A nondestructive testing method for fiber distribution of steel fiber shield tunnel segments

Abstract

The present invention discloses a nondestructive detection method for fiber distribution of steel fiber shield tunnel segments, including: obtaining the acoustic wave signal and the resonance signal of the steel fiber shield tunnel segment to be detected; preprocessing the acoustic wave signal and the resonance signal to obtain the preprocessed acoustic wave signal and the resonance signal; analyzing the preprocessed acoustic wave signal and the resonance signal based on the steel fiber distribution nondestructive detection model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and outputting the corresponding quality detection result. The present invention analyzes and classifies the acoustic signals of the steel fiber shield tunnel segment through a deep learning network model, and then obtains the steel fiber content of each point, thereby reducing the detection cost of the steel fiber tunnel segment and realizing the function of nondestructive detection of fiber distribution of the shield tunnel pipe in the tunnel scene.

Description

Technical Field

[0001] The present invention relates to the technical field of steel fiber detection, in particular to a non-destructive detection method for fiber distribution of steel fiber shield tunnel segments. Background Art

[0002] The scale, number, and total length of tunnel construction in China are increasing annually. Shield tunneling has been widely used in urban subways and cross-river and sea tunnels, particularly in conditions with complex geological conditions and high water levels. Tunnel linings are subject to significant bending loads during operation, including traffic loads, soil loads, water loads, and seismic loads. Traditional reinforced concrete tunnel linings are susceptible to cracking and spalling under concentrated loads, seriously affecting the mechanical properties and durability of the segments. During operation, stray currents can cause corrosion of the steel reinforcement, compromising the durability of the structure. Cathodic protection is typically used to prevent corrosion, further increasing project costs. The use of steel fiber tunnel segments can alleviate these problems. However, while conventional reinforced concrete has a complex quality inspection process, steel fiber tunnel segments typically require drilling and sampling, and specialized testing equipment. This not only results in high costs for testing steel fiber tunnel segments, but also limits their application in tunnel inspection scenarios.

[0003] Therefore, the existing technology needs to be improved. Summary of the Invention

[0004] The technical problem to be solved by the present invention is that, in response to the defects of the existing technology, the present invention provides a non-destructive detection method for fiber distribution of steel fiber shield tunnel segments to solve the problems of high cost and inability to realize detection in specific scenarios of the existing steel fiber tunnel segment detection method.

[0005] The technical solutions adopted by the present invention to solve the technical problems are as follows:

[0006] In a first aspect, the present invention provides a method for nondestructive detection of fiber distribution in steel fiber shield tunnel segments, comprising:

[0007] Obtain the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be tested;

[0008] Preprocessing the acoustic wave signal and the resonance signal to obtain preprocessed acoustic wave signal and resonance signal;

[0009] The pre-processed acoustic wave signal and the resonance signal are analyzed based on the steel fiber distribution non-destructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and output the corresponding quality test results.

[0010] In one implementation, the step of obtaining the acoustic wave signal and the resonance signal of the steel fiber shield tunnel segment to be inspected includes:

[0011] Obtaining the resonance frequency and hammer gravity parameters corresponding to the steel fiber shield tunnel segment to be tested;

[0012] According to the resonant frequency and the hammer gravity parameter, the intelligent hammer is controlled to sequentially strike a plurality of inspection points on the steel fiber shield tunnel segment to be inspected;

[0013] The intelligent hammer collects the acoustic wave signals and resonance signals generated after each detection point is struck.

[0014] In one implementation, preprocessing the acoustic wave signal and the resonance signal to obtain the preprocessed acoustic wave signal and resonance signal includes:

[0015] De-noising and filtering are performed on the acoustic wave signal and the resonance signal, and amplitude normalization is performed on the filtered signals to obtain the pre-processed acoustic wave signal and resonance signal.

[0016] In one implementation, the steel fiber distribution nondestructive testing model is a model trained based on a 2D CNN-LSTM hybrid framework, and the steel fiber distribution nondestructive testing model includes: a 2D-CNN network and an LSTM network; wherein, the 2D-CNN network includes: an input layer, three convolutional layers, three pooling layers, a fully connected layer and an output layer; the LSTM network includes: an input layer, multiple hidden LSTM layers and an output layer.

[0017] In one implementation, the pre-processed acoustic wave signal and the resonance signal are analyzed based on the steel fiber distribution non-destructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and output the corresponding quality test results, including:

[0018] Inputting the preprocessed acoustic wave signal and the resonance signal into the 2D-CNN network for analysis, obtaining the spectral characteristics and Mel-frequency cepstral coefficients of the acoustic wave signal, and obtaining the frequency domain characteristics and time-frequency domain characteristics of the resonance signal;

[0019] The data output by the 2D-CNN network is input into the LSTM network. After feature merging and assimilation, the steel fiber distribution data of the corresponding steel fiber shield tunnel segment is obtained. The corresponding quality grade is obtained based on the steel fiber distribution data, and the corresponding quality inspection result is output.

[0020] In one implementation, inputting the pre-processed acoustic wave signal and the resonance signal into the 2D-CNN network for analysis includes:

[0021] The preprocessed acoustic wave signal and resonance signal are sent as matrices to the input layer of the 2D-CNN network for convolution, pooling, and nonlinear transformation processing, and the features obtained after the pooling operation are flattened into a fully connected layer, and the features corresponding to the acoustic wave signal and the features corresponding to the resonance signal are output.

[0022] In one implementation, the data output by the 2D-CNN network is input into the LSTM network, and after feature merging and assimilation, the steel fiber distribution data of the corresponding steel fiber shield tunnel segment is obtained, and the corresponding quality grade is obtained based on the steel fiber distribution data, including:

[0023] The data output by the 2D-CNN network is input into the LSTM network, and after feature merging and assimilation, the steel fiber content of the corresponding steel fiber shield tunnel segment is obtained;

[0024] By calculating the variance of the steel fiber content at each detection point, the corresponding steel fiber distribution data is obtained;

[0025] The steel fiber distribution data is compared with the preset steel fiber content, and the ratio is used as a quality control index. The quality grade corresponding to the steel fiber shield tunnel segment to be tested is evaluated based on the quality control index.

[0026] In a second aspect, the present invention provides a non-destructive detection device for fiber distribution of steel fiber shield tunnel segments, comprising:

[0027] A signal acquisition module is used to obtain the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be tested;

[0028] a signal preprocessing module, configured to preprocess the acoustic wave signal and the resonance signal to obtain preprocessed acoustic wave signals and resonance signals;

[0029] The quality detection module is used to analyze the preprocessed acoustic wave signal and the resonance signal based on the steel fiber distribution nondestructive detection model, obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and output the corresponding quality detection result.

[0030] In a third aspect, the present invention provides a terminal comprising: a processor and a memory, wherein the memory stores a nondestructive detection program for fiber distribution of steel fiber shield tunnel segments, and when the nondestructive detection program for fiber distribution of steel fiber shield tunnel segments is executed by the processor, it is used to implement the operation of the nondestructive detection method for fiber distribution of steel fiber shield tunnel segments as described in the first aspect.

[0031] In a fourth aspect, the present invention also provides a medium, which is a computer-readable storage medium, and which stores a non-destructive detection program for the fiber distribution of steel fiber shield tunnel segments. When the non-destructive detection program for the fiber distribution of steel fiber shield tunnel segments is executed by a processor, it is used to implement the operation of the non-destructive detection method for the fiber distribution of steel fiber shield tunnel segments as described in the first aspect.

[0032] The present invention adopts the above technical solution to achieve the following effects:

[0033] The present invention obtains the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be tested, preprocesses the acoustic wave signal and resonance signal to obtain the preprocessed acoustic wave signal and resonance signal; then analyzes the preprocessed acoustic wave signal and resonance signal based on the steel fiber distribution nondestructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment and output the corresponding quality test results. The present invention analyzes and classifies the acoustic signals of the steel fiber shield tunnel segment through a deep learning network model, and then obtains the steel fiber content at each point, reducing the cost of steel fiber tunnel segment testing and realizing the function of nondestructive testing of shield tunnel fiber distribution in tunnel scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0035] Figure 1 It is a flow chart of the nondestructive detection method for fiber distribution of steel fiber shield tunnel segments in the present invention.

[0036] Figure 2 It is a schematic diagram of the framework of the steel fiber distribution nondestructive testing model in the present invention.

[0037] Figure 3 It is a schematic diagram of the segment knocking points in the present invention.

[0038] Figure 4 It is a structural diagram of the 2D-CNN network in the present invention.

[0039] Figure 5 It is a structural diagram of the LSTM network in the present invention.

[0040] Figure 6 Schematic diagram of signal transmission of LSTM network in the present invention.

[0041] Figure 7It is a functional principle diagram of a terminal in one implementation of the present invention.

[0042] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the present invention more clear and distinct, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0044] Exemplary Methods

[0045] At present, the quality inspection process of ordinary reinforced concrete is complete, while steel fiber tunnel segments usually require drilling sampling and inspection, and require professional inspection equipment for inspection. On the one hand, this leads to high costs for steel fiber tunnel segment inspection, and on the other hand, it limits its application in tunnel inspection scenarios.

[0046] In response to the above technical problems, an embodiment of the present invention provides a non-destructive testing method for fiber distribution of steel fiber shield tunnel segments. The method obtains the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be tested, and pre-processes the acoustic wave signal and resonance signal to obtain the pre-processed acoustic wave signal and resonance signal; and analyzes the pre-processed acoustic wave signal and resonance signal based on the steel fiber distribution non-destructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and outputs the corresponding quality test results. The embodiment of the present invention analyzes and classifies the acoustic signals of the steel fiber shield tunnel segment through a deep learning network model, and then obtains the steel fiber content of each point, thereby reducing the detection cost of the steel fiber tunnel segment and realizing the function of non-destructive testing of the fiber distribution of the shield tunnel pipe in the tunnel scenario.

[0047] like Figure 1 As shown, an embodiment of the present invention provides a method for nondestructive detection of fiber distribution in steel fiber shield tunnel segments, comprising the following steps:

[0048] Step S100: Acquire the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be inspected.

[0049] Typically, steel fiber shield tunnel segments undergo bending, pullout, and seepage tests, as well as three-ring horizontal assembly tests. However, the distribution of steel fibers within a steel fiber tunnel segment directly impacts its performance. Therefore, testing the fiber distribution of steel fiber tunnel segments is crucial for their engineering applications.

[0050] In this embodiment, a smart hammer is used to strike multiple evenly distributed points on the steel fiber shield tunnel segment to be tested to generate acoustic signals and resonance signals, and then the receiver on the smart hammer is used to record these acoustic signals; the acoustic signals are analyzed and classified by a deep learning network model to obtain the steel fiber content of each point, thereby realizing the function of detecting the steel fiber content of n points; finally, by performing variance analysis on the steel fiber content of each point, the function of detecting the steel fiber distribution and segment quality of the steel fiber segment is realized.

[0051] Specifically, in one implementation of this embodiment, step S100 includes the following steps:

[0052] Step S101, obtaining the resonance frequency and hammer gravity parameters corresponding to the steel fiber shield tunnel segment to be tested;

[0053] Step S102, controlling the intelligent hammer to sequentially strike a plurality of inspection points on the steel fiber shield tunnel segment to be inspected according to the resonant frequency and the hammer gravity parameter;

[0054] Step S103: collecting the sound wave signals and resonance signals generated by striking each detection point through the intelligent hammer.

[0055] In this embodiment, the steel fiber shield tunnel segment to be inspected is struck by an intelligent hammer to obtain the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be inspected; specifically, the intelligent hammer is composed of a hammer head, a hammer rod, a sensor element and an output connector; wherein the hammer head, the hammer rod and the output connector are connected in sequence, and the sensor element is not limited to a position; it can be understood that the intelligent hammer can be controlled by a mechanical device (for example, a robotic arm) to achieve precise control of the striking force and striking frequency; of course, the intelligent hammer can also be a striking device with a variable hammer head weight to adjust the striking force on the steel fiber shield tunnel segment to be inspected according to actual detection needs.

[0056] In this embodiment, before testing the steel fiber shield tunnel segment to be tested, it is necessary to set the testing points (i.e., knocking points) of the steel fiber shield tunnel segment to be tested. When actually testing the segment, the following can be selected: Figure 3 The 18 points shown are evenly distributed points with equal horizontal distances.

[0057] After setting the detection points, the corresponding resonant frequency and hammering gravity parameters are obtained according to the material information of the steel fiber shield tunnel segment to be detected and the historical detection parameters; then, according to the resonant frequency and the hammering gravity parameters, the intelligent hammer is controlled by the robotic arm to knock on multiple detection points on the steel fiber shield tunnel segment to be detected in turn; after each knock, the sensor element on the intelligent hammer is used to collect the sound wave signal and resonance signal generated after the knocking of each detection point, and these signals are transmitted to the signal collection intelligent device via the output connector to collect and preprocess the sound wave signal and resonance signal; wherein, the signal collection intelligent device is a device that can record sound wave data and resonance information (for example, a smart phone or a collector).

[0058] like Figure 1 As shown, an embodiment of the present invention provides a method for nondestructive detection of fiber distribution in steel fiber shield tunnel segments, comprising the following steps:

[0059] Step S200 , preprocessing the acoustic wave signal and the resonance signal to obtain preprocessed acoustic wave signal and resonance signal.

[0060] In this embodiment, the collected acoustic wave signals and resonance signals generated after the tapping of each detection point need to undergo a series of preprocessing processes to obtain slice data for the analysis of the steel fiber distribution non-destructive detection model.

[0061] Specifically, in one implementation of this embodiment, step S200 includes the following steps:

[0062] Step S201 , performing denoising and filtering processing on the acoustic wave signal and the resonance signal, and performing amplitude normalization on the filtered signals to obtain the pre-processed acoustic wave signal and resonance signal.

[0063] In this embodiment, the data is sliced ​​during the preprocessing process to obtain slice features at different times; specifically, first, the acoustic wave signal and the resonance signal are denoised and filtered to remove noise signals in these signals; then, the amplitude of the filtered signal is normalized to obtain the acoustic wave signal and resonance signal data in the form of data slices.

[0064] like Figure 1 As shown, an embodiment of the present invention provides a method for nondestructive detection of fiber distribution in steel fiber shield tunnel segments, comprising the following steps:

[0065] Step S300: Analyze the pre-processed acoustic wave signal and the resonance signal based on the steel fiber distribution non-destructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment and output the corresponding quality test result.

[0066] In this embodiment, the pre-processed acoustic wave signal and resonance signal are analyzed by the steel fiber distribution non-destructive testing model to obtain the classification of the steel fiber content at each point; wherein, the steel fiber distribution non-destructive testing model is a model trained based on the 2D CNN-LSTM hybrid deep learning framework; Figure 2 As shown, the steel fiber distribution nondestructive testing model includes: 2D-CNN network and LSTM network; wherein, Figure 4 As shown in , the 2D-CNN network includes: an input layer, three convolutional layers, three pooling layers, a fully connected layer and an output layer; Figure 5 As shown, the LSTM network includes: an input layer, multiple hidden LSTM layers and an output layer.

[0067] In this example, the steel fiber distribution nondestructive testing model was developed through indoor experimental training. Specifically, a large number of concrete cubic test blocks with a fiber content of 0-2% (150*150*150mm) were prepared. The concrete blocks were struck using an intelligent tapping device, and the acoustic and resonance signals from each block were recorded using a collection device to create a training dataset. This training dataset was used to train the 2DCNN-LSTM framework, thereby developing the steel fiber distribution nondestructive testing model.

[0068] After obtaining the trained steel fiber distribution nondestructive testing model, the preprocessed acoustic wave signal and resonant signal are analyzed based on the steel fiber distribution nondestructive testing model to achieve nondestructive testing of the quality of the steel fiber reinforced concrete segment. Compared to traditional steel fiber distribution testing methods, the method in this embodiment uses the steel fiber distribution nondestructive testing model to detect the steel fiber content at each point to indirectly assess the segment quality, without the need for signal conversion or manual feature extraction.

[0069] Specifically, in one implementation of this embodiment, step S300 includes the following steps:

[0070] Step S301: Input the preprocessed acoustic wave signal and the resonance signal into the 2D-CNN network for analysis to obtain the spectral characteristics and Mel-frequency cepstral coefficients of the acoustic wave signal, and obtain the frequency domain characteristics and time-frequency domain characteristics of the resonance signal;

[0071] In step S302, the data output by the 2D-CNN network is input into the LSTM network. After feature merging and assimilation, the steel fiber distribution data of the corresponding steel fiber shield tunnel segment is obtained. The corresponding quality grade is obtained based on the steel fiber distribution data, and the corresponding quality inspection result is output.

[0072] Specifically, in one implementation of this embodiment, the preprocessed acoustic wave signal and resonance signal are input into the 2D-CNN network for analysis, including: sending the preprocessed acoustic wave signal and resonance signal as matrices into the input layer of the 2D-CNN network for convolution, pooling, and nonlinear transformation processing, and flattening the features obtained after the pooling operation into a fully connected layer, and outputting the features corresponding to the acoustic wave signal and the features corresponding to the resonance signal.

[0073] In this embodiment, the 2D-CNN network generates two-dimensional features by convolving a set of two-dimensional kernels. For acoustic signals, the 2D-CNN network processes them to obtain their spectral features and Mel-frequency cepstral coefficients; for resonant signals, the 2D-CNN network processes them to obtain their frequency domain features and time-frequency domain features.

[0074] like Figure 4 As shown, the 2D-CNN network includes an input layer, three convolutional layers, three pooling layers, a fully connected layer and an output layer. The data sets corresponding to the preprocessed acoustic wave signal and the resonance signal are fed into the input layer as matrices and convolved using a 3*3 convolution kernel. After three convolutions, features with 32, 64 and 128 different kernel numbers are obtained respectively. There is a pooling layer after each convolution to reduce the size of the feature map. After each convolution and pooling operation, the rectified linear unit (ReLU) activation function is used to convert the linear transformation into a nonlinear transformation to improve the representation of the CNN network. The features obtained after the last pooling operation are flattened into a fully connected layer to reduce the dimensionality. The fully connected layer is used to calculate the output layer of size 5. The softmax function is applied before the output layer so that the results of the classification model are in a probabilistic form.

[0075] The 2D CNN algorithm process and principle are as follows:

[0076] Take x(i,j) as the input of the 2D CNN and apply the convolution kernel w(i,j) with a size of 3×3 to the signal:

[0077]

[0078] Pooling operation:

[0079]

[0080] where Ω k represents the pooling area with index k, and Represents the output and input features of the l-th max pooling layer with indices k and p.

[0081] Among them, Softmax is a commonly used mathematical function. The Softmax function converts a set of values ​​into a vector representing probability distribution. The formula is as follows:

[0082]

[0083] Where xi is the i-th element in the input vector, and N is the length of the input vector. The Softmax function converts each element in the vector to a value between 0 and 1, and the sum of these values ​​is 1.

[0084] Specifically, in one implementation of this embodiment, the data output by the 2D-CNN network is input into the LSTM network, and after feature merging and assimilation, the steel fiber distribution data of the corresponding steel fiber shield tunnel segment is obtained, and the corresponding quality grade is obtained based on the steel fiber distribution data evaluation, including: inputting the data output by the 2D-CNN network into the LSTM network, and after feature merging and assimilation, the steel fiber content of the corresponding steel fiber shield tunnel segment is obtained; obtaining the corresponding steel fiber distribution data by calculating the variance of the steel fiber content of each detection point; comparing the steel fiber distribution data with the preset steel fiber content, and using the ratio as a quality control index, and obtaining the corresponding quality grade of the steel fiber shield tunnel segment to be detected based on the quality control index evaluation.

[0085] In this embodiment, the LSTM network belongs to the RNN network and is very effective in processing sequence data. Figure 5 As shown, the LSTM network is an architecture consisting of an input layer, multiple hidden LSTM layers, and an output layer. The input layer transmits the data sequence (i.e., the features output by the 2D-CNN network) to the first LSTM layer, and the output layer converts the output of the last LSTM layer into a response sequence. The hidden LSTM layers are composed of identical LSTM units, expanded in the time domain, and mapped sequence to sequence.

[0086] The LSTM unit in the LSTM structure introduces three different gates, namely the forget gate, input gate, and output gate. Figure 6 As shown, for the LSTM unit structure, the output h of time step t t Depends on the network's current long-term memory (c t-1 ), the previous time step (h t-1 ) and the output of the current time step (x t ). The forget gate determines the useful portion of long-term memory, which depends on the previous output and the current input. This is achieved by multiplying the long-term memory with the filter vector generated by the sigmoid activation function. The input gate accepts the same input as the forget gate and adds new information to the network's long-term memory.

[0087]

[0088]

[0089]

[0090]

[0091]

[0092] c t represents the LSTM unit state; W, U and b are parameter matrices and vectors; f t ,i t ,o t is the gate vector, σ ij is the sigmoid function, σ c ,σ z represents the hyperbolic tangent; The operator represents the Hadamard product.

[0093] Similar to the forget gate, the input gate generates a filter vector and multiplies it with the new memory update vector to extract useful information and update the information into the long-term memory. The output gate finally generates a filter that selects the necessary information from the updated new long-term memory and passes it to the next time step.

[0094] Among them, the Sigmoid function is used as an activation function to map any real value to a value between 0 and 1. It has the characteristics of smoothness and differentiability, which makes it have good properties in optimization such as gradient descent. The specific definition is as follows:

[0095]

[0096] Where e is the base of the natural logarithm and x is the input value.

[0097] The tanh function is short for the hyperbolic tangent function, which helps maintain the stability of the gradient when training neural networks. Its form is as follows:

[0098]

[0099] The tanh function maps real values ​​to [-1, 1].

[0100] As an example, when testing steel fiber shield tunnel segments, the following Figure 3At the 18 locations shown, acoustic signals were acquired using an intelligent tapping device, and the acoustic and resonance signals were recorded using an intelligent data collection device. These acoustic and resonance signals were then fed into a 2DCNN-LSTM model to determine the steel fiber content at each location.

[0101] Afterwards, the corresponding steel fiber distribution data is obtained by calculating the variance of the steel fiber content at each detection point; the steel fiber distribution data is compared with the preset steel fiber content, and the ratio is used as a quality control indicator. The quality grade corresponding to the steel fiber shield tunnel segment to be detected is evaluated based on the quality control indicator; wherein, the larger the variance, the worse the steel fiber distribution at the detection point.

[0102] Here is an example where the ratio of the variance to the steel fiber content in the segment design is used as a quality control indicator:

[0103] High quality (uniform distribution): The ratio of the variance to the designed steel fiber content is less than or equal to 0.2, indicating that the steel fiber content is evenly distributed and the quality of the corresponding test points is high.

[0104] Meets the standard (good distribution): The ratio of the variance to the designed steel fiber content is greater than 0.2 and less than or equal to 0.5, indicating that the steel fiber content is well distributed and the quality of the corresponding test points meets the standard.

[0105] Not meeting the standards (poor distribution): The ratio of the variance to the designed steel fiber content is greater than 0.5, indicating that the steel fiber content is poorly distributed and the quality of the corresponding test points does not meet the standards.

[0106] This embodiment evaluates the quality level of the entire steel fiber shield tunnel segment to be inspected by inspecting the quality level of each point, and outputs the entire quality inspection result.

[0107] This embodiment achieves the following technical effects through the above technical solution:

[0108] This embodiment obtains the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be inspected, preprocesses the acoustic wave signal and resonance signal to obtain the preprocessed acoustic wave signal and resonance signal; then analyzes the preprocessed acoustic wave signal and resonance signal based on the steel fiber distribution nondestructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment and output the corresponding quality inspection results. This embodiment analyzes and classifies the acoustic signals of the steel fiber shield tunnel segment using a deep learning network model to obtain the steel fiber content at each point, reducing the cost of steel fiber tunnel segment inspection and realizing the function of nondestructive testing of shield tunnel fiber distribution in tunnel scenarios.

[0109] Exemplary devices

[0110] Based on the above embodiments, the present invention further provides a non-destructive detection device for fiber distribution of steel fiber shield tunnel segments, comprising:

[0111] A signal acquisition module is used to obtain the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be tested;

[0112] a signal preprocessing module, configured to preprocess the acoustic wave signal and the resonance signal to obtain preprocessed acoustic wave signals and resonance signals;

[0113] The quality detection module is used to analyze the preprocessed acoustic wave signal and the resonance signal based on the steel fiber distribution nondestructive detection model, obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and output the corresponding quality detection result.

[0114] This embodiment achieves the following technical effects through the above technical solution:

[0115] This embodiment obtains the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be inspected, preprocesses the acoustic wave signal and resonance signal to obtain the preprocessed acoustic wave signal and resonance signal; then analyzes the preprocessed acoustic wave signal and resonance signal based on the steel fiber distribution nondestructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment and output the corresponding quality inspection results. This embodiment analyzes and classifies the acoustic signals of the steel fiber shield tunnel segment using a deep learning network model to obtain the steel fiber content at each point, reducing the cost of steel fiber tunnel segment inspection and realizing the function of nondestructive testing of shield tunnel fiber distribution in tunnel scenarios.

[0116] Based on the above embodiment, the present invention further provides a terminal, whose principle block diagram can be shown as follows: Figure 7 shown.

[0117] The terminal includes: a processor, memory, interface, display screen and communication module connected via a system bus; wherein the processor of the terminal is used to provide computing and control capabilities; the memory of the terminal includes a storage medium and an internal memory; the storage medium stores an operating system and computer programs; the internal memory provides an environment for the operation of the operating system and computer programs in the storage medium; the interface is used to connect to external devices; the display screen is used to display corresponding information; and the communication module is used to communicate with a cloud server or other devices.

[0118] When the computer program is executed by a processor, it is used to implement the operation of a non-destructive detection method for fiber distribution of steel fiber shield tunnel segments.

[0119] It will be understood by those skilled in the art that Figure 7The principle block diagram shown in the figure is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal to which the solution of the present invention is applied. The specific terminal may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0120] In one embodiment, a terminal is provided, which includes: a processor and a memory, wherein the memory stores a non-destructive detection program for fiber distribution of steel fiber shield tunnel segments, and when the non-destructive detection program for fiber distribution of steel fiber shield tunnel segments is executed by the processor, it is used to implement the operation of the above-mentioned non-destructive detection method for fiber distribution of steel fiber shield tunnel segments.

[0121] In one embodiment, a storage medium is provided, wherein the storage medium stores a nondestructive detection program for fiber distribution of steel fiber shield tunnel segments. When the nondestructive detection program for fiber distribution of steel fiber shield tunnel segments is executed by a processor, it is used to implement the operation of the above-mentioned nondestructive detection method for fiber distribution of steel fiber shield tunnel segments.

[0122] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile storage medium. When executed, the computer program can include the processes in the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include both non-volatile and volatile memory.

[0123] In summary, the present invention provides a method for non-destructive testing of fiber distribution of steel fiber shield tunnel segments, comprising: obtaining an acoustic wave signal and a resonance signal of a steel fiber shield tunnel segment to be tested; pre-processing the acoustic wave signal and the resonance signal to obtain a pre-processed acoustic wave signal and a resonance signal; analyzing the pre-processed acoustic wave signal and the resonance signal based on a non-destructive testing model for steel fiber distribution to obtain steel fiber distribution data of the corresponding steel fiber shield tunnel segment, and outputting a corresponding quality test result. The present invention analyzes and classifies the acoustic signals of steel fiber shield tunnel segments through a deep learning network model, thereby obtaining the steel fiber content at each point, reducing the cost of steel fiber tunnel segment testing and realizing the function of non-destructive testing of shield tunnel tube fiber distribution in a tunnel scenario.

[0124] It should be understood that the application of the present invention is not limited to the above examples. For those skilled in the art, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A nondestructive detection method for fiber distribution of steel fiber shield tunnel segments, characterized in that: include: Obtain the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be tested; Preprocessing the acoustic wave signal and the resonance signal to obtain preprocessed acoustic wave signal and resonance signal; Analyze the preprocessed acoustic wave signal and the resonance signal based on the steel fiber distribution nondestructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and output the corresponding quality test result; The steel fiber distribution nondestructive testing model is a model obtained by training based on a 2D CNN-LSTM hybrid framework, and the steel fiber distribution nondestructive testing model includes: a 2D-CNN network and an LSTM network; wherein the 2D-CNN network includes: an input layer, three convolutional layers, three pooling layers, a fully connected layer and an output layer; the LSTM network includes: an input layer, multiple hidden LSTM layers and an output layer; The pre-processed acoustic wave signal and the resonance signal are analyzed based on the steel fiber distribution non-destructive testing model to obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and output the corresponding quality test results, including: Inputting the preprocessed sound wave signal and the resonance signal into the 2D-CNN network for analysis, obtaining the frequency spectrum characteristics and Mel-frequency cepstrum coefficients of the sound wave signal, and obtaining the frequency domain characteristics and time-frequency domain characteristics of the resonance signal; The data output by the 2D-CNN network is input into the LSTM network, and the corresponding steel fiber distribution data of the steel fiber shield tunnel segment is obtained after feature merging and assimilation. The corresponding quality grade is obtained according to the steel fiber distribution data, and the corresponding quality inspection result is output.

2. The nondestructive testing method for fiber distribution of steel fiber shield tunnel segments according to claim 1 is characterized in that: The method of obtaining the acoustic wave signal and the resonance signal of the steel fiber shield tunnel segment to be detected includes: Obtaining the resonance frequency and hammer gravity parameters corresponding to the steel fiber shield tunnel segment to be tested; According to the resonant frequency and the hammer gravity parameter, the intelligent hammer is controlled to knock on multiple detection points on the steel fiber shield tunnel segment to be detected in sequence; The intelligent hammer collects the sound wave signals and resonance signals generated after each detection point is struck.

3. The nondestructive testing method for fiber distribution of steel fiber shield tunnel segments according to claim 1 is characterized in that: The preprocessing of the sound wave signal and the resonance signal to obtain the preprocessed sound wave signal and the resonance signal comprises: The sound wave signal and the resonance signal are subjected to denoising and filtering, and the amplitude of the filtered signals is normalized to obtain the pre-processed sound wave signal and the resonance signal.

4. The nondestructive testing method for fiber distribution of steel fiber shield tunnel segments according to claim 1 is characterized in that: The step of inputting the preprocessed acoustic wave signal and the resonance signal into the 2D-CNN network for analysis comprises: The preprocessed sound wave signal and resonance signal are sent as matrices to the input layer of the 2D-CNN network for convolution, pooling and nonlinear transformation processing, and the features obtained after the pooling operation are flattened into a fully connected layer, and the features corresponding to the sound wave signal and the features corresponding to the resonance signal are output.

5. The nondestructive testing method for fiber distribution of steel fiber shield tunnel segments according to claim 1 is characterized in that: The data output by the 2D-CNN network is input into the LSTM network, and after feature merging and assimilation, the steel fiber distribution data of the corresponding steel fiber shield tunnel segment is obtained, and the corresponding quality grade is obtained according to the steel fiber distribution data evaluation, including: The data output by the 2D-CNN network is input into the LSTM network, and the steel fiber content of the corresponding steel fiber shield tunnel segment is obtained after feature merging and assimilation; By calculating the variance of the steel fiber content at each detection point, the corresponding steel fiber distribution data is obtained; The steel fiber distribution data is compared with the preset steel fiber content, and the ratio is used as a quality control index. The quality grade corresponding to the steel fiber shield tunnel segment to be tested is evaluated based on the quality control index.

6. A nondestructive detection device for fiber distribution of steel fiber shield tunnel segments, used to implement the nondestructive detection method for fiber distribution of steel fiber shield tunnel segments as claimed in any one of claims 1 to 5, characterized in that: include: A signal acquisition module is used to acquire the acoustic wave signal and resonance signal of the steel fiber shield tunnel segment to be tested; A signal preprocessing module, used for preprocessing the acoustic wave signal and the resonance signal to obtain a preprocessed acoustic wave signal and a resonance signal; The quality detection module is used to analyze the preprocessed acoustic wave signal and the resonance signal based on the steel fiber distribution nondestructive detection model, obtain the corresponding steel fiber distribution data of the steel fiber shield tunnel segment, and output the corresponding quality detection result.

7. A terminal, characterized in that: include: A processor and a memory, wherein the memory stores a nondestructive detection program for fiber distribution of steel fiber shield tunnel segments, and when the nondestructive detection program for fiber distribution of steel fiber shield tunnel segments is executed by the processor, it is used to implement the operation of the nondestructive detection method for fiber distribution of steel fiber shield tunnel segments as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a nondestructive detection program for fiber distribution of steel fiber shield tunnel segments, which, when executed by a processor, is used to implement the operation of the nondestructive detection method for fiber distribution of steel fiber shield tunnel segments as described in any one of claims 1 to 5.

Images