基于T-S模糊神经网络预测算法的管道泄漏定位研究与分析

长输管道的泄漏检测和定位对管道安全平稳运行意义极其重大，在以软件计算为主的检测方法里，模糊神经网络模型综合了模糊算法和神经网络模型的优点，能较好地适应长输管道的非线性特征。本文采用中亚地区某管道某相邻两站场的历史数据训练基于高木-关野（Takagi-Sugeno）模糊神经网络的预测模型，使用STONER管道仿真软件产生实时数据，用一种较简单的软方法较好的实现管道泄漏定位，该种方法对中亚某长输管道这类没有专门硬件泄漏检测设备和系统的管线有一定的实用意义。

Study and Analysis of Gas Pipeline Leak Location Based on T-S Fuzzy Neural Network Prediction Algorithm

Leak detection and location of long-distance pipeline are of significance for safe operation. Among the detection methods based on computing software, the fuzzy neural network model combines the advantages of fuzzy algorithm and neural network model, and can better adapt to long-distance pipeline characterized by nonlinearity. This paper used the historical data from two neighbored gas compressor stations of Central Asia Gas Pipeline to train the model based on Takagi-Sugeno fuzzy neural network forecasting algorithm, and used STONER pipeline simulation software to generate real-time data. Thus, a relatively simple method on soft computing realized leak location of pipeline, which is of some practical significance to the pipelines without special leak detection devices, such as Central Asia Gas Pipeline.