Acoustic Emission Intelligent Detecting Technique And System Development For Internal Valve Leakage

In oil and natural gas industries,valves are extensive and frequent used machinery universal products,which have a variety of types,huge number and different structures and works in highly corrosive places and hostile natural environment.Due to the effect of medium process parameters that include high temperature,pressure and corrosive etc.,there are many internal valve leakage accidents occurred.Internal valve leakage faults destroy the structural integrity of equipment and lead to medium leaking unexpectedly causing fire,explosion,medium pollution and poisoning etc.more serious consequences disastrous accident.Therefore,the developing of acoustic emission detection system with the functions of recognizing internal valve leakage failure modes and predicting internal valve leakage rate can provide support for early warning technology,which has important engineering research significance and applied value.The whole process of converting AE source to AE signals was researched based on the fundamental principles of AE detecting techniques.On the basis of this choosing valves internal leakage as object,the type of internal valve leakage signal was analyzed and the affect factors of valve medium and external environment of valve to internal valve leakage were concluded.Analyzed the relationship between the internal valve leakage rate to physical size of valve and Fluid characteristic parameters,and finding out the relationship model between internal valve leakage rate and AE signal characteristic parameters.Internal valve leakage AE testing experiment system was established,using this experiment system,collecting ball valve,butterfly valve,manual ball valve internal leakage AE signals in eight failure modes under the pressure of 0.1-1.0MPa at the same time recording the internal valve leakage rate.The influence of valve type to AE signal,the correlation between internal valve leakage and AE signal characteristic parameter and the AE signal spectrum signature differences of different valve failure mode were analyzed.In order to get the input data for support vector machine(SVM),filtering and de-noising the original waveform signal as pretreatment of data,recognizing ball valve in good condition,ball valve in tinny opening and ball valve core scratch three fault pattern.Taking ball valve core scratch fault pattern for example,30 sets of acoustic emission signal characteristic parameters were extracted as input data of BP neural network,25 sets data were selected as training data and 5 sets data were used to test the accuracy of the BP neural network to predict the internal valve leakage rate.According to the basic principle of acoustic emission testing system and special requirements of valve leakage detection,the AE sensors,acquisition card and a tablet supported by Windows 7/8 system were selected to complete hardware combination of acoustic emission testing system.The internal valve leakage acoustic emission test system software was developed by Microsoft Foundation Classes(MFC),which realized AE signal acquisition function by calling API functions of acquisition board and internal valve leakage failure modes recognition and internal valve leakage rate accurate inversion functions by calling dynamic link libraries from MATLAB.