| How to locate the pipeline disease point at the fastest speed and obtain the information on the development degree of the underground pipeline leakage disease and the disease location is the focus of the emergency treatment of the municipal pipeline today,and it is also a hot research issue in the municipal planning and construction.Compared with traditional detection methods,ground penetrating radar technology has the characteristics of high detection efficiency,high spatial resolution,wide range of detection targets,non-destructive detection,strong anti-interference ability and wide application environment,and has advantages in the detection of hidden underground conditions.However,with the continuous development and utilization of underground space today,the environment of underground pipelines is becoming increasingly complex,and the types and diameters of pipelines are gradually diversified.While the limitations of traditional detection methods have become increasingly prominent,they have also brought new problems and challenges to interpretation technology.At present,ground penetrating radar technology still has problems such as manual interpretation of radar echo images,difficulty in identifying the degree of pipeline disease,low accuracy in pipeline disease recognition,strong underground clutter interference,and difficulty in identifying disease features in complex environments.In response to the above problems,this paper uses forward simulation and on-site detection of diseases of underground pipelines on campus to study the characteristics of radar echoes of different diseases of common materials of underground pipelines and different disease locations.Analyze the reflection and scattering phenomena in the pipeline disease echo propagation process under complex conditions,summarize the disease echo characteristics,and combine different detection algorithms to detect and select the pipeline disease echo characteristics.Finally,the selected pipeline disease features are applied to the support vector machine model to identify the radar echo of the pipeline disease,and the accuracy of the recognition of the SVM model is evaluated by the accuracy of the pipeline disease recognition.The main findings of the full text are as follows:(1)Considering the detection depth of the underground pipeline radar and the resolution of the echo image,the center frequency of the 500MHz antenna is selected through experiments to be the best,and its resolution and reflected energy are balanced.Experiments show that the echo of a healthy PVC pipe presents a complete hyperbolic shape,and the continuous echo envelope is clear.The echo envelope of healthy concrete pipes is short and incomplete with two leaves,and the continuous echo envelope is discontinuous.The echo hyperbola of the healthy metal steel pipe is clear and complete,the echo envelope is eye-catching and fine,and there is no continuous echo envelope.(2)For different disease levels of pipelines,in the initial stage of disease,the non-metallic tube echo hyperbola has an incomplete shape and the envelope is short and fuzzy,and the metal tube echo hyperbola has a complete shape and a fine and complete envelope.In the developmental stage of the disease,the echo opening of the non-metallic tube becomes larger and the envelope is complete,and the two leaves of the echo envelope of the metal steel tube become thicker.During the disease formation period,the echo envelope shape is relatively complete,the continuous echo envelope is clear,the echo envelope of the concrete pipe becomes significantly longer,the hyperbolic opening of the metal pipe becomes larger,and the envelope is clear and complete.(3)For different locations of pipeline diseases,at the bottom,the echo envelope of non-metallic pipe is complete,the opening is large,the top of continuous echo overlaps indistinctly,the echo envelope of metal pipe is fine and complete,and the envelope of single diffraction wave is incomplete.In the top position,the opening of echo envelope of non-metallic pipe is small,the continuous echo presents a clear X-shape,and the X-shape echo characteristics of metallic pipe overlap clearly.In the lateral position,the opening of hyperbolic echo of non-metallic tube is large,the envelope is clear,the continuous echo shape is incomplete,the echo envelope of metallic tube is fine,the right half of diffraction echo envelope overlaps indistinctly,and the left half is continuous.(4)Through the forward modeling of common material small diameter pipelines,large diameter pipelines,and pipeline diseases under the rough layered medium with the characteristics of real underground random distribution,the difference in imaging characteristics of ground penetrating radar echoes of pipeline diseases under complex conditions is studied,and corresponding diseases are established Feature data set.(5)Experiments show that the use of the gray gradient edge detection algorithm to extract the edge of the radar echo image feature is better,the echo feature of the pipeline disease is clearly distinguished,and the noise is reduced.The electromagnetic wave spectrum time-frequency domain S-transform method has high quality of distinguishing the radar echo characteristics of pipeline diseases,and the noise filtering is significant.The detection and identification of pipeline diseases based on the support vector machine(SVM)method shows that due to the complexity of the actual underground environment,the target area of interest is covered by the noise signal,and the recognition of the characteristic area is missing and enlarged to a certain extent.Conventional pipelines and large-diameter pipelines have high detection and recognition rates,while small-diameter pipelines and pipelines under layered media have low detection and recognition rates.In general,the recognition results of pipeline disease echo characteristics are highly consistent with the actual situation,and the overall pipeline disease detection and recognition rate is 81%,with high-efficiency recognition rate and accuracy. |
PDF Full Text Request