人工蜂群优化BP神经网络在输油管道泄漏检测中的应用  

Application of Oil Pipeline Leakage Detection Technology Based on BP Neural Network Optimized by Artificial Bee Colony

作　　者：刘胜楠[1] 边海容 富宽 张弢甲 施宁[1] 郑健峰[1] 

机构地区：[1]中国石油管道科技研究中心油气管道输送安全国家工程实验室,河北廊坊065000 [2]中国石油管道局工程有限公司设计分公司,河北廊坊065000

出　　处：《管道技术与设备》2017年第2期1-4,共4页Pipeline Technique and Equipment

摘　　要：BP神经网络的泛化性能很大程度上取决于模型的参数选择,针对此问题提出基于人工蜂群算法(ABC)的BP神经网络参数优化算法,并将其应用于输油管道的泄漏检测。该方法将BP神经网络的连接权值和阈值一一对应于人工蜂群算法优化问题每个可行解的各维度值,找到问题的最优解,训练出符合精度要求的理想模型。将该方法应用于输油管道泄漏信号的识别,提取管道各类工况下压力信号部分非线性特征作为检测模型训练样本集,建立输油管道泄漏检测模型并进行识别测试,有效提高了泄漏检测的准确率。The generalization performance of BP neural network was largely dependent on the choice of its parameters. A parameter optimization method based on artificial bee colony algorithm （ABC） was proposed to solve this problem and applied to oil pipeline leak detection. In this method, a sort of one-to-one correspondence was established between the weights and thresholds of BP neural network and each dimension of feasible solution of ABC optimization problem. The most optimal solution was found to achieve the ideal model of meeting required precision. The method was applied to leakage signal identification of oil pipeline. The partial nonlinear characteristics of pipeline pressure signals in the various conditions were extracted as the training sample set of detection model. Pipeline leak detection model was established and identification test was marched, improving the leakage detec- tion accuracy rate.

关 键 词：人工蜂群算法 BP神经网络 输油管道 泄漏检测 

分 类 号：TE88[石油与天然气工程—油气储运工程]