摘要
为了及时准确地检测管道泄漏,提出了一种基于模糊神经网络的管道泄漏检测的方法。首先以管道内流体在出口与入口之间的压力差与流量差信号作为网络输入,以泄漏的尺寸大小作为网络输出,构建了管道泄漏检测与泄漏尺寸估计的模糊神经网络结构,进而选取实际管道数据,对网络的参数进行离线训练,并得出网络权值。最后以借鉴某管道泄漏的部分先验知识建立模糊规则的基础上,通过仿真验明了方法在管道泄漏诊断中的有效性和可行性。
In order to detect pipe - line leakage quickly and accurately, an efficient method is proposed based on fuzzy neural network. First this algorithm uses the pressure difference and flow difference of the flow between the inlet and outlet of the pipeline as the input signal and uses the leakage area size as the network output to build a fuzzy - neural network for diagnosing the pipe - line leakage detection and estimating the leakage size. The real pipeline data are chosen, then training the neural network parameters offline to get network weight. Lastly, based on drawing lessons from partial transcendental knowledge of a certain pipeline leakage to establish fuzzy rules, and throught simulation, the method has proved to be effective and practicable for the pipeline leakage diagnosis.
出处
《计算机仿真》
CSCD
北大核心
2009年第2期190-192,232,共4页
Computer Simulation
基金
教育部"春晖计划"(Z2005-1-62001)
兰州理工大学特色学术梯队基金项目(0950)
关键词
管道
泄漏检测
模糊规则
神经网络
Pipeline
Leak detection
Fuzzy - rule
Neural network