基于遗传优化的模糊神经网络在管道泄漏检测中的应用研究

Application Research Based on Genetic Optimization Fuzzy-neural Network in the Detection of the Pipeline Leakage

提出了一种基于遗传优化的模糊神经网络的管道泄漏检测方法。针对BP算法易受初始权值影响陷入局部极小、收敛速度慢的缺点,根据遗传算法具有全局寻优特点,将二者结合起来训练模糊神经网络,进而获得更接近全局最优的网络参数以提高泄漏的估计精度。最后通过对管道泄漏检测的实际数据进行仿真测试,表明该算法可以有效、可靠地运用于管道泄漏检测中。

A kind of pipeline leak detection method based on fuzzy-neural network of genetic optimization is applied. According to the disadvantage of BP algorithm which can be easily affected by the initiative weight value and trapped into a local optimum and low convergence speed as well as the advantage of globe optimal searching of genetic algorithm, combing these two algorithms to train the fuzzy-neural network can get network parameters which are more closer to globe optimization and the leak evaluating precision can be promoted. Through the simulation test of real data been acquired by pipeline leak detection, it suggests that this algorithm can be used in diagnosing the pipeline leakage fault effectively and reliably.