模糊小波神经网络在现役油田玻璃钢管的运行风险评价

Risk Assessment of In-service FRP Pipe Security in Oilfield Based on Fuzzy Wavelet Neural Network

玻璃钢管受到环境和腐蚀等各种因素的影响造成各种管道事故.将模糊评价法、小波分析和人工神经网络理论有机结合,建立了基于模糊小波神经网络的在役玻璃钢管运行风险评价模型.该模型采用模糊评价法对风险因素的指标进行量化,将模糊系统的输出作为神经网络的输入,构造模糊小波神经网络并加以训练.通过选择合适的算法,设计相应的小波神经网络,进行实例仿真.仿真结果表明,该模型能够有效解决现有评价方法主观随意性大、结论模糊等不足,并实时动态再现专家评价结果,其最大相对误差不超过0.3779%.

Some pipeline accidents occurred in oilfield because fiberglass reinforced plastics pipes subjected to the influence of some factors like environment and corrosion, etc. To cope with this problem,a model of risk assessment of pipe security based on fuzzy wavelet neural network （FWNN） was established. The index of fiberglass reinforced plastics pipeline security risk factors was quantized by fuzzy evaluation method. The outputs of fuzzy system were taken as the inputs of ANN. The fuzzy wavelet neural was established and trained. The simulation was made after choosing suitable training algorithm and setting up the wavelet neural network. The simulation result shows that this method can overcome the insufficiencies of arbitrariness, complicated algorithm and fuzzy conclusions in existing traditional evaluation methods, which realizes real-time dynamic reproduction the expert evaluation results,and its maximum relative error is less than 0. 3779 %.