Application And Research On Bootstrap And Fuzzy Cluster Method In Pipeline Leak Detection System

Long transportation Pipeline has become a major transportation way because of it's high effective, convention, environmental protection. However, because ageing, erode, beening mangled by human, the leakage events is often happen. Those events cause bigness economy losts and environment pollution. Therefore, pipeline leak detection is an important issue in the pipeline transportation industry in the world wide range. Hardware-based leak detection can't operate continuously, the characters of real time,reliability, precision, sensitivity of software-based leak detection are unsatisfactory. During the far-ranging use of SCADA systems in pipeline transportation industry, the complex software-based leak detection based on the information process has become a focus and tidal current and get much attentions by academe. On the base of previous works, following works been done in this paper:First, we need to identify the system sample frequency which can reflect the fundamental structures and features of pipeline small leaks and regulating pumps through analyzing the pipeline small leaks and regulating pumps or valves pressure waveforms in different sample frequency, then normalize the corresponding signals of negative press wave.Second, as for the incomplete information problem occurred in the negative pressure wave signal of the process of pipeline leak. Using the Bootstrap method of the small sample statistics, sampling simulation and statistical analysis the observation sample to provide a number of valid data samples for the distinguish between small leaks and pressure waveforms induced by normal work conditions. Then, build a standard sample base in different work conditions.Third, extract some reliable indexes and parameters which can reflect the feature of sample from a standard sample base in different work conditions as index vectors, then, being the input of fuzzy c-means clustering. Fuzzy cluster can be used for iteration method to the extracted features and continuous modification clustering center until it was convergence according to the change of features. Once acquired the fuzzy clustering center in different work conditions, we can identify the operating modes of pipeline leaks and normal work conditions and decrease the system false alarm rate at the pipeline small leaks. The simulation resultsdemonstrated that this method is simple, practical and feasible in the application of pipe leakage detection system.At last, based on the pre-phase developed software of leakage detection and location of pipeline, we proposed a preliminary plan of realization of fuzzy cluster software by using the important design thoughts and experience parameters provided by the simulate experiment of fuzzy cluster.