Fuzzy Neural Network Modeling And Pipeline Monitoring System Design

Many chemical processes are multivariable and nonlinear, such as a continuous stirred tank reactor (CSTR). In order to improve the control performance of CSTR, advanced control techniques shall be employed, while the first thing is to model the dynamic process. The pH-mixture tank is a serious nonlinear system, so that it is difficult to model the system with general neural networks. In this paper, a Compensate Fuzzy Neural Network (CFNN) for modeling the pH-mixture tank dynamic system is studied. Compared with those conventional neural networks, the Compensate Fuzzy Neural Network has advantages in training step, training time and precision. Moreover its convergency is faster and more stable.Oil is one of the most important fuels and often transported by pipeline. With the time lapse, aging and eroding of pipeline often result in perforation. That will not only lose a lot of oils but also pollute the environment, which cause serious economic losses.In this paper, a monitoring system for oil pipeline transportation is developed. The paper realizes the pipelines leakage detection system combining correlation analysis to process the pressure wave signal with a fast differential algorithm to capture the signal features. Furthermore, the paper introduces the method of using wavelet algorithm and a fast differential algorithm. The system is demonstrated to be available to detect and location leakage for oil-gas pipelines, and the two methods are simple and easy to be utilized.