Underground pipeline detection and identification method and system based on ground penetrating radar and neural network

Abstract

The invention discloses an underground pipeline detection and identification method and system based on a ground penetrating radar and a neural network, and the method comprises the steps: firstly setting a transmitting antenna and a receiving antenna of the ground penetrating radar, transmitting a radar signal through the transmitting antenna, receiving an echo signal during the detection of the ground penetrating radar through the receiving antenna, and generating an underground pipeline target detection map P1; then, a neural network is utilized to obtain a high-resolution underground pipeline target detection map; the front-end conditioning circuit designed by the invention can amplify or attenuate a signal to a proper range according to requirements. For high-frequency echo signals, a high-speed data acquisition system is designed, and a high-speed data acquisition circuit is designed on the basis of an FPGA (Field Programmable Gate Array) and two domestic high-speed ADCs (Analog to Digital Converters). A neural network framework is further designed, efficient and rapid multi-scale feature integration can be achieved, high-performance underground pipeline contour extraction is achieved, and the problem of detection accuracy in the prior art is effectively solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of underground pipeline detection, and relates to an underground pipeline detection and identification method, system and product, and in particular to an underground pipeline detection and identification method and system based on ground penetrating radar and neural network. Background Art

[0002] Underground pipelines are the "blood vessels" of the city and the "lifeline" to ensure the normal operation of the city. With the rapid advancement of urbanization, the number and scale of urban underground pipelines are increasing, and the structure is becoming more and more complex. During the construction process, due to the unclear distribution of underground pipelines, it brings many difficulties to urban construction and renovation as well as the use and maintenance of pipelines.

[0003] Ground Penetrating Radar (GPR) is an electromagnetic technology for locating invisible targets or interfaces underground or inside objects. Ground penetrating radar has become a powerful tool for geophysical exploration due to its high resolution and high efficiency. With the continuous development of signal processing technology and electronic technology, the increase in engineering practice and the accumulation of rich experience, ground penetrating radar technology has further developed, instruments have been continuously updated, and the scope of application has gradually expanded. It has now been widely used in construction quality control, pavement structure and material research, measurement equipment application technology research, design data survey and collection, disease and hidden danger detection, monitoring and arbitration of engineering quality accident causes, evaluation and reconstruction of old roads, road structure layer thickness detection, underground pipeline detection and other fields.

[0004] Ground penetrating radar uses electromagnetic waves as information carriers. Since electromagnetic waves decay quickly during propagation, the detection accuracy of underground pipelines is related to the center frequency of electromagnetic waves. The higher the center frequency of electromagnetic waves, the higher the detection accuracy, but the faster the decay. It is suitable for shallow high-precision detection, and vice versa for deep detection. Therefore, a high-power high-frequency signal is usually generated, which is transmitted underground through the antenna, and then the echo signal is received by the antenna for analysis and processing to achieve the detection of underground pipeline conditions.

[0005] Due to the working characteristics of ground penetrating radar, various clutter and noise will be received during the actual detection of underground pipelines, including electromagnetic interference caused by mobile phones, wires, etc., resulting in poor accuracy of the images obtained by current detection. Therefore, it is an urgent problem to study reliable ground penetrating radar echo signal processing methods and image processing methods. Summary of the invention

[0006] In order to solve the technical problem that the existing ground penetrating radar cannot effectively filter the electromagnetic interference in the echo signal during signal acquisition, resulting in poor image accuracy, the present invention provides an underground pipeline detection and identification method and system based on ground penetrating radar and neural network.

[0007] The technical solution adopted by the method of the present invention is: a method for detecting and identifying underground pipelines based on ground penetrating radar and neural network, comprising the following steps:

[0008] Step 1: Set up the transmitting antenna and receiving antenna of the ground penetrating radar, transmit the radar signal through the transmitting antenna, receive the echo signal during ground penetrating radar detection through the receiving antenna, and generate the underground pipeline target detection map P 1 ;

[0009] Step 2: The underground pipeline target detection map P 1 Input the neural network to obtain a high-resolution underground pipeline target detection map;

[0010] The neural network consists of a feature extraction subnetwork and a feature fusion subnetwork;

[0011] The feature extraction subnetwork is composed of a first downsampling layer, a second downsampling layer and a third downsampling layer connected in sequence, and is used to convert the input P 1 The hierarchical image features of different sizes are obtained by downsampling in sequence P = {P 2 , P 3 , P 4}; the P 4 is a low-resolution coarse detection image, denoted as P′ 4 The feature fusion subnetwork is composed of a first upsampling operation layer, a first fusion layer, a first feature fusion decoder, a second upsampling operation layer, a second fusion layer, a second feature fusion decoder, a third upsampling operation layer, a third fusion layer and a third feature fusion decoder, and is used to convert P′ 4 After the upsampling process of the first upsampling operation layer, the P′ 4 , P 3 After adding pixel by pixel through the first fusion layer, it is input into the first feature fusion decoder for processing and outputs P′ 3 ; Then P′ 3 After the upsampling process through the second upsampling operation layer, the 4 , P′ 3 , P 2 After pixel-by-pixel addition through the second fusion layer, it is input into the second feature fusion decoder for processing, and the output is P′ 2 ; Finally, P′ 2 After the upsampling process through the third upsampling operation layer, the 4 , P′ 3 , P′ 2 , P 1 After pixel-by-pixel addition through the third fusion layer, it is input into the third feature fusion decoder for processing to obtain a high-resolution underground pipeline target detection map.

[0012] Preferably, the feature fusion decoder is composed of two branches arranged in parallel; the first branch is composed of a dimensionality reduction convolution layer, an attention convolution layer, and a classifier connected in sequence; the second branch is composed of a dimensionality reduction convolution layer, an attention convolution layer, an upsampling layer, and a classifier connected smoothly; wherein the output of the attention convolution layer of the first branch is added pixel by pixel to the upsampling layer of the second branch and then input into the classifier of the first branch; the high-resolution feature map is processed by the first branch to output a high-resolution detail map, and the low-resolution feature map is processed by the second branch to output a low-resolution detail map; the high-resolution detail map is added pixel by pixel to the low-resolution detail map to output a high-resolution prediction map.

[0013] The technical solution adopted by the system of the present invention is: an underground pipeline detection and identification system based on ground penetrating radar and neural network, which is composed of a ground penetrating radar, an echo signal acquisition processor and a host computer; the ground penetrating radar is composed of a main control unit, a transmitter, a transmitting antenna, a receiver, and a receiving antenna; the echo signal acquisition processor is composed of an echo conditioning circuit, an ADC module, an FPGA module and an STM32 module connected in sequence; the STM32 module is connected to the host computer for communication;

[0014] The echo conditioning circuit is used to transform, program-control and amplify the echo signal received by the ground-penetrating radar receiver; the ADC module is used to perform analog-to-digital conversion on the conditioned echo signal to obtain a parallel data stream; the FPGA module is used to control the analog-to-digital conversion process of the ADC module, and to receive the parallel data stream and send it to the STM32 module after data processing; the STM32 module processes the received data stream and inputs it into the host computer, and sends a reverse control gain instruction to the FPGA module, controls the gain of the echo conditioning circuit through the reverse control gain instruction, and realizes program-controllable amplification or attenuation of the echo signal.

[0015] Preferably, the echo conditioning circuit comprises an RF transformer, an RF amplifier with variable gain and a fully differential amplifier connected in sequence; the RF transformer is used to transform the echo signal received, and the RF amplifier with variable gain is used to perform program-controlled amplification or program-controlled attenuation on the transformed echo signal before inputting it into the fully differential amplifier.

[0016] Preferably, the ADC module uses two analog-to-digital converters connected in parallel, and first samples the echo signal output by the echo conditioning circuit through the DES bilateral sampling mode in combination with the bilateral edges of the sampling clock provided by the FPGA module to obtain a parallel high-speed data stream; and then sends the parallel high-speed data stream to the FPGA module through the LVDS low-voltage differential transmission method.

[0017] Preferably, the FPGA module obtains two sampling clocks with a phase difference of 90° through PLL phase-locked loop frequency multiplication and phase shift processing; the FPGA module inputs sampling clocks with the same frequency and different phases into the ADC module through a high-speed data acquisition unit.

[0018] Preferably, the FPGA module comprises a high-speed data acquisition unit, a data cache unit and a data superposition unit connected in sequence; the data superposition unit is connected to the STM32 module; the FPGA module is also provided with a control gain unit, which is respectively connected to the STM32 module and the echo conditioning circuit, and is used to adjust the gain of the echo conditioning circuit;

[0019] The data buffer unit is used to cache the received parallel high-speed data streams, and re-join the parallel high-speed data streams into serial data, and input them into the data superposition unit for data processing; the data superposition unit corrects and de-noises the joined high-speed data streams and then sends them to the STM32 module.

[0020] Preferably, the data superposition unit is used to perform zero-time correction and superposition noise reduction processing on the spliced ​​high-speed data stream.

[0021] Preferably, the zero-time correction is performed by fixing the relative position between the transmitting antenna and the receiving antenna and using the time when the direct wave arrives as a reference to correct the time difference.

[0022] Preferably, the superposition noise reduction process uses an average superposition noise reduction algorithm to process the high-speed data stream.

[0023] Compared with the prior art, the beneficial effects of the present invention include:

[0024] (1) The present invention designs a densely connected coarse-to-fine feature fusion decoder architecture to achieve efficient and fast multi-scale feature integration and output high-resolution underground pipeline detection maps.

[0025] (2) The present invention designs an echo conditioning circuit to perform front-end conditioning on the echo signal, and then uses FPGA and two ADCs to achieve 3.6 GSPS high-speed data acquisition, with a fast acquisition speed.

[0026] (3) In order to reduce the influence of noise, the present invention designs real-time superposition average noise reduction. At the same time, in view of the jitter of the signal source, the present invention performs zero-time correction processing on the collected echo signal. When the ground penetrating radar is working, the FPGA will perform real-time high-speed superposition average filtering on the received echo signal data. After reaching a certain number of times, the filtered data will be sent to the host computer, and the detection accuracy is high.

[0027] (4) The present invention is well suited for use in various complex and changeable ground exploration environments and is highly practical. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The technical solution of the present invention is further described below using embodiments and specific implementation methods. In addition, in the process of describing the technical solution, some drawings are also used. For those skilled in the art, other drawings and the intention of the present invention can also be obtained based on these drawings without paying creative work.

[0029] Figure 1 A schematic diagram of a neural network structure according to an embodiment of the present invention;

[0030] Figure 2 This is a structural diagram of a feature fusion decoder according to an embodiment of the present invention;

[0031] Figure 3 is a system principle diagram of an embodiment of the present invention;

[0032] Figure 4 It is a structural diagram of an echo conditioning circuit according to an embodiment of the present invention;

[0033] Figure 5 This is a schematic diagram of an ADC module according to an embodiment of the present invention;

[0034] Figure 6 Schematic diagram of experimental results of an embodiment of the present invention, wherein (a) is the original radar waveform output by the ground penetrating radar, and (b) is a high-resolution underground pipeline target detection image output by the neural network. DETAILED DESCRIPTION

[0035] In order to facilitate ordinary technicians in the field to understand and implement the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the implementation examples described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0036] Please see Figure 1 The present embodiment provides a method for detecting and identifying underground pipelines based on ground penetrating radar and neural network, comprising the following steps:

[0037] Step 1: Set up the transmitting antenna and receiving antenna of the ground penetrating radar, transmit the radar signal through the transmitting antenna, receive the echo signal during ground penetrating radar detection through the receiving antenna, and generate the underground pipeline target detection map P 1 ;

[0038] Step 2: The underground pipeline target detection map P 1 Input the neural network to obtain a high-resolution underground pipeline target detection map;

[0039] The neural network consists of a feature extraction subnetwork and a feature fusion subnetwork;

[0040] The feature extraction subnetwork is composed of a first downsampling layer, a second downsampling layer and a third downsampling layer connected in sequence, and is used to convert the input P 1 The hierarchical image features of different sizes are obtained by downsampling in sequence P = {P 2 , P 3 , P 4}; the P 4 is a low-resolution coarse detection image, denoted as P′ 4 The feature fusion subnetwork is composed of a first upsampling operation layer, a first fusion layer, a first feature fusion decoder, a second upsampling operation layer, a second fusion layer, a second feature fusion decoder, a third upsampling operation layer, a third fusion layer and a third feature fusion decoder, and is used to convert P′ 4 After the upsampling process of the first upsampling operation layer, the P′ 4 , P 3 After adding pixel by pixel through the first fusion layer, it is input into the first feature fusion decoder for processing and outputs P′ 3 ; Then P′ 3 After the upsampling process through the second upsampling operation layer, the 4 , P′ 3 , P 2 After pixel-by-pixel addition through the second fusion layer, it is input into the second feature fusion decoder for processing, and the output is P′ 2 ; Finally, P′ 2 After the upsampling process through the third upsampling operation layer, the 4 , P′ 3 , P′ 2 , P 1 After pixel-by-pixel addition through the third fusion layer, it is input into the third feature fusion decoder for processing to obtain a high-resolution underground pipeline target detection map.

[0041] In this embodiment, the input feature P is considered i In different number of feature channels, first for each input feature P i Applying a dimensionality reduction convolutional layer Compress the number of feature channels c to the pre-screened size, represents a convolutional layer with a kernel size of k×k and an output channel of c. After that, the reduced-dimensional features are input into the attention convolutional layer to generate an attention map in the spatial dimension, suppressing background information, highlighting the bottom line pipeline area, and improving model performance.

[0042] Please see Figure 2In one embodiment, the feature fusion decoder is composed of two branches arranged in parallel; the first branch is composed of a dimensionality reduction convolution layer, an attention convolution layer, and a classifier connected in sequence; the second branch is composed of a dimensionality reduction convolution layer, an attention convolution layer, an upsampling layer, and a classifier connected smoothly; wherein the output of the attention convolution layer of the first branch is added pixel by pixel to the upsampling layer of the second branch and then input into the classifier of the first branch; the high-resolution feature map is processed by the first branch to output a high-resolution detail map, and the low-resolution feature map is processed by the second branch to output a low-resolution detail map; the high-resolution detail map is added pixel by pixel to the low-resolution detail map to output a high-resolution prediction map.

[0043] The dimension reduction convolution layer described in this embodiment is composed of a 1×1 convolution layer, a batch normalization layer, a ReLU activation layer, a 3×3 convolution layer, a batch normalization layer, and a ReLU activation layer connected in sequence. The attention convolution layer is composed of a mean pooling layer, a first 3×3 convolution layer, a Sigmoid activation layer, a second 3×3 convolution layer, a batch normalization layer, and a ReLU activation layer connected in sequence; wherein the original input is multiplied pixel by pixel with the Sigmoid activation layer and then input into the second 3×3 convolution layer for subsequent processing.

[0044] The neural network used in this embodiment is a pre-trained network, and the training scheme adopted is the existing technical scheme, which will not be further described here.

[0045] Please see Figure 3 , this embodiment also provides an underground pipeline detection and identification system based on ground penetrating radar and neural network, including a ground penetrating radar, an echo signal acquisition processor and a host computer; the ground penetrating radar is composed of a main control unit, a transmitter, a transmitting antenna, a receiver, and a receiving antenna; the echo signal acquisition processor is composed of an echo conditioning circuit, an ADC module, an FPGA module and an STM32 module connected in sequence; the STM32 module is connected to the host computer for communication;

[0046] The echo conditioning circuit is used to transform, program-control and amplify the echo signal received by the ground-penetrating radar receiver; the ADC module is used to perform analog-to-digital conversion on the conditioned echo signal to obtain a parallel data stream; the FPGA module is used to control the analog-to-digital conversion process of the ADC module, and to receive the parallel data stream and send it to the STM32 module after data processing; the STM32 module processes the received data stream and inputs it into the host computer, and sends a reverse control gain instruction to the FPGA module, controls the gain of the echo conditioning circuit through the reverse control gain instruction, and realizes program-controllable amplification or attenuation of the echo signal.

[0047] Please see Figure 4In one embodiment, the echo conditioning circuit includes an RF transformer, a variable gain RF amplifier, and a fully differential amplifier connected in sequence; the RF transformer is used to transform the received echo signal, and the variable gain RF amplifier is used to programmatically amplify or programmatically attenuate the transformed echo signal before inputting it into the fully differential amplifier. The echo conditioning circuit first passes the echo signal through an RF transformer to complete the electrical isolation of the signal. Afterwards, due to the limited input amplitude of the high-performance ADC, it is necessary to pass through a variable gain RF amplifier to realize program-controllable amplification or attenuation of the input signal through the FPGA module. Finally, the input signal needs to pass through a fully differential amplifier to offset the interference of the common-mode signal, and then the signal differential is input into the ADC module; in the design of the echo conditioning circuit, stable signal transmission is achieved through impedance matching.

[0048] In one embodiment, the RF transformer can use a RF transformer with model number T3002NL to complete the electrical isolation of the signal; the variable gain amplifier can use a variable gain amplifier with model number ADL5201ACPZ-R7 to perform programmable amplification or attenuation; the fully differential amplifier can use a fully differential amplifier with model number LMH6554 to provide low-distortion amplification for broadband differential signals.

[0049] Please see Figure 5 In one implementation, the ADC module uses two analog-to-digital converters connected in parallel. First, the echo signal output by the echo conditioning circuit is sampled through the DES bilateral sampling mode in combination with the bilateral edges of the sampling clock provided by the FPGA module to obtain a parallel high-speed data stream; then the parallel high-speed data stream is sent to the FPGA module through the LVDS low-voltage differential transmission method.

[0050] In one embodiment, the ADC module can use an ADC of model MXT2001. MXT2001 is a dual-channel, high-performance ADC with a sampling accuracy of 8 bits and a variety of program-controllable acquisition and data-side transmission methods. Under dual-channel working conditions, each channel can reach a maximum of 1.0GSPS. It supports time-interleaved mode sampling, and can reach a maximum of 2.0GSPS under bilateral sampling. The ADC with two channels is integrated on the chip, and the single-chip ADC has four parallel high-speed data streams. Through TIADC technology, the sampling clock is phase-shifted so that the two ADCs are sampled alternately, doubling the real-time sampling rate to 3.6GSPS. The converted data adopts LVDS parallel transmission mode, and with the data synchronization clock, the data is sent to the FPGA module in parallel at a speed of up to 500MHz.

[0051] In one embodiment, the FPGA module includes a high-speed data acquisition unit, a data cache unit and a data superposition unit connected in sequence; the data superposition unit is connected to the STM32 module; the FPGA module is also provided with a control gain unit, which is respectively connected to the STM32 module and the echo conditioning circuit, and is used to adjust the gain of the echo conditioning circuit;

[0052] The data buffer unit is used to cache the received parallel high-speed data streams, re-join the parallel high-speed data streams into serial data, and input them into the data superposition unit for data processing; the data superposition unit corrects and de-noises the joined high-speed data streams and then sends them to the STM32 module.

[0053] The FPGA module obtains two sampling clocks with a phase difference of 90° through PLL phase-locked loop frequency multiplication and phase shift processing; the FPGA module inputs sampling clocks with the same frequency and different phases into the ADC module through the high-speed data acquisition unit set up. The sampling clocks with the same frequency and different phases are used to control multiple ADCs to work together. The sampling clocks are phase-shifted and then distributed to each sub-ADC as the sampling clock of each sub-ADC, so that the same signal is collected at different phases, and the collected data is processed and rearranged to finally restore the original signal.

[0054] In one implementation, the FPGA chip model of the FPGA module may be 10CL055YF484C6G.

[0055] When the system of this embodiment is working, various interferences and noises often occur, and due to the influence of device consistency and temperature on the analog circuit, the delay between the final output of the signal and the trigger signal may fluctuate. In order to reduce the influence of noise interference on detection accuracy and stability, it is necessary to perform zero-time correction and superposition noise reduction processing on the data and send it to the STM32 module, and then transmit it to the host computer. The STM32 module mainly plays the role of function control and data interaction.

[0056] In one implementation, for zero-time correction, the idea is to fix the relative position between the transmitting and receiving antennas and use the arrival time of the direct wave as a reference to correct the time difference. For random noise in the environment, an average superposition noise reduction algorithm is used to highlight the effective reflected waveform information and improve the signal-to-noise ratio.

[0057] In one implementation, zero-time correction and noise reduction can be integrated into a module, which includes a superposition operator, a traversal search module, a data selector, a shift divider, and a dual-port data cache module generated by the M9K memory resource inside the FPGA. The processed data is sent to the STM32, which temporarily stores the data and reversely controls the gain. Finally, during actual detection, a normal echo signal can be observed on the host computer.

[0058] The effects of the present invention are further verified by specific experiments below.

[0059] This experiment was conducted on a municipal road in Wuhan. The ground penetrating radar used in the experiment was BD-GPR-200 single-frequency ground penetrating radar, with an antenna center frequency of 200MHz, 512 sampling points, and a sampling rate of 5GHz. The detection results can be found in Figure 6 , where (a) is the original radar waveform output by the ground penetrating radar, and (b) is the high-resolution underground pipeline target detection map output by the neural network. Figure 6 It can be seen that this scheme can accurately and clearly detect underground pipelines buried at a depth of 3.8m. This scheme can provide an important reference for the subsequent municipal pipeline design and construction safety.

[0060] It should be understood that the embodiments described above are part of the embodiments of the present invention, rather than all of the embodiments. In addition, the technical features in the various embodiments or single embodiments provided by the present invention can be combined with each other arbitrarily to form a feasible technical solution. Such combination is not restricted by the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0061] It should be understood that the above description of the preferred embodiment is relatively detailed and cannot be regarded as limiting the scope of patent protection of the present invention. Under the enlightenment of the present invention, ordinary technicians in this field can also make substitutions or modifications without departing from the scope of protection of the claims of the present invention, which all fall within the scope of protection of the present invention. The scope of protection requested for the present invention shall be based on the attached claims.

Claims

1. A method for detecting and identifying underground pipelines based on ground penetrating radar and neural network, characterized in that: The following steps are involved: Step 1: Set up the transmitting antenna and receiving antenna of the ground penetrating radar, transmit the radar signal through the transmitting antenna, receive the echo signal during ground penetrating radar detection through the receiving antenna, and generate the underground pipeline target detection map P1; Step 2: Input the underground pipeline target detection map P1 into the neural network to obtain a high-resolution underground pipeline target detection map; The neural network consists of a feature extraction subnetwork and a feature fusion subnetwork; The feature extraction subnetwork is composed of a first downsampling layer, a second downsampling layer and a third downsampling layer connected in sequence, and is used to sequentially downsample the input P1 to obtain hierarchical image features P = {P2, P3, P4} of different sizes; the P4 is a low-resolution coarse detection image, denoted as P′4; the feature fusion subnetwork is composed of a first upsampling operation layer, a first fusion layer, a first feature fusion decoder, a second upsampling operation layer, a second fusion layer, a second feature fusion decoder, a third upsampling operation layer, a third fusion layer and a third feature fusion decoder, and is used to convert P′4 through the first upsampling operation layer After upsampling, P′4 and P3 are added pixel by pixel through the first fusion layer and then input into the first feature fusion decoder for processing, and P′3 is output; then P′3 is upsampled through the second upsampling operation layer, added pixel by pixel with P′4, P′3, and P2 through the second fusion layer, and then input into the second feature fusion decoder for processing, and P′2 is output; finally, P′2 is upsampled through the third upsampling operation layer, added pixel by pixel with P′4, P′3, P′2, and P1 through the third fusion layer, and then input into the third feature fusion decoder for processing, so as to obtain a high-resolution underground pipeline target detection map.

2. The underground pipeline detection and identification method based on ground penetrating radar and neural network according to claim 1 is characterized in that: The feature fusion decoder is composed of two branches arranged in parallel; the first branch is composed of a dimensionality reduction convolution layer, an attention convolution layer, and a classifier connected in sequence; the second branch is composed of a dimensionality reduction convolution layer, an attention convolution layer, an upsampling layer, and a classifier connected smoothly; wherein the output of the attention convolution layer of the first branch is added pixel by pixel to the upsampling layer of the second branch and then input into the classifier of the first branch; the high-resolution feature map is processed by the first branch to output a high-resolution detail map, and the low-resolution feature map is processed by the second branch to output a low-resolution detail map; the high-resolution detail map is added pixel by pixel to the low-resolution detail map to output a high-resolution prediction map.

3. An underground pipeline detection and identification system based on ground penetrating radar and neural network, used to implement the method described in claim 1 or 2; characterized in that: It is composed of a ground penetrating radar, an echo signal acquisition processor and a host computer; the ground penetrating radar is composed of a main control unit, a transmitter, a transmitting antenna, a receiver, and a receiving antenna; the echo signal acquisition processor is composed of an echo conditioning circuit, an ADC module, an FPGA module and an STM32 module connected in sequence; the STM32 module is connected to the host computer for communication; The echo conditioning circuit is used to transform, program-control and amplify the echo signal received by the ground-penetrating radar receiver; the ADC module is used to perform analog-to-digital conversion on the conditioned echo signal to obtain a parallel data stream; the FPGA module is used to control the analog-to-digital conversion process of the ADC module, and to receive the parallel data stream and send it to the STM32 module after data processing; the STM32 module processes the received data stream and inputs it into the host computer, and sends a reverse control gain instruction to the FPGA module, controls the gain of the echo conditioning circuit through the reverse control gain instruction, and realizes program-controllable amplification or attenuation of the echo signal.

4. The underground pipeline detection and identification system based on ground penetrating radar and neural network according to claim 3 is characterized in that: The echo conditioning circuit includes a radio frequency transformer, a radio frequency amplifier with variable gain, and a fully differential amplifier connected in sequence; the radio frequency transformer is used to transform the echo signal received, and the radio frequency amplifier with variable gain is used to perform program-controlled amplification or program-controlled attenuation on the transformed echo signal before inputting it into the fully differential amplifier.

5. The underground pipeline detection and identification system based on ground penetrating radar and neural network according to claim 3 is characterized in that: The ADC module uses two analog-to-digital converters connected in parallel. First, the echo signal output by the echo conditioning circuit is sampled through the DES bilateral sampling mode in combination with the bilateral edges of the sampling clock provided by the FPGA module to obtain a parallel high-speed data stream; then the parallel high-speed data stream is sent to the FPGA module through the LVDS low-voltage differential transmission method.

6. The underground pipeline detection and identification system based on ground penetrating radar and neural network according to claim 3 is characterized in that: The FPGA module obtains two sampling clocks with a phase difference of 90° through PLL phase-locked loop frequency multiplication and phase shift processing; the FPGA module inputs sampling clocks with the same frequency and different phases into the ADC module through a high-speed data acquisition unit.

7. The underground pipeline detection and identification system based on ground penetrating radar according to claim 3 is characterized by: The FPGA module includes a high-speed data acquisition unit, a data cache unit and a data superposition unit connected in sequence; the data superposition unit is connected to the STM32 module; the FPGA module is also provided with a control gain unit, which is respectively connected to the STM32 module and the echo conditioning circuit, and is used to adjust the gain of the echo conditioning circuit; The data buffer unit is used to cache the received parallel high-speed data streams, re-join the parallel high-speed data streams into serial data, and input them into the data superposition unit for data processing; the data superposition unit corrects and de-noises the joined high-speed data streams and then sends them to the STM32 module.

8. The underground pipeline detection and identification system based on ground penetrating radar and neural network according to claim 7 is characterized in that: The data superposition unit is used to perform zero-time correction and superposition noise reduction processing on the spliced ​​high-speed data stream.

9. The underground pipeline detection and identification system based on ground penetrating radar and neural network according to claim 8, characterized in that: The zero-time correction is to correct the time difference by fixing the relative position between the transmitting antenna and the receiving antenna and taking the arrival time of the direct wave as a reference.

10. The underground pipeline detection and identification system based on ground penetrating radar and neural network according to claim 8, characterized in that: The superposition noise reduction process uses an average superposition noise reduction algorithm to process the high-speed data stream.

Images