Research On Pressure Pipeline Leakage Detection And Warning Based On CFD-LSTM-BP

In chemical plants,pressure process piping as an important transport equipment is subject to fluid washout and corrosion for a long time,especially at high temperature and pressure,even if a small leak occurs,the leak hole will expand rapidly,resulting in the pipeline,equipment and plant then enter a high-risk state or even cause an explosion and other accidents,resulting in immeasurable economic losses and damage to the environment.There is an urgent need for a technology to quickly detect and locate leaks in pressurized process pipelines in open spaces,so that measures can be taken in time for small leaks.The current leak detection method used in chemical plants is the concentration detection alarm,whose detection effect is greatly affected by the wind speed and direction in the flow field,as well as the installation location,while the acoustic method as an effective method of pipeline leak detection is commonly used in buried long-distance pipelines,there is no research on the application of process pipelines in chemical plants.In view of this,this study uses the sound source generated by pressure pipeline leakage as the signal source,and adopts the idea of acoustic wave method for pressure pipeline leakage detection and location,based on 3D modeling software spaceclaim combined with fluid dynamics calculation software(CFD)simulation to establish the pipeline leakage flow field in the turbulent leakage basin of pressure pipeline,to explore the type of sound source,sound field characteristics and acoustic wave attenuation law of its generation of aerodynamic noise,and The acoustic model based on the leakage flow field and the characteristics of leakage acoustic wave is established,and the model for pipeline leakage detection and localization in open space is established by LSTM-BP neural network to realize the rapid output of the pressure of the leaking pipeline and the distance between the leakage point and the detection point,so as to accurately determine the location of the leakage.The results show that the CFD-LSTM-BP based pressure pipeline leakage early warning model is an effective and highly sensitive model for locating and detecting the leakage of toxic and hazardous gases in pipelines and timely warning,and this early warning technology has important scientific and theoretical significance.Based on the CFD software Fluent,we simulated the leakage flow field of pipeline and obtained the characteristics of jet velocity,flow line and diffusion material distribution in the leakage gas flow field under different working conditions.Using a grid partitioning scheme with a grid number of 1.7 million,we simulated the characteristics of jet velocity,streamline and diffusion material distribution in the flow field of leaking gas under different working conditions,and obtained the following behaviors of external meteorological conditions,leakage aperture,pipe pressure and other conditions on the leakage flow field: 1)The wind direction and wind speed in the flow field basically have no effect on the velocity and form of the leaking jet,but have a greater effect on the diffusion of leaking material larger.In the absence of wind or downwind(wind speed less than 2m/s),the concentration distribution of leaking substances outside the jet is too low or does not reach the lower limit of the detection concentration measurement area,there will be existing toxic or combustible gas concentration detection alarm instrument can not effectively detect the leak;2)when the pressure of the leaking pipe is less than or equal to 11 MPa,the concentration distribution of leaking substances in the flow field does not change much,and can be detected in the simulation area;3)when the initial velocity and range of the leaky jet in the leaky flow field increase with the increase in pipe pressure,the leaky aperture is positively correlated with the range of the sound source and negatively correlated with the initial velocity of the jet,i.e.,negatively correlated with the intensity of the generated pneumatic noise source,and the leaky point in the Different tube shapes produce different jet velocities.Combined with the generation of pneumatic noise,leak source type and Fourier transform theory,the acoustic signal characteristics of different operating conditions are extracted to study the impact of environmental and other factors on the acoustic field results are as follows: 1)In the flow field of pressure pipe leaks almost does not exist a clear main frequency,but the acoustic signal receiver arranged in the leak point outside 1m can collect the acoustic signal of the leak.The attenuation of acoustic waves is mainly concentrated in the area close to the source(5-100mm),in the study of the more distant area(100-1000mm)in the acoustic attenuation becomes slow.2)For the same conditions at the receiving point of the sound pressure level spectrogram curve presents the trend and waveform consistency,can be judged by the spectrogram under different operating conditions of the pipeline leakage.Leakage environment wind speed and wind direction on the intensity of the leak source and the impact of acoustic attenuation is mainly concentrated in the 100 mm range of the leak,the attenuation value can reach about 50%,in 100 mm outside the attenuation of its sound waves moderate a lot,in the simulation area at any point can still receive the leak signal.3)The pressure of the pipe and its leak aperture have a certain impact on the intensity of the sound source,the pressure of the leak pipe and the leak source of sound The pressure of the leaking pipe is positively correlated with the intensity of the leaking sound source,and the pressure of the pipe can reach 102 d B sound pressure level when there is a leak at 3 MPa,and the sound pressure level of the sound source increases when the pressure increases.Leakage aperture diameter and the intensity of the sound source is negatively correlated,the sound pressure level at the source of the leak is also 80 d B when the aperture diameter is 8mm,which again proves that the detection of pipe leakage in open space by acoustic method can be detected in this simulation calculation domain.The study is based on the deep learning neural network model established by LSTM-BP,through the LSTM neural network on the numerical simulation results of the acoustic field to build the acoustic signal learning in the case of leakage and non-leakage in the data set,the pipeline leakage detection early warning model was constructed,and the accuracy of the data test reached more than95%.Then after the BP network-based model can quickly output the leaky pipeline pressure and the distance between the leak point and the detection point,the leak point is accurately located,and its accuracy rate reaches 99.3%.After operation,the leak point localization model based on LSTM-BP deep learning neural network shows good generalization ability and has the value of further research and promotion.