Research And Design Of The Fault Testing And Analysis Model In The Heating Pipeline Network

This subject takes leakage positioning in heating pipeline network as the core research content, and studies pipeline leakage detection methods so as to minimize leakage, aiming to improve the water supply enterprises optimal economic and social benefits. The heating pipeline network system with characteristics such as large-scale, widely distributed, and can not be intuitive, as well as not be found the moment the pipeline leakage accident occurred, which brings a very serious impact to urban construction, living and economic construction. Many domestic companies are undertaking related research of technology, but mostly there are large investment with low accuracy, so the theory is still in the research stage, and the practical application is difficult.Based on BP neural network, the analysis model of fault detection in distribution network brought out by domestic scholars breaks the traditional methods rely on manual testing. With calculation of logical reasoning of problems, the model can get conclusion to provides a new idea to the study of the research of the heating pipeline network system. But considering the problems of the BP neural network that it is a slow convergence model with hard training and local optimization in analysis of the pipe network, this paper presents the probability of detection model based on neural network. Probabilistic neural network has many successful applications in other fields such as mechanical fault diagnosis, which is introduced into the field of water supply network troubleshooting. Taking normal and abnormal patterns of pipe network as a type of operation, the probabilistic neural network may diagnose pipe network for state. At the same time, the use of Bayesian theory and the theory of Fisher may optimize model parameters.Then, in order to solve the problems that the aggravation of network errors due to large pipe network scale together with heavier model computational burden, the paper introduces cluster analysis theory into the heating pipeline network system partition. The number of pipe within the region is significantly less than the number of the entire pipe network, which improves the accuracy and efficiency of the neural network model. Meanwhile, in order to obtain operational data line pipe network leakage in the state to ensure that the correct tube Fabric zoning category, the paper uses actual pipe network in related regions as an example to simulate leakage accident, and introduces test methods during the experiment of simulating fire hydrant accident. Thus, it provides a reliable data support to establishes an accurate, reliable model.Finally, based on the above study, we use VC ++ and MATLAB jointly to develope a prototype system of pipeline leakage detection and positioning. The system can effectively and accurately diagnose and locate network faults, which reducing the manual burden.