultaneous minimization of failure probability and construction cost can be hardly realized. Second, in the optimized layout with the smallest score, the failure probability can be minimized while the construction cost is nearly equal to the minimum value. In this way, failure probability and construction cost are optimized simultaneously. Third, the pipeline network layout optimization scheme shall not simply consider a single factor, and the optimal one shall be determined according to the weight relationship between the failure probability and the construction cost, so as to maximize the economic benefits. In conclusion, the research results can realize the simultaneous optimization of failure probability and construction cost and they are of a greater popularization value.