Pipeline detection equipment and method for predicting shock wave bearing capacity of pipeline

Abstract

The invention provides pipeline detection equipment and a method for predicting the bearing capacity of a pipeline, and the equipment comprises an equipment housing which is used for being assembled in a detected pipeline; the advancing device is arranged outside the equipment shell and is used for driving the pipeline detection equipment to move in the detected pipeline; the detection device is arranged on the equipment shell and is used for detecting corrosion data of the detected pipeline; the data processing device is connected with the detection device and is configured to obtain the corrosion data, the pipeline type of the detected pipeline, the environment data and the explosion data; and by adopting an SOM-BP neural network model, at least according to the corrosion data, predicting the minimum explosive amount capable of causing damage to the tested pipeline under the condition that the explosive of a set type is detonated at a preset position. According to the technical scheme provided by the invention, the minimum explosive amount causing damage to the tested pipeline can be obtained, so that the purpose of improving the safety of the tested pipeline is achieved.

Description

Technical Field

[0001] The invention relates to the field of prediction of pipeline's ability to withstand shock waves, and in particular to a pipeline detection device and a method for predicting pipeline's ability to withstand shock waves. Background Art

[0002] Urban pipelines are an important part of urban infrastructure. They are pipeline systems used to transport various materials and energy. They are usually buried underground. According to different usage scenarios and functions, urban pipelines can be divided into water supply pipelines, drainage pipelines, gas pipelines, thermal pipelines and industrial pipelines. Pipelines buried underground for a long time will be corroded due to environmental influences, which will weaken the pressure-bearing capacity of urban pipelines. When there are blasting operations around urban pipelines or explosions inside, the shock waves generated will impact urban pipelines and damage the locations where urban pipelines are prone to corrosion, making urban pipelines unable to be used normally, affecting economic production and residents' lives.

[0003] Based on the above reasons, in order to ensure the safety of urban pipeline operation, it is necessary to detect the bearing capacity of urban pipelines to determine the amount of explosives used in blasting operations, ensure that the impact force generated by blasting operations will not cause damage to urban pipelines, and improve the safety of urban pipelines. The current method for obtaining the degree of corrosion of urban pipelines is to detect the degree of corrosion of urban pipelines through non-excavation means such as magnetic leakage technology, ultrasonic technology, and eddy current technology, and then use machine learning algorithms or single neural network models to predict the ability of the tested pipeline to withstand shock waves based on the degree of corrosion.

[0004] However, the methods for obtaining the shock wave bearing capacity of urban pipelines in the prior art have poor accuracy in predicting the shock wave bearing capacity of urban pipelines, and cannot provide a reliable basis for determining the amount of explosives during blasting operations. Summary of the invention

[0005] The present invention provides a pipeline detection device and a method for predicting the ability of a pipeline to withstand shock waves, which are used to accurately predict the minimum amount of explosives that can cause damage to the pipeline under test when a set type of explosive is detonated at a preset position, thereby avoiding damage to the pipeline under test by blasting operations, so as to achieve the purpose of improving the safety of the pipeline under test.

[0006] In a first aspect, the present invention provides a method for predicting the ability of a pipeline to withstand shock waves, comprising:

[0007] Acquiring the corrosion data, wherein the corrosion data includes a pipeline type, environmental data, and explosion data of the pipeline under test;

[0008] Get the pre-trained SOM-BP neural network model;

[0009] constructing input data of the SOM-BP neural network model according to one or more of the corrosion data, pipeline type, environmental data and explosion data;

[0010] The input data is input into the SOM-BP neural network model to predict the minimum amount of explosives that can cause damage to the corroded pipeline when a set type of explosive is detonated at a preset position.

[0011] Furthermore, the step of obtaining the corrosion data includes:

[0012] The detection data of a plurality of ultrasonic probe groups are obtained, and the corrosion data is obtained according to an average value of each of the detection data.

[0013] Furthermore, the step of obtaining the corrosion data includes:

[0014] Acquire detection data of at least one pulsed eddy current probe group, and acquire the corrosion data according to the detection data.

[0015] Furthermore, the step of obtaining the corrosion data includes:

[0016] Acquire first detection data of the ultrasonic probe group and second detection data of the pulsed eddy current probe group;

[0017] A weighted sum is performed on the first detection data and the second detection data to obtain the corrosion data.

[0018] Furthermore, after the step of predicting the minimum amount of explosives that can cause damage to the corroded pipeline when the explosives of the set type are detonated at the preset position, the method further includes:

[0019] The minimum amount of explosive is sent to the user.

[0020] Furthermore, the step of obtaining the pre-trained SOM-BP neural network model includes:

[0021] Obtain training data set and initialized SOM-BP neural network model;

[0022] The initialized SOM-BP neural network model is trained according to the training data set by using a mean square error function as a loss function and a gradient descent optimization algorithm to obtain the pre-trained SOM-BP neural network model.

[0023] In a second aspect, the present invention further provides a pipeline detection device, comprising:

[0024] A device housing, which is used to be assembled inside the pipeline to be tested;

[0025] A traveling device, which is arranged outside the device housing and is used to drive the pipeline detection device to move inside the detected pipeline;

[0026] A detection device, which is arranged on the equipment housing and is used to detect corrosion data of the tested pipeline;

[0027] A data processing device is connected to the detection device and is configured to execute any of the above methods for predicting the pipeline's ability to withstand shock waves.

[0028] Furthermore, the traveling device includes a power mechanism, a support structure and a roller, wherein the support structure is used to assemble the roller on the equipment housing, and the power mechanism is connected to the roller through transmission, so that the pipeline detection equipment moves inside the detected pipeline by driving the roller.

[0029] Furthermore, the support structure is movably installed on the device housing, a central body is arranged inside the device housing, and the central body is connected to the support structure through an elastic telescopic member.

[0030] Furthermore, the detection device includes an ultrasonic probe group and / or a pulsed eddy current array, wherein:

[0031] The ultrasonic probe group includes a plurality of ultrasonic probes, and the plurality of ultrasonic probes are evenly distributed around the circumference of the device housing;

[0032] The pulsed eddy current array includes a plurality of pulsed eddy current probes, and the plurality of pulsed eddy current probes are evenly distributed around the circumference of the device housing.

[0033] According to the technical solution of the present invention, a traveling device and a detection device are arranged on the equipment housing of the pipeline detection device. After the pipeline detection device is assembled into the pipeline to be detected, the pipeline detection device can be driven to travel in the pipeline to be detected by the traveling device, and the corrosion data of the pipeline to be detected can be detected by the detection device during the traveling process, thereby performing a comprehensive detection on the pipeline to be detected, and improving the reliability of the corrosion data detection of the pipeline to be detected. In addition, in the technical solution of the present invention, the data processing device of the pipeline detection device can adopt a pre-trained SOM-BP neural network model to predict the minimum amount of explosives that can cause damage to the pipeline to be detected when the set type of explosives is detonated at a preset position according to the corrosion data, pipeline type, environmental data and explosion data of the pipeline to be detected; since the SOM-BP neural network model is a composite neural network model based on the SOM neural network and the BP neural network, compared with the method of adopting a single neural network model, the accuracy of the prediction of the minimum amount of explosives can be improved.

[0034] Based on the following detailed description of specific embodiments of the present invention in conjunction with the accompanying drawings, those skilled in the art will become more aware of the above and other objects, advantages and features of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Hereinafter, some specific embodiments of the present invention will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the accompanying drawings:

[0036] Figure 1 is a schematic structural diagram of a pipeline detection device according to an embodiment of the present invention;

[0037] Figure 2 is a schematic flow chart of a method for predicting the ability of a pipeline to withstand shock waves according to an embodiment of the present invention;

[0038] Figure 3 is a schematic structural diagram of a SOM-BP neural network model according to an embodiment of the present invention;

[0039] Figure 4 is a schematic flow chart of clustering input data of corrosion points by a SOM neural network sub-model according to an embodiment of the present invention;

[0040] Figure 5 is a schematic structural diagram of a pipeline detection device according to another embodiment of the present invention;

[0041] Figure 6 is a schematic structural diagram of a pipeline detection device according to another embodiment of the present invention;

[0042] Figure 7 is a schematic structural diagram of a pipeline detection device according to another embodiment of the present invention;

[0043] Figure 8 is a schematic structural diagram of a pipeline detection device according to another embodiment of the present invention;

[0044] Fig. 9 is a schematic structural diagram of a pipeline detection device according to another embodiment of the present invention;

[0045] Fig.10 is a schematic structural diagram of a pipeline detection device according to another embodiment of the present invention;

[0046] Fig.11 is a schematic structural diagram of a pipeline detection device according to another embodiment of the present invention. DETAILED DESCRIPTION

[0047] Refer to the following Figures 1 to 11 To describe a pipeline detection device and a method for predicting the ability of a pipeline to withstand shock waves in an embodiment of the present invention. In the description of this embodiment, it should be understood that the terms "first" and "second" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features, that is, include one or more of the features. In the description of the present invention, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. When a feature "includes or contains" one or some of the features it covers, unless otherwise specifically described, this indicates that other features are not excluded and may further include other features.

[0048] See also Figure 1 , Figure 1 What is shown is a schematic structural diagram of a pipeline detection device in one embodiment of the present invention. The pipeline detection device is used to detect corrosion data of the pipeline under test, and accurately predict the minimum amount of explosives that can cause damage to the pipeline under test when a set type of explosive is detonated at a preset position based on at least the detected corrosion data, so as to improve the safety of the pipeline under test during blasting operations.

[0049] exist Figure 1 In the structure shown, the pipeline detection device includes a device housing 1, on which a detection device 2, a traveling device 3 and a data processing module 4 are arranged, wherein the shape of the device housing 1 matches the shape of the pipeline to be detected. Since the shape of the pipeline to be detected in the prior art is mostly cylindrical, the device housing 1 in this embodiment is cylindrical. When the pipeline detection device is used to detect the device to be detected, the device housing 1 can be assembled inside the pipeline to be detected.

[0050] The traveling device 3 in this embodiment is arranged on the outside of the equipment housing 1. After the pipeline detection equipment is assembled into the interior of the pipeline to be detected, the traveling device 3 contacts the interior of the pipeline to be detected, which can not only provide support for the pipeline detection equipment, but also move on the inner wall of the pipeline to be detected to drive the pipeline detection equipment to move inside the pipeline to be detected.

[0051] When the pipeline detection equipment of this embodiment moves inside the pipeline under test, the detection device 2 can send out a detection signal to detect the corrosion data of the pipeline under test; the data processing module 4 can adopt a logic control device with data processing function such as a single-chip microcomputer, and is connected to the detection equipment 2, and can receive the corrosion data of the pipeline under test detected by the detection device 2, and predict the bearing capacity of the pipeline under test based on the corrosion data, and the bearing capacity includes: the minimum amount of explosives that can cause damage to the pipeline under test when a set type of explosives is detonated at a preset position.

[0052] In this embodiment, the method for the data processing module 4 to predict the bearing capacity of the measured pipeline according to the corrosion data of the measured pipeline is as follows: Figure 2 As shown, the following steps are included:

[0053] Step S101: Obtain the pipeline type, corrosion data and explosion data of the pipeline under test;

[0054] Step S102: obtaining a pre-trained SOM-BP neural network model;

[0055] Step S103: constructing input data of the SOM-BP neural network model according to the pipeline type, corrosion data and explosion data of the tested pipeline;

[0056] Step S104: inputting the input data into the pre-trained SOM-BP neural network model to obtain the minimum amount of explosives that can cause damage to the tested pipeline when the set type of explosives is detonated at the preset position.

[0057] In the above step S101, the corrosion data of the measured pipeline are obtained, including the corrosion position, corrosion thickness and corrosion area of ​​each corrosion point on the measured pipeline, and the explosion data are obtained, including the soil parameters around the measured pipeline, the explosion pressure and the distance from the explosion source.

[0058] In this embodiment, the corrosion coefficient of each corrosion position can be calculated according to the corrosion thickness of each corrosion position, and the corrosion degree of the measured pipeline at the corresponding position can be obtained according to the corrosion coefficient. The formula for calculating the corrosion coefficient of the measured pipeline in this embodiment is:

[0059]

[0060] Among them, Y i is the corrosion coefficient of the i-th section of the tested pipeline, P i -1 is the limit pressure of the i-th section of the tested pipeline, d″ i is the corrosion thickness of the i-th section of the tested pipeline, d′ i Indicates the thickness of the i-th section of the measured pipeline.

[0061] In the above step S102, a training data set may be obtained, and the training data set may be used to train the SOM-BP neural network model to obtain a pre-trained SOM-BP neural network model. The pre-trained SOM-BP neural network model may also be stored in a database, and the pre-trained SOM-BP neural network model may be read from the database during the execution of step S102.

[0062] The SOM-BP neural network model in this embodiment includes a SOM neural network sub-model and a BP neural network sub-model. Figure 2 As shown, the SOM neural network sub-model includes an input layer and a competition layer, and the BP neural network sub-model includes a hidden layer and an output layer.

[0063] In the above step S103, the corrosion thickness, corrosion area, surrounding soil parameters, distance from the explosion source and explosion pressure of each corrosion point are respectively used as input data corresponding to each corrosion point, thereby completing the construction of input data.

[0064] In the above step S104, after the input data of each corrosion point is input into the SOM-BP neural network model, the SOM neural network sub-model in the SOM-BP neural network model can cluster the input data of each corrosion point respectively to obtain the damage type of each corrosion point; then the damage type of each corrosion point is input into the BP neural network sub-model, and the BP neural network sub-model predicts the minimum amount of explosives that can cause damage to the measured pipeline when the set type of explosives is detonated at a preset position according to the damage type of each corrosion point.

[0065] The SOM neural network sub-model in this embodiment is an unsupervised self-organizing learning network model based on the SOM neural network, which can automatically recognize and classify input data without prior knowledge. In this embodiment, the SOM neural network sub-model includes an input layer and a competition layer, wherein the input layer transmits the input data to the competition layer through a weight vector; the competition layer adopts a two-dimensional array topology similar to the image of the cerebral cortex, and clusters the input data through a number of neurons in parallel.

[0066] Taking one of the corrosion points as an example, let the number of nodes in the input layer be N, and the number of nodes in the competition layer be M. Each node in the input layer corresponds to a type of data in the input data. The i-th type of data in the input data of the corrosion point is x i , then the input data of the corrosion point is:

[0067] X={x1,x2,x3,...,x N}

[0068] The method of clustering the input data of the corrosion point by the SOM neural network sub-model is as follows: Figure 4 As shown, the following steps are included:

[0069] Step S201: normalizing the input data of the corrosion points;

[0070] Step S202: Calculate the Euclidean distance between each neuron in the competition layer and the input data of the corrosion point;

[0071] Step S203: taking the neuron with the smallest Euclidean distance to the input data of the corrosion point in the competition layer as the winning neuron, and obtaining the winning neighborhood in the competition layer according to the winning neuron;

[0072] Step S204: updating the connection weight vectors of neurons in the winning neighborhood of the competition layer according to the learning rate and the update neighborhood function;

[0073] Step S205: determine whether the learning rate is 0;

[0074] If yes, execute step S206;

[0075] If not, execute step S207;

[0076] Step S206: Update the learning rate and update the neighborhood function, and then return to step S204;

[0077] Step S207: obtaining a clustering result of the input data according to the updated superior neighborhood, and obtaining the damage type of the tested pipeline at the corrosion position according to the clustering result;

[0078] In the above step S201, the maximum value of the input data of each corrosion point is assumed to be x max , the minimum value is x min , let the input data of the corrosion point after normalization be X′, where the i-th type of data is x′ i ,but

[0079] X′={x′1, x′2, x′3,...,x′ N}

[0080]

[0081] In the above step S202, the method for calculating the Euclidean distance between each neuron in the output layer and the input data includes:

[0082] Let the connection weight vector between the jth neurons in the competition layer be W j , the i-th component vector in the connection weight vector W is W ij ,but

[0083] W j = {W 1j , W 2j , W 3j , ..., W Nj}

[0084] Let the Euclidean distance between the input data and the jth neuron in the competitive layer be d j ,but

[0085]

[0086] In the above step S203, the winning neighborhood in the competition layer obtained refers to the area in the competition layer whose Euclidean distance to the winning neural unit is within a preset distance range; in this embodiment, after obtaining the winning neuron, the Euclidean distance between each neuron in the competition layer and the winning neuron is calculated, and the neurons whose Euclidean distance to the winning neuron is less than the set distance threshold range are regarded as neurons in the winning neighborhood.

[0087] In the above step S204, assuming that after the t-th iteration update, the connection weight vector of the j-th neuron in the winning neighborhood is Wj(t), and the i-th component vector in the connection weight vector is Wj(t), then the connection weight vector of the neuron in the winning neighborhood is updated by the following formula:

[0088] W′ ij (t+1)=W′ ij (t)+α(t)h jm (t)[xi-W′ ij (t)]

[0089] Among them, α(t) is the learning rate updated at the tth iteration, α(0) is the initial learning rate, and h jm (t) is the updated neighborhood function after the tth iteration.

[0090] In the above step S206, the calculation formula used to update the learning rate is:

[0091] α(t)=α(0)×(1-t / T)

[0092] Where T is the total number of iterations set.

[0093] The calculation formula used to update the update neighborhood function is:

[0094]

[0095] where σ t is the radius of the winning neighborhood at the tth iteration, D is the Euclidean distance between the jth neuron in the winning neighborhood and the winning neuron, and:

[0096] σ t =σ0exp(-t / T)

[0097] In the above formula, σ0 is the initial radius of the winning neighborhood.

[0098] In the above step S207, four damage types can be set first, namely, no deformation, deformation but no damage, damage but damage diameter is less than a set value, and damage diameter is greater than a set value, and the clustering result corresponding to each loss type is set. After the clustering result of the input data of the corrosion point is obtained, the damage type corresponding to the clustering result can be queried, and the damage type is the damage type of the corrosion point of the measured pipeline.

[0099] Since the location of the corrosion point is most likely to be damaged after the tested pipeline is impacted, after the damage type of each corrosion point of the tested pipeline is obtained in sequence through the above method, the damage type of each corrosion point is first normalized, and then the damage type of each corrosion point after normalization is input into the BP neural network sub-model. After receiving the damage type of each corrosion point on the tested pipeline, the BP neural network sub-model can predict the minimum amount of explosives required to cause damage to the tested pipeline when the explosives are detonated at a specific location based on the loss type of each corrosion point.

[0100] In this embodiment, the maximum output value of the SOM neural network is y max , the minimum output value is y min , the i-th data in the output data is y i , then after normalization:

[0101]

[0102] In this embodiment, the BP neural network sub-model includes an input layer, an output layer and 4 hidden layers, wherein the hidden layer includes 12 neurons and uses the Sigmoid logarithmic function as the activation function. The BP neural network sub-model sets the minimum amount of explosives that can cause damage to the pipeline under test when the type of explosives is detonated at a preset position according to the damage type of each corrosion point on the pipeline under test, including:

[0103] Firstly, the BP neural network sub-model is used to predict the corresponding relationship between the intensity of the shock wave emitted at the set position and the damage area caused to each corrosion point on the measured pipeline according to the damage type of each corrosion point on the measured pipeline.

[0104] Then, based on the corresponding relationship between the shock wave intensity and the damaged area, the minimum intensity of the shock wave required to be generated at a specific position in order to damage the tested pipeline is determined;

[0105] Next, according to the preset type of explosive, the amount of explosive required to generate the above-mentioned minimum intensity shock wave is calculated, which is the minimum amount of explosive that causes damage to the measured pipeline when a specific type of explosive is detonated at a specific position;

[0106] In this embodiment, there is the following corresponding relationship between the volume of the explosive and the intensity of the generated shock wave:

[0107]

[0108] Among them, V is the volume of the explosive, A, B, R1, R2 and ω are the matching constants of the explosive, E is the energy of the explosive, and A, B, R1, R2, ω and E can all be obtained according to the type of explosive.

[0109] In this embodiment, the matching constants of the set type of explosive are as follows: A is 209.7 GPa, B is 3.5 GPa, R1 is 5.76, R2 is 1.29, ω is 0.39, and E is 6.0 J·m -3 .

[0110] Finally, the obtained minimum amount of explosives is denormalized.

[0111] In this embodiment, it is assumed that the minimum amount of explosives required to cause damage to the measured pipeline when the explosives are detonated at a specific position according to the BP neural network sub-model is Z, and the minimum amount of explosives obtained after denormalization of the minimum amount of explosives Z is K p ,but

[0112] K p =x min +(Z-1)(x max -x min )

[0113] According to the above content, the pipeline detection equipment of this embodiment is provided with a traveling device and a detection device on the equipment housing. After the pipeline detection equipment is assembled into the pipeline to be detected, the pipeline detection equipment can be driven to travel in the pipeline to be detected by the traveling device, and the corrosion data of the pipeline to be detected can be detected by the detection device during the traveling process, thereby performing a comprehensive detection on the pipeline to be detected, and improving the reliability and comprehensiveness of the corrosion data monitoring of the pipeline to be detected. The data processing device of the pipeline detection equipment can use a pre-trained SOM-BP neural network model to predict the minimum amount of explosives that can cause damage to the pipeline to be detected when the set type of explosives is detonated at a preset position according to the corrosion data, pipeline type, environmental data and explosion data of the pipeline to be detected; since the SOM-BP neural network model is a composite neural network model based on the SOM neural network and the BP neural network, it can improve the accuracy of the prediction of the minimum amount of explosives compared with the method of using a single neural network model.

[0114] In some embodiments of the present invention, Figure 5As shown, a plurality of traveling devices 3 are arranged on the equipment housing 1, wherein a portion of the traveling devices 3 are arranged at a first end in the length direction of the equipment housing 1 and are evenly distributed around the circumference of the equipment housing 1 at the first end; another portion of the traveling devices 3 are arranged at a second end in the length direction of the equipment housing 1 and are evenly distributed around the circumference of the equipment housing 1 at the second end.

[0115] In this embodiment, the traveling device 3 includes a power mechanism, a support structure and a roller 32, wherein in this embodiment, the support structure includes a first support member 311 and a second support member 312 arranged on the equipment housing 1, and the first support member 311 and the second support member 312 form a U-shaped structure; the roller 32 is assembled at a position between the first support member 311 and the second support member 312, so that the roller 32 is assembled to the equipment housing 1 through the support structure 31.

[0116] In this embodiment, the power mechanism of the traveling device 3 can adopt a DC motor, which is connected to the corresponding roller 32 and can drive the corresponding roller 32 to rotate; after the pipeline detection equipment is assembled into the pipeline to be detected, each roller 32 contacts the inner wall of the pipeline to be detected, and each power mechanism can drive each roller 32 to move the pipeline detection equipment inside the pipeline to be detected, thereby controlling the pipeline detection equipment to move to different positions of the pipeline to be detected, so as to detect the corrosion data of the pipeline to be detected at different positions.

[0117] Through the technical solution of this embodiment, the pipeline detection equipment can be driven to move inside the pipeline under test by controlling the roller 32 on the traveling device 3 to detect the corrosion data of each position of the pipeline under test, thereby performing a comprehensive detection of the pipeline under test, thereby improving the reliability of the detection of the pipeline under test.

[0118] In the above embodiment, the traveling device 3 uses rollers to drive the pipeline detection equipment to move; in other application scenarios, the traveling device 3 can use other implementation methods, such as using a robotic arm to drive the pipeline detection equipment to crawl inside the pipeline under test.

[0119] In some embodiments of the present invention, the device housing 1 is a cylindrical structure, and a plurality of through holes are provided on the cylindrical structure, each through hole corresponds to a traveling device 3, and the supporting structures on each traveling device 3 are movably arranged in each through hole.

[0120] like Figure 6As shown, in this embodiment, a central body 5 and a plurality of elastic expansion members 51 are further provided inside the device housing 1, wherein the elastic expansion members 51 may be springs. One end of each elastic expansion member 51 is connected to the central body 5, and the other end is connected to the corresponding support structure 31. After the device housing 1 is assembled inside the measured pipeline, the roller 32 on the traveling device 3 can be made to fit the inner wall of the measured pipeline under the action of the elastic expansion member 51.

[0121] Through the technical solution of this embodiment, the roller 32 can be fitted with the inner wall of the measured pipe through the elastic telescopic member 51, which can not only improve the reliability of the assembly of the pipeline detection equipment in the measured pipe, but also increase the friction between the roller 32 and the inner wall of the measured pipe, so that the pipeline detection equipment can move stably inside the measured pipe, thereby improving the reliability of the detection of the measured pipe.

[0122] In some embodiments of the present invention, Figure 7 As shown, the detection device 2 includes an ultrasonic array, each of which includes a plurality of ultrasonic probes 211, and each ultrasonic probe 211 is evenly distributed around the circumference of the detection housing 1. Each ultrasonic probe 211 can detect the pipeline under test through ultrasonic signals to obtain corrosion data of the pipeline under test.

[0123] After the pipeline under test is corroded, a material interface will be formed on the pipeline under test; the ultrasonic probe 211 includes an ultrasonic generator and an ultrasonic receiver, wherein the ultrasonic generator can send an ultrasonic signal to the pipeline under test, and the ultrasonic signal will be reflected to form a reflected signal wave when it reaches the material interface; after receiving the reflected signal wave, the ultrasonic receiver can calculate the distance between the pipeline detection device and the material interface according to the time difference between sending the ultrasonic signal and receiving the reflected signal wave, and the propagation speed of the ultrasonic signal in the pipeline under test. Since the material interface is formed after the pipeline under test is corroded, the difference between the standard distance and the distance can be calculated after the distance between the pipeline detection device and the material interface. If the difference is 0, it is determined that the detected position is not corroded; if the difference is greater than 0, it is determined that the detected position is corroded, and the difference is used as the corrosion thickness of the pipeline under test. Adjacent corroded positions are divided into the same corrosion point to obtain the position of the corrosion point, and the maximum corrosion thickness in the corrosion point is used as the corrosion thickness of the corrosion point, thereby realizing the detection of the measured thickness.

[0124] Through the technical solution of this embodiment, an ultrasonic array can be used to detect the corrosion data of the measured pipeline in a non-contact manner, which can not only improve the accuracy of the corrosion data detection of the measured pipeline, but also can detect different positions of the measured pipeline as the pipeline detection equipment moves, so as to improve the convenience of detecting the measured pipeline.

[0125] In some embodiments of the present invention, Figure 8 As shown, the detection device 2 includes multiple ultrasonic arrays, and the pipeline detection device can be controlled to move in the pipeline to be detected, so that each ultrasonic array detects the same position of the pipeline to be detected, so as to obtain multiple corrosion thicknesses of the position detected by each ultrasonic array; and then the average value of the multiple corrosion thicknesses is used as the corrosion thickness of the position.

[0126] Through the technical solution of this embodiment, multiple ultrasonic arrays can be used to perform multiple detections on the same position of the pipeline to be detected, thereby improving the accuracy and reliability of the detection of the pipeline to be detected.

[0127] In some embodiments of the present invention, Fig. 9 As shown, the detection device 2 includes a pulsed eddy current array, which includes a plurality of pulsed eddy current probes 221, and each pulsed eddy current probe 221 is evenly distributed around the device housing 1. Each pulsed eddy current probe 221 can detect the pipeline under test through a pulse signal to obtain corrosion data of the pipeline under test.

[0128] In this embodiment, each pulsed eddy current probe 221 includes an excitation coil and a receiving coil, wherein the number of the excitation coils is two, each of which has 200 turns, and contains ferrite inside to ensure the fixed direction and strength of the magnetic field; there are two receiving coils, each of which has 400 turns. The excitation coil is used to generate an electromagnetic interference signal to form an eddy current on the measured pipeline, and the receiving coil can detect the eddy current on the measured pipeline and generate an induced voltage signal, and then judge the corrosion thickness of the measured pipeline by the attenuation change of the voltage signal.

[0129] In this embodiment, the pulsed eddy current array does not need to be in direct contact with the measured pipeline, and is suitable for various metal measured pipelines with complex shapes and large sizes. It is more accurate in detecting the corrosion degree of thin-walled metal measured pipelines and the corrosion degree of the metal pipeline anti-corrosion layer, and will not cause damage to the measured pipeline. The pulsed eddy current probe 221 is very sensitive to tiny eddy current signals and can accurately detect tiny corrosion defects in the measured pipeline. In addition, the pulsed eddy current probe 221 is insensitive to electromagnetic interference in the environment, and can intelligently identify and eliminate interference factors at the detection site to ensure the accuracy of the detection results.

[0130] In some embodiments of the present invention, Fig.10As shown, the detection device 2 includes an ultrasonic array and a pulsed eddy current array, wherein the ultrasonic array includes a plurality of ultrasonic probes 211, and each ultrasonic probe 211 is evenly distributed around the circumference of the detection shell 1; the pulsed eddy current array includes a plurality of pulsed eddy current probes 221, and each pulsed eddy current probe 221 is evenly distributed around the device shell 1. After obtaining the corrosion thickness detected by the ultrasonic array and the corrosion thickness detected by the pulsed eddy current array, the actual corrosion data of the tested pipeline is obtained by weighted summation.

[0131] For example, assuming that the ultrasonic array detects that the corrosion thickness at a location on the tested pipeline is H1, and the pulsed eddy current array detects that the corrosion thickness at the location is H2, then the actual corrosion thickness H at the location is:

[0132] H=a1H1+a2H2

[0133] Where a1 is the weight coefficient of the ultrasonic array, a2 is the weight coefficient of the pulsed eddy current array, and

[0134] a1+a2=1

[0135] Through the technical solution of this embodiment, the detection data of the ultrasonic array and the pulsed eddy current array can be fused by weighted summation to obtain the corrosion data of the measured pipeline, thereby improving the accuracy and reliability of the detection of the measured pipeline.

[0136] In some embodiments of the present invention, Fig.11 As shown, a data output module 6 is also provided on the device housing 1 of the pipeline detection device. The data output module 6 is connected to the data processing module 4 and can exchange information with the data processing module 4 .

[0137] In this embodiment, after predicting the shock wave bearing capacity of the tested pipeline, the data processing module 4 of the pipeline detection equipment can send the shock wave bearing capacity to the data output module 6, and the data output module 6 outputs the shock wave bearing capacity of the tested pipeline to the user.

[0138] In this embodiment, the data output module 6 can use a display device such as a display screen, and after obtaining the bearing capacity of the tested pipeline, the bearing capacity is displayed on the display device. Alternatively, the data output module 6 can use a data sending device, and the data sending device can use a wireless communication device such as a Bluetooth, USB interface or a wireless communication device, and the data sending device can establish a communication connection with a terminal device such as a mobile phone or a tablet computer; after obtaining the shock wave bearing capacity of the tested pipeline, the data output module 6 can send the shock wave bearing capacity of the tested pipeline to the terminal device, so that the user can view the shock wave bearing capacity of the tested pipeline on the terminal device.

[0139] The embodiment of the present invention also provides a method for predicting the ability of a pipeline to withstand shock waves. The method is applied to the pipeline detection equipment provided in the above embodiments, and the process of the method is as follows: Figure 2 Since the prediction method has been introduced in detail in the above embodiments, no further description will be given in this embodiment to avoid redundancy.

[0140] In some embodiments of the present invention, the method for obtaining a pre-trained SOM-BP neural network model includes:

[0141] First, obtain the training data set and the initialized SOM-BP neural network model;

[0142] Then, the mean square error function is used as the loss function, and the gradient descent optimization algorithm is used to train the initialized SOM-BP neural network model according to the training data set to obtain a pre-trained SOM-BP neural network model.

[0143] In this embodiment, multiple groups of pipeline types, corrosion data, explosion data and actual pipeline damage areas can be obtained, and 80% of the groups of pipeline types, explosion data and actual pipeline damage areas are used as training data sets, and the remaining pipeline types, explosion data and actual pipeline damage areas are used as test data sets.

[0144] The BP neural network sub-model in the SOM-BP neural network model can predict the intensity of the shock wave generated at a specific position and the damage area of ​​each corrosion position on the measured pipeline according to the damage type of each corrosion point on the measured pipeline. Therefore, in the process of training the SOM-BP neural network model, the accuracy of the SOM-BP neural network model can be determined according to the damage area of ​​each corrosion point on the measured pipeline.

[0145] Based on the above reasons, in the process of training the SOM-BP neural network model in this embodiment, the loss function used is

[0146]

[0147] Among them, y i is the ith actual damage area, is the i-th predicted damage area, m is the number of damages, and F(t) is the loss function value of the SOM-BP neural network model after the t-th iterative update.

[0148] During the training of the SOM-BP neural network model, if the loss function value of the SOM-BP neural network model is less than the set loss threshold, the training ends, and a test data set is used to test whether the accuracy of the trained SOM-BP neural network model is greater than the preset accuracy. The test method includes:

[0149] The pipeline type, corrosion data and explosion data of the pipeline in the test data set test are input into the trained SOM-BP neural network model to obtain the predicted damage area of ​​the pipeline; then it is determined whether the difference between the preset damage area and the corresponding actual loss area is less than the set difference threshold; if so, it is determined that the accuracy of the SOM-BP neural network model is greater than the preset accuracy; if not, it is determined that the accuracy of the SOM-BP neural network model is not greater than the preset accuracy, and the SOM-BP neural network model is retrained.

[0150] If the loss function value of the SOM-BP neural network model is greater than the set loss value, the gradient descent optimization algorithm is used to update the weights of each neuron in the hidden layer of the BP neural network sub-model and the threshold of each iteration.

[0151] In this embodiment, the calculation formula used in the process of updating the weight of each neuron in the hidden layer of the BP neural network sub-model is:

[0152] V ij (t+1)=V ij (t)+β(t)×F(t)×y i (t)

[0153] Among them, V ij (t) is the jth component vector in the weight vector of the i-th neuron in the hidden layer after the t-th iterative update, and β(t) is the learning rate after the t-th iterative update.

[0154] In the process of updating the iteration threshold, the calculation formula used is:

[0155]

[0156] Among them, θ(t) is the threshold vector after the t-th iteration update.

[0157] The output value of the SOM-BP neural network model can be calculated by the weight of each neuron and the above threshold vector. The calculation formula used is:

[0158]

[0159] Among them, Z j (t) is the jth component of the minimum amount of explosives output by the SOM-BP neural network model after the tth iteration update. After obtaining the components of the minimum amount of explosives after the tth iteration update, the modulus value of the minimum amount of explosives after the tth iteration update can be calculated according to the components, and the modulus value is used as the value of the minimum amount of explosives after the tth iteration update.

[0160] Through the technical solution of this embodiment, in the process of training the SOM-BP neural network model, the mean square error function is used as the loss function to determine whether the trained SOM-BP neural network model has converged, thereby improving the efficiency of training the SOM-BP neural network model and the accuracy of the SOM-BP neural network model. Furthermore, the technical solution of this embodiment uses a gradient descent algorithm to update the weights of each neuron in the hidden layer of the BP neural network sub-model to improve the convergence speed of the SOM-BP neural network model and improve the work efficiency of training the SOM-BP neural network model.

[0161] The flow chart provided by the present embodiment is not intended to indicate that the operation of the method will be performed in any particular order, or that all operations of the method are included in all every case. In addition, the method may include additional operations. Within the scope of the technical ideas provided by the present embodiment method, additional changes can be made to the above method.

[0162] At this point, those skilled in the art should recognize that, although multiple exemplary embodiments of the present invention have been shown and described in detail herein, many other variations or modifications that conform to the principles of the present invention can still be directly determined or derived based on the content disclosed in the present invention without departing from the spirit and scope of the present invention. Therefore, the scope of the present invention should be understood and recognized as covering all these other variations or modifications.

Claims

1. A method for predicting the ability of a pipeline to withstand shock waves, characterized in that: include: Acquiring the corrosion data, wherein the corrosion data includes a pipeline type, environmental data, and explosion data of the pipeline under test; Get the pre-trained SOM-BP neural network model; constructing input data of the SOM-BP neural network model according to one or more of the corrosion data, pipeline type, environmental data and explosion data; The input data is input into the SOM-BP neural network model to predict the minimum amount of explosives that can cause damage to the corroded pipeline when a set type of explosive is detonated at a preset position.

2. The method for predicting the pipeline's ability to withstand shock waves according to claim 1, characterized in that: The step of obtaining the corrosion data comprises: The detection data of a plurality of ultrasonic probe groups are obtained, and the corrosion data is obtained according to an average value of each of the detection data.

3. The method for predicting the pipeline's ability to withstand shock waves according to claim 1, characterized in that: The step of obtaining the corrosion data comprises: The detection data of at least one pulsed eddy current probe group is acquired, and the corrosion data is acquired according to the detection data.

4. The method for predicting the pipeline's ability to withstand shock waves according to claim 1, characterized in that: The step of obtaining the corrosion data comprises: Acquire first detection data of the ultrasonic probe group and second detection data of the pulsed eddy current probe group; A weighted sum is performed on the first detection data and the second detection data to obtain the corrosion data.

5. The method for predicting the pipeline's ability to withstand shock waves according to claim 1, characterized in that: After the step of predicting the minimum amount of explosives that can cause damage to the corroded pipeline when the explosives of the set type are detonated at the preset position, the method further includes: The minimum amount of explosive is sent to the user.

6. The method for predicting the pipeline's ability to withstand shock waves according to claim 1, characterized in that: The step of obtaining the pre-trained SOM-BP neural network model includes: Obtain training data set and initialized SOM-BP neural network model; The initialized SOM-BP neural network model is trained according to the training data set by using a mean square error function as a loss function and a gradient descent optimization algorithm to obtain the pre-trained SOM-BP neural network model.

7. A pipeline detection device, characterized in that: include: A device housing, which is used to be assembled inside the measured pipeline; A traveling device, which is arranged outside the device housing and is used to drive the pipeline detection device to move inside the detected pipeline; A detection device, which is arranged on the equipment housing and is used to detect corrosion data of the tested pipeline; A data processing device is connected to the detection device and is configured to execute the method for predicting the pipeline's ability to withstand shock waves as described in any one of claims 1 to 6.

8. The pipeline detection device according to claim 7, characterized in that: The traveling device comprises a power mechanism, a supporting structure and a roller, wherein the supporting structure is used to assemble the roller on the device housing, and the power mechanism is connected to the roller by transmission, so that the pipeline detection device moves inside the detected pipeline by driving the roller.

9. The pipeline detection device according to claim 8, characterized in that: The supporting structure is movably arranged on the device housing, a central body is arranged inside the device housing, and the central body is connected to the supporting structure through an elastic telescopic member.

10. The pipeline detection device according to claim 7, characterized in that: The detection device includes an ultrasonic probe group and / or a pulsed eddy current array, wherein: The ultrasonic probe group includes a plurality of ultrasonic probes, and the plurality of ultrasonic probes are evenly distributed around the circumference of the device housing; The pulsed eddy current array includes a plurality of pulsed eddy current probes, and the plurality of pulsed eddy current probes are evenly distributed around the circumference of the device housing.

Images