CN-120652421-B - Urban drainage pipeline anomaly identification method based on shallow radar wave feedback model

CN120652421BCN 120652421 BCN120652421 BCN 120652421BCN-120652421-B

Abstract

The invention relates to the technical field of urban underground pipeline detection and discloses an urban drainage pipeline anomaly identification method based on a shallow radar wave feedback model, which comprises the steps of dynamically adjusting radar wave emission parameters based on a time-space joint entropy value of real-time radar echo, so that signal energy is adaptively focused in a high information entropy area; and fusing multi-scale defect characteristics by adopting a cross-scale attention pyramid network, and generating an anti-sample fine-tuning main model on line by combining a lightweight discrimination network. According to the invention, through the cooperative optimization of dynamic radar wave regulation and physical propagation model, the compatibility detection capability of metal and nonmetal pipelines is obviously improved, and the noise sensitivity feedback mechanism is utilized to convert environmental noise into model optimization parameters, so that stable identification under complex working conditions is realized.

Inventors

LIANG GUANHUA
FANG HAITAO
WANG HONGFENG
LI SHITANG
LUO HAOBIN
LI ZHONGE
ZHANG YONGMING
XIAO FEI
YANG XIAO
ZHANG RUNZHI
CHEN RUIJIE
LUO PUGUAN

Assignees

广东天驰智慧管网有限公司
广州科学城排水管理有限公司

Dates

Publication Date: 20260508
Application Date: 20250729

Claims (10)

1. A city drainage pipeline anomaly identification method based on a shallow radar wave feedback model is characterized by comprising the following steps: step S1, calculating a time domain and space domain joint entropy value through a sliding window based on a received real-time radar echo signal The size of the sliding window and the diameter of the detected pipeline are in a negative correlation relationship, and radar wave emission parameters are dynamically adjusted according to the joint entropy value so as to enhance the signal intensity of a high information entropy area and generate a self-adaptive radar wave signal; S2, constructing a priori propagation path of missing radar echo data by using a radar wave propagation physical model, repairing the missing data by adopting priority-guided cavity convolution based on the priori propagation path, wherein the repaired priority and the attenuation coefficient of the radar wave in a pipeline medium In inverse proportion to each other, where Representing the type of pipeline material; s3, extracting multi-scale defect characteristics of the repaired radar echo data through a cross-scale attention pyramid network, and dynamically adjusting weight distribution of different scale characteristics in the cross-scale attention pyramid network based on prior probability of defect types to realize fusion of the multi-scale defect characteristics; And S4, constructing a lightweight discrimination network, evaluating the credibility of the defect identification result in real time, generating an countermeasure sample based on the evaluation result, and performing on-line fine tuning on parameters of the defect identification main model to form a dynamic optimization closed loop, wherein the countermeasure sample is generated for a region with the confidence lower than a preset threshold in the current detection.
2. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 1, wherein in the step S1, the radar wave emission parameters include pulse width And frequency modulation slope And the dynamic adjustment is such that when the joint entropy value Increasing pulse width or adjusting frequency modulation slope to enhance signal energy when the joint entropy value Reducing pulse width or adjusting frequency modulation slope to reduce energy output, wherein For a first preset threshold value, A second preset threshold.
3. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 1, wherein in the step S2, the attenuation coefficient of the radar wave in the pipeline medium According to the material type of the pipeline And dynamically determining, wherein the missing points of the near-end signals are repaired preferentially for the metal pipeline, and the missing points of the far-end signals are repaired preferentially for the nonmetal pipeline.
4. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 1, wherein in the step S3, the defect types include cracks and corrosion, a pipe-defect type mapping table is pre-established to record defect characteristic frequency bands corresponding to different pipes, and the prior probability of the defect types is counted and updated based on occurrence frequencies of different types of defects in historical detection data.
5. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 4, wherein in the step S4, the generation of the challenge sample further synchronously introduces a dynamic noise feature mapping mechanism, comprising the following sub-steps: s4.1, extracting noise area signals which are not covered by defect characteristics in radar echo data in real time, and constructing a noise baseline template based on the spectrum distribution characteristics of the noise area signals; S4.2, generating a noise sensitivity weight matrix by comparing the offset of the current noise area signal and the noise baseline template, wherein the noise sensitivity weight matrix is used for quantifying the interference intensity of different frequency band noises on defect identification; And S4.3, reversely associating the noise sensitivity weight matrix with the gradient disturbance direction of the countermeasure sample, and dynamically adjusting the generation path of the countermeasure sample to enable the fine adjustment of the model to preferentially inhibit the influence of high interference noise.
6. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 5, wherein in the step S4.1, a defect feature mask filtering technology is adopted to automatically shield the identified defect area in the time-frequency diagram so as to extract a clean noise area signal, and the size of the defect feature mask is dynamically adjusted based on the size of a sliding window.
7. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 5, wherein in the step S4.2, the threshold value is obtained by frequency band energy duty ratio And judging a noise area, and marking the corresponding area as the noise area when the energy ratio of the non-defective frequency band is higher than a preset ratio, wherein the preset ratio is automatically adjusted according to the frequency domain distribution of the defect characteristics in the historical detection data.
8. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 5, wherein in the process of generating the noise sensitivity weight matrix, a coupling attenuation factor is introduced The coupling attenuation factor is calculated by the following formula: , Wherein, the Representing the energy of the noise band, Indicating the energy of the frequency band of the defect characteristic, And representing a time domain correlation coefficient, and dynamically loading the determination of the defect characteristic frequency band energy according to the pipe-defect type mapping table.
9. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 1, further comprising storing historical detection data after the step S4, wherein parameters of the lightweight discrimination network and the defect identification main model are iteratively updated based on the countermeasure sample generated in real time and the historical detection data.
10. The urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model according to claim 1, wherein the method is applied to an urban drainage pipeline system comprising cast iron pipes and/or PVC pipes.

Description

Urban drainage pipeline anomaly identification method based on shallow radar wave feedback model Technical Field The invention relates to an urban drainage pipeline anomaly identification method based on a shallow radar wave feedback model, and belongs to the technical field of urban underground pipeline nondestructive testing. Background The urban drainage pipeline is used as an important component of an underground pipe network, and the real-time monitoring of the structural health state of the urban drainage pipeline is important for urban safety operation and maintenance. At present, shallow radar wave detection technology is commonly adopted in the industry, and the characteristics of defects such as cracks, corrosion and the like in a pipeline are analyzed by transmitting electromagnetic waves with specific parameters and receiving reflected signals. Typical methods generally employ a fixed waveform parameter configuration, combined with a time-frequency analysis algorithm to perform static feature extraction. However, the following technical bottlenecks exist in practical application: 1. The existing method adopts preset radar wave pulse width and frequency modulation parameters, and is difficult to adjust the signal intensity in real time according to dynamic changes such as the humidity and sediment distribution in the pipeline. For example, metallic pipe near-end signals are subject to multiple reflection interference, while non-metallic pipe far-end signals experience signal-to-noise ratio dips due to material attenuation. The industry usually adopts manual recheck or multiple scanning compensation, but the detection efficiency is obviously reduced. 2. Aiming at the problem of radar echo signal missing, the prior art mostly adopts interpolation or neighborhood mean filling algorithm, and does not consider the propagation attenuation characteristic of radar waves in pipeline media. For example, if conventional interpolation is adopted for near-end signal deletion of a metal pipeline, false reflection peaks are introduced to cause erroneous judgment, and if material attenuation coefficients are ignored for far-end deletion repair of a non-metal pipeline, the true defect contour is difficult to recover. 3. The contradiction between multi-scale defect identification and noise suppression is that the traditional method relies on a single-scale convolution network to extract characteristics, and cannot capture high-frequency details of millimeter-scale cracks and low-frequency structural characteristics of centimeter-scale corrosion at the same time. The industry attempts to treat defects of different scales respectively by cascading a plurality of detection modules, but the characteristic coupling among the modules is insufficient and is easy to be interfered by water flow noise in a pipeline, so that the false detection rate is increased. In order to cope with the above problems, some improvement schemes attempt to introduce adaptive filtering or static noise library matching, but such methods have limitations, such as that a noise suppression module and a defect identification model operate independently, collaborative optimization in a dynamic environment is difficult to realize, a physical propagation model is separated from a data driving algorithm, and a repaired signal has deviation from a real propagation path. Therefore, how to realize the dynamic regulation and control of radar wave parameters, data restoration under physical propagation constraint, and the collaborative optimization of multi-scale feature decoupling and noise suppression becomes the technical problem to be solved by the invention. Disclosure of Invention The invention provides a shallow radar wave feedback model-based urban drainage pipeline anomaly identification method, which mainly aims to solve the problems of insufficient dynamic environment adaptability, physical misalignment of data restoration and difficult compatibility of multi-scale detection and noise suppression. In order to achieve the above purpose, the urban drainage pipeline anomaly identification method based on the shallow radar wave feedback model provided by the invention comprises the following steps: step S1, calculating a time domain and space domain joint entropy value through a sliding window based on a received real-time radar echo signal The size of the sliding window and the diameter of the detected pipeline are in a negative correlation relationship, and radar wave emission parameters are dynamically adjusted according to the joint entropy value so as to enhance the signal intensity of a high information entropy area and generate a self-adaptive radar wave signal; S2, constructing a priori propagation path of missing radar echo data by using a radar wave propagation physical model, repairing the missing data by adopting priority-guided cavity convolution based on the priori propagation path, wherein the repaired priority and the attenuation coe