A buried pipeline deformation monitoring and early warning method

Abstract

The present invention relates to the fields of municipal engineering and geotechnical engineering, and in particular to a method for monitoring and providing early warning of buried pipeline deformation. This method combines a neural network time series model with the Monte Carlo method to monitor and provide early warning of buried pipeline deformation trends. The advantages of this method include: timely detection of sudden damage to pipelines caused by construction, natural disasters, and other factors during routine monitoring; simultaneous prediction of medium- and long-term pipeline deformation trends; and early identification of unsafe conditions and specific locations. This allows for the identification of causes and the implementation of countermeasures to prevent potential damage.

Description

Technical Field

[0001] The present invention relates to the technical fields of municipal engineering and geotechnical engineering, and in particular to a buried pipeline deformation monitoring and early warning method. Background Art

[0002] Deformation of buried pipelines occurs due to vertical deformation of the pipeline and the underlying soil, caused by the weight of the pipeline, the soil covering the pipeline, and additional ground loads. Uniform deformation generally poses little threat to the pipeline, but if localized deformation is excessive, exceeding the strain limit of the pipe material, the pipeline can rupture, leaking the contents. In severe cases, it can even cause soil loss and ground collapse, resulting in accidents.

[0003] In recent years, many cities have experienced ground collapse accidents. Investigations and analysis have shown that approximately 60% of these incidents were caused by leaks in pipelines surrounding the accident sites, resulting in sand flow. Leaks also affect the normal operation of pipelines, increasing maintenance and repair costs.

[0004] There are a lot of research results on the deformation laws and theoretical calculations of buried pipelines, and there are also a variety of relatively reliable monitoring methods, but the existing methods focus on alarms after the pipeline has ruptured or even caused an accident.

[0005] At present, traditional research methods mainly start from the perspective of the causal relationship between force and deformation, and propose different methods for calculating the soil pressure on the top of the pipe. Marston first proposed the soil pressure theory of buried pipelines based on the relative displacement of pipes and soil; Gu Anquan proposed a calculation method for the vertical soil pressure of buried pipelines based on indoor model tests and a large amount of field measurement data; An Zongwen et al. considered the influence of principal stress and backfill soil and established a new method for calculating the vertical soil pressure of straight trench buried pipelines; Zhe Xuesen et al. made a breakthrough by considering the influence of valley topography on pipeline soil pressure and proposed a soil pressure calculation method that can be applied to various buried pipelines. In addition, with the popularization of finite element calculation software such as ABAQUS, many scholars have also adopted the finite element method to model and calculate the theoretical deformation of pipelines.

[0006] Traditional causal relationship research methods are suitable for application in the structural design stage when there is a lack of measured data. Their limitation is that they require the assumption of many boundary conditions, which often change greatly during the pipeline construction and use stages. This leads to a large deviation between the theoretical calculation results and the actual results.

[0007] Machine learning algorithms do not consider the causal relationships of the research subjects; they simply adjust model weights based on large amounts of monitoring data to achieve predictions. However, their disadvantage is that they require the identification of appropriate parameter combinations to achieve good prediction results. However, the theory of optimal parameter combinations is still immature, and manual trial and error adjustments are often required in many engineering practices. Summary of the Invention

[0008] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and provide a method for monitoring and early warning of buried pipeline deformation. By analyzing the deformation monitoring data of buried pipelines and combining a neural network time series model with the Monte Carlo method, a method and process for predicting deformation trends is provided. This method enables timely detection of sudden damage to pipelines caused by construction, natural disasters, etc. during daily monitoring, and at the same time predicts the medium- and long-term deformation trends of pipelines, thereby detecting unsafe conditions and specific locations in advance, identifying the causes and taking countermeasures to prevent them from happening.

[0009] The purpose of the present invention is achieved by the following technical solutions:

[0010] A method for monitoring and early warning of deformation of buried pipelines, characterized in that the method comprises the following steps:

[0011] Inputting historical deformation monitoring data of the buried pipeline into a neural network autoregressive model to determine the position of the pipeline where the deformation of the buried pipeline is the largest;

[0012] The predicted values ​​of the monitoring points on both sides of the pipeline position with the maximum deformation of the buried pipeline are used as the expected values. The standard deviation of the deformation monitoring historical data of the monitoring points is calculated at the same time. Two normally distributed random variables N1 and N2 are constructed. The two random variables are then subtracted to obtain a normally distributed pipeline deformation difference random variable N3. The Monte Carlo method is used for simulation calculation to obtain the mean and standard deviation of N. The confidence level of N is determined based on the importance of the buried pipeline, thereby obtaining the confidence interval of N.

[0013] The ultimate deformation threshold of the buried pipeline is determined according to the structural form and material form of the buried pipeline. If the ultimate deformation threshold is within the confidence interval of N, it is determined that the position of the buried pipeline may be damaged due to deformation.

[0014] The deformation history detection data of the buried pipeline is obtained by arranging monitoring points at intervals along the pipeline laying direction of the buried pipeline, and each monitoring point is provided with a monitoring device.

[0015] The monitoring equipment is buried in the cushion layer below the pre-buried pipeline, or buried above the pre-buried pipeline; a protective layer is provided on the periphery of the detection equipment buried above the pre-buried pipeline.

[0016] The buried pipeline is monitored daily. The standard deviation of the deformation monitoring historical data of the previous monitoring cycle is calculated, the monitoring data of the previous day is used as the mean, and the confidence level is determined according to the importance of the buried pipeline, thereby obtaining a confidence interval as a reasonable range of change for the day.

[0017] The neural network autoregressive model is a neural network time series model. The deformation monitoring historical data of the buried pipeline is used as the training set of the neural network time series model to predict the cumulative deformation prediction value of the buried pipeline after n days. The cumulative deformation of each monitoring point before the training set is added to the predicted value to obtain Pn. The first-order difference of Pn of each adjacent monitoring point is performed to obtain a predicted deformation difference sequence of each adjacent monitoring point on the nth day. The position between the two monitoring points associated with the maximum absolute value in the predicted deformation difference sequence is the pipeline position of the buried pipeline with the maximum deformation.

[0018] The standard deviation of the two monitoring points in the pipeline deformation monitoring historical data is added to the cumulative deformation values ​​of the two before the monitoring period, and the expected value and the standard deviation are used to form two normally distributed random variables N1 and N2.

[0019] The pipeline deformation monitoring history is a data set updated at certain time intervals, and the data set contains buried pipeline deformation monitoring history data within a monitoring period.

[0020] The advantages of the present invention are: it can carry out daily monitoring of the deformation of buried pipelines, and predict deformation trends and dangerous parts at the same time with high prediction accuracy, which has significant guiding significance for subsequent repair and maintenance operations; based on the prediction results, by finding the causes and taking countermeasures, the hidden dangers of buried pipelines can be ultimately eliminated; it can not only reduce the ground collapse accidents caused by them and ensure the normal use of pipelines, but also save a lot of pipeline repair and maintenance costs; it has significant social and economic benefits and is suitable for promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is the buried pipeline monitoring arrangement form I in the present invention;

[0022] Figure 2 for Figure 1 AA section view;

[0023] Figure 3 This is the buried pipeline monitoring arrangement form II of the present invention;

[0024] Figure 4 for Figure 3 BB cross-sectional view;

[0025] Figure 5 It is a deformation prediction schematic diagram of the present invention;

[0026] Figure 6 This is a structural diagram of the neural network model in the present invention. DETAILED DESCRIPTION

[0027] The features of the present invention and other related features are further described in detail below through embodiments in conjunction with the accompanying drawings to facilitate understanding by those skilled in the art:

[0028] like Figure 1-6 As shown, the marks in the figure respectively represent: buried pipeline 1, soil 2, monitoring equipment 3, base 4, cushion layer 5, protective layer 6, monitoring point 1#, monitoring point 2#, monitoring point 3#, and monitoring point 4#.

[0029] Example: Figure 1 or Figure 3 As shown, the buried pipeline deformation monitoring and early warning method in this embodiment is used to monitor the deformation of the buried pipeline 1 buried in the soil 2, and adopts a neural network time series model combined with the Monte Carlo method to predict the deformation trend based on the deformation monitoring data, thereby triggering an early warning before the buried pipeline 1 ruptures due to deformation, so as to find the cause, take countermeasures, and ultimately eliminate the hidden dangers of the buried pipeline.

[0030] First, this embodiment provides two buried pipeline deformation monitoring arrangement forms. According to actual construction conditions and design requirements, either of the two arrangements can be used to monitor the deformation of the buried pipeline.

[0031] like Figure 1 and Figure 2 As shown, the buried pipeline 1 is buried in the soil 2 by being erected on a base 4, and a cushion layer 5 is provided below the base 4. The cushion layer 5 is used to provide support to the base 4 and the buried pipeline 1 erected above the base 4 to prevent the buried pipeline 1 from sinking as much as possible.

[0032] The monitoring device 3 is buried in the cushion layer 5. A plurality of monitoring devices 3 are arranged at intervals in the cushion layer 5 along the laying direction of the buried pipeline 1. Each monitoring device 3 is used to monitor deformation at a different position of the buried pipeline 1. Figure 1 The illustration shows only four monitoring points arranged within the cushion layer 5 beneath this section of buried pipeline 1: Monitoring Point 1#, Monitoring Point 2#, Monitoring Point 3#, and Monitoring Point 4#. However, in actual design and construction, the number of monitoring points and the spacing between adjacent monitoring points can be varied as needed. For example, to more precisely locate the location of pipeline damage or the most likely location of damage, a larger number of monitoring points with closer spacing could be installed beneath buried pipeline 1, but this would increase construction costs. The same applies vice versa.

[0033] like Figure 3 and Figure 4As shown, another form of deformation monitoring of the buried pipeline 1 is to arrange the monitoring equipment 3 at the top position of the buried pipeline 1; and in order to protect the monitoring equipment 3, a protective layer 6 is provided on its periphery, and the monitoring equipment 3 is protected by the protective layer 6 to avoid being adversely affected by factors such as the soil 2 or groundwater, acidic and corrosive substances contained in the soil 2.

[0034] The deformation of the buried pipeline 1 can be monitored by both of the above two deformation monitoring modes, and the monitoring data of the monitoring equipment 3 can be centrally collected and processed and stored accordingly.

[0035] The buried pipeline deformation monitoring and early warning method of this embodiment uses a neural network time series autoregressive model (NAR). First, the historical monitoring data collected by the monitoring device 3 is input to solve the model's hidden layer nodes and activation function weights. Then, a multi-step prediction is performed to determine the pipeline location with the largest local deformation, i.e., the location where the buried pipeline has been damaged or is most likely to be damaged.

[0036] In order to overcome the cumulative errors that may be caused by multi-variable and multi-step predictions of the neural network model, the local deformation prediction results are characterized in the form of random variables.

[0037] Using the predicted values ​​of two adjacent monitoring points at the previously determined maximum local pipeline deformation location as the expected value, and calculating the standard deviation of the historical monitoring data, we construct two normally distributed random variables, N1 and N2. Subtracting these two random variables yields a normally distributed random variable, N3, representing the pipeline deformation difference. Using the Monte Carlo method (a statistical test method), we perform numerous simulations to determine the mean and standard deviation of N. Based on the pipeline's importance, we determine the confidence level for N, thereby obtaining a confidence interval for N.

[0038] Based on the pipeline's structure and material properties (e.g., round tube, square tube, reinforced concrete, plastic, steel, etc.), theoretical calculations, engineering tests, or literature review are used to determine the ultimate deformation threshold for pipeline failure. If the threshold falls within the confidence interval of the predicted random variable N, the pipeline is likely to fail.

[0039] Specifically, if Figure 5 and Figure 6 As shown, the method in this embodiment includes the following steps:

[0040] 1. Sudden deformation monitoring:

[0041] Daily monitoring: The standard deviation of the data from the previous monitoring cycle is calculated for each monitoring point. The previous day's monitoring data is used as the mean. A confidence level is determined based on the importance of the pipeline, and thus a confidence interval is determined as the reasonable range of variation for the day. If the monitoring data for the day exceeds this range, an alarm is triggered.

[0042] 2. Early warning of medium- and long-term pipeline deformation:

[0043] Phase 1: Data Preparation

[0044] 1) Obtain historical data on pipeline deformation monitoring for the most recent monitoring period (e.g., 30 days) (excluding the construction phase);

[0045] 2) Calculate the cumulative deformation Qn of each monitoring point before the monitoring period.

[0046] Phase 2: Determine the location of the pipe segment with the maximum local deformation

[0047] 3) Use the data set in step 1 as the training set and use Figure 6 The neural network time series model (NAR) shown predicts the cumulative deformation after n days (multiple predictions are averaged);

[0048] 4) Add the cumulative deformation of each monitoring point before the training set to the predicted value obtained in step 3 to obtain Pn;

[0049] 5) Make the first-order difference of Pn of each adjacent monitoring point to obtain the predicted deformation difference sequence of each adjacent monitoring point on the nth day;

[0050] 6) Filter the absolute maximum value in the deformation difference sequence obtained in step 5, and record the two monitoring points related to it as Jm and Jm+1. Monitoring point Jm and monitoring point Jm+1 are the two monitoring points adjacent to the position of maximum local deformation of the pipeline; (this determines the position of the pipeline segment with maximum local deformation).

[0051] Phase 3: Predicting deformation range

[0052] 7) Calculate the standard deviations dm and dm+1 of the two monitoring points Jm and Jm+1 in the data set of step 1 respectively;

[0053] 8) Add the predicted values ​​of the monitoring points Jm and Jm+1 obtained in step 3 to the accumulated deformation in step 2 as the expected value and the standard deviation obtained in step 7 to form two normally distributed random variables N1(Jm, dm) and N2(Jm+1, dm+1).

[0054] 9) The deformation difference random variable N3 = N2 - N1 is simulated multiple times using the Monte Carlo method to determine the mean and standard deviation of N3. Since N3 also obeys the normal distribution, a confidence level can be manually specified based on the importance of the pipeline to determine the confidence interval (for example, for important pipelines, the confidence level is specified as 95%, and its confidence interval can be determined as: mean ± 1.96 × standard deviation).

[0055] Stage 4: Deformation Warning

[0056] 10) Compare the ultimate deformation threshold of the buried pipeline (which can be determined in advance through theoretical calculations, engineering tests, literature review, etc.) with the confidence interval determined in step 9. If the ultimate deformation threshold of the buried pipeline does not enter the confidence interval, it is judged that the pipeline deformation is safe and will not rupture. Conversely, if the ultimate deformation threshold enters the confidence interval, it is judged that the pipeline at this location is likely to be damaged.

[0057] 11) Investigate pipeline locations predicted to be at risk of damage and take countermeasures.

[0058] 3. Update the dataset

[0059] 12) Update the dataset regularly (e.g., update it every 10 days, while maintaining a 30-day dataset, excluding data from 30-21 days before the update date and adding the latest 10 days of data). Repeat steps 2-9 to provide early warning of mid-term pipeline trends.

[0060] When the monitoring and early warning method in this embodiment is applied to a certain engineering case, the following steps and application effects are involved:

[0061] 1) Install buried pipeline deformation monitoring sensors and collect data;

[0062] 2) This section of pipeline has 20 monitoring points, spaced 1 meter apart. Each monitoring period is 30 days (for example, February 22, 2022, to March 23, 2022). The cumulative deformation value, Qn, of each monitoring point before the current monitoring period is calculated. The goal is to predict whether the pipeline will be at risk of damage at the point where the maximum local deformation occurs 10 days later (April 2).

[0063] 3) Import all monitoring point data of this detection cycle into the neural network model (NAR), set parameters such as the number of hidden layers and time lag, train the neural network and output the predicted value Yn for each monitoring point 10 days later (April 2).

[0064] 4) The predicted value Yn of each monitoring point is added to the historical cumulative deformation value Qn obtained in step 2 to obtain the cumulative predicted result Pn. Then, the adjacent monitoring points are differentiated (P(n+1)-P(n)) to obtain the differential deformation values ​​between the monitoring points. The pipeline segment with the largest absolute value is the one with the largest local deformation (in this case, the maximum deformation difference between the 14th and 15th monitoring points is 12.36), as shown in the following table:

[0065]

[0066] 5) Calculate the standard deviations dm and dm+1 of the two adjacent monitoring points in the monitoring period data set for the pipe segment with the maximum local deformation (d14 = 0.1233 and d15 = 0.1233, respectively, in this case).

[0067] 6) Construct the predicted values ​​of monitoring points 14# and 15# on April 2 expressed in the form of normally distributed random variables: monitoring point 14# is N14 (-1.6621, 0.1233), and monitoring point 15# is N15 (-2.2614, 0.1233).

[0068] 7) Calculate the differential deformation of the pipeline between monitoring points 14# and 15# as N = N15 - N14. Use the Monte Carlo method to perform multiple simulations to determine the mean and standard deviation of the normally distributed variable N. In this case, after 1000 simulations, the mean was determined to be -0.59 and the standard deviation was 0.17. Specifying a 95% confidence level, the confidence interval is the mean ±1.96 × 0.17. When applied to engineering monitoring, the mean should be added to the cumulative value of -11.7644 prior to this monitoring period. The predicted differential deformation range for this pipe segment 10 days later (April 2nd) is [-12.688, -12.021].

[0069] 8) According to finite element analysis, the maximum limit deformation value of the pipeline is -20. Therefore, the predicted range of differential deformation calculated in step 7 does not exceed the limit value and there is no risk of damage.

[0070] 9) Ten days later, update the monitoring data set from step 2 to include data from March 4th to April 2nd. Predict the deformation data for April 12th, ten days later, and repeat steps 2-8. Simultaneously, verify on April 2nd that the buried pipeline segment is structurally stable and has not suffered any damage.

[0071] 10) Daily Monitoring: At each monitoring point, the standard deviation is calculated based on the data from the previous period (in this case, February 22, 2022, to March 23, 2022). The previous day's monitoring data is used as the mean. A confidence level is determined based on the pipeline's importance, and thus a confidence interval is determined, which serves as the reasonable range of variation for the day. If the monitoring data for the day exceeds this range, an alarm is issued.

[0072] Although the above embodiments have described in detail the concepts and embodiments of the present invention with reference to the accompanying drawings, ordinary technicians in this field can recognize that various improvements and modifications can still be made to the present invention without departing from the scope of the claims, so they are not described in detail here.

Claims

1. A buried pipeline deformation monitoring and early warning method, characterized by: The method comprises the following steps: Inputting historical deformation monitoring data of the buried pipeline into a neural network autoregressive model to determine the position of the pipeline where the deformation of the buried pipeline is the largest; The predicted values ​​of the monitoring points on both sides of the pipeline position with the maximum deformation of the buried pipeline are used as the expected values. The standard deviation of the deformation monitoring historical data of the monitoring points is calculated at the same time. Two normally distributed random variables N1 and N2 are constructed. The two random variables are then subtracted to obtain a normally distributed pipeline deformation difference random variable N3. The Monte Carlo method is used for simulation calculation to obtain the mean and standard deviation of N. The confidence level of N is determined based on the importance of the buried pipeline, thereby obtaining the confidence interval of N. determining a limit deformation threshold of the buried pipeline according to the structure and material of the buried pipeline, and if the limit deformation threshold is within the confidence interval of N, determining that the position of the buried pipeline may be damaged due to deformation; The neural network autoregressive model is a neural network time series model, which uses the deformation monitoring historical data of the buried pipeline as a training set of the neural network time series model to predict the cumulative deformation prediction value of the buried pipeline after n days; The cumulative deformation of each monitoring point before the training set is added to the predicted value to obtain Pn. The first-order difference of Pn of each adjacent monitoring point is performed to obtain the predicted deformation difference sequence of each adjacent monitoring point on the nth day. The position between the two monitoring points associated with the maximum absolute value in the predicted deformation difference sequence is the pipeline position with the largest deformation of the buried pipeline.

2. The buried pipeline deformation monitoring and early warning method according to claim 1 is characterized by: The deformation history detection data of the buried pipeline is obtained by arranging monitoring points at intervals along the pipeline laying direction of the buried pipeline, and each monitoring point is provided with a monitoring device.

3. The method for monitoring and early warning of buried pipeline deformation according to claim 2, characterized in that: The monitoring device is buried in the cushion layer below the pre-buried pipeline, or buried above the pre-buried pipeline; a protective layer is provided on the periphery of the detection device buried above the pre-buried pipeline.

4. The method for monitoring and early warning of buried pipeline deformation according to claim 1, characterized in that: The buried pipeline is monitored daily. The standard deviation of the deformation monitoring historical data of the previous monitoring cycle is calculated, the monitoring data of the previous day is used as the mean, and the confidence level is determined according to the importance of the buried pipeline, thereby obtaining a confidence interval as a reasonable range of change for the day.

5. The method for monitoring and early warning of buried pipeline deformation according to claim 1, characterized in that: The standard deviation of the two monitoring points in the pipeline deformation monitoring historical data is added to the cumulative deformation values ​​of the two before the monitoring period, and the expected value and the standard deviation are used to form two normally distributed random variables N1 and N2.

6. The buried pipeline deformation monitoring and early warning method according to claim 1 is characterized by: The pipeline deformation monitoring history is a data set updated at certain time intervals, and the data set contains buried pipeline deformation monitoring history data within a monitoring period.

Images