Method and system for detecting strength of building concrete
By laying detection points in building concrete, using ultrasonic rebound method and graph neural network model, combining data processing and image analysis, the detection point spatial distribution topology matrix is constructed, waveform features and topological features are extracted, and concrete strength distribution feature matrix is generated. The destructive and inefficient problems of traditional concrete strength detection methods are achieved, and intelligent strength detection and efficient quality control of building concrete are realized.
Patent Information
- Application Number
- CN202411066628.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional concrete strength detection methods are time-consuming and labor-intensive, may cause damage to the structure, and cannot fully reflect the concrete quality of the entire structure, and lack non-destructive and efficient.
Multiple intensity detection points are arranged in building concrete, and the position information of the detection point and the ultrasonic rebound curve are obtained by using ultrasonic rebound method. Through the graph neural network model, the spatial distribution topology matrix of the detection point is constructed and the waveform features are extracted through the graph neural network model combined with data processing and image analysis. and topological features, generate a concrete strength distribution feature matrix, and input the decoder for strength measurement.
It realizes intelligent strength detection of building concrete, avoids the destructiveness and inefficiency of traditional methods, provides more effective means and support, and provides more accurate and efficient inspection methods for building project quality control.
Smart Images

Figure CN118937481A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of building concrete, and in particular, to a strength detection method and system for building concrete. Background Art
[0002] In the construction industry, architectural concrete is an artificial stone made of cement, sand, aggregate and water in a certain proportion. It is widely used in structural components, foundations, walls and other parts of construction projects. The performance of concrete directly affects the safety, durability and bearing capacity of buildings. The strength of concrete is a key indicator to measure its performance, usually referring to its ability to resist compression, tension or flexure. The detection of concrete strength is crucial to ensure the structural bearing capacity and safety, as well as to evaluate the durability of buildings and predict their service life. It is used to guide the engineering design and construction process and ensure the safety and reliability of building structures.
[0003] However, traditional concrete strength testing methods, such as core sampling and pressure testing, are not only time-consuming and labor-intensive, but also require sampling from the building structure, which may cause damage to the structure or weaken its integrity. In addition, traditional concrete strength testing methods usually take a long time to prepare samples, conduct tests, and analyze results, which increases the cost and time of testing, and cannot fully reflect the concrete quality of the entire structure. Therefore, the development of non-destructive testing technology is of great significance to improving the efficiency and reliability of concrete strength testing.
[0004] Therefore, an optimized strength testing solution for architectural concrete is desired. Summary of the invention
[0005] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.
[0006] In a first aspect, the present application provides a method for testing the strength of building concrete, the method comprising:
[0007] Arrange multiple strength detection points on the building concrete to be detected, and obtain position information of the multiple strength detection points;
[0008] Performing strength testing at the plurality of strength testing points using an ultrasonic rebound method to obtain ultrasonic rebound curves of the plurality of strength testing points;
[0009] Based on the position information of the multiple intensity detection points, construct a detection point spatial distribution topological matrix between the multiple intensity detection points, wherein the eigenvalue of each position on the non-diagonal position in the detection point spatial distribution topological matrix is the Euclidean distance between the corresponding two intensity detection points;
[0010] Taking each of the ultrasonic rebound curves of the multiple strength detection points as a node and the detection point spatial distribution topological matrix as an edge between nodes, the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple strength detection points are input into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected;
[0011] The strength distribution characteristic matrix of the building concrete to be detected is input into a strength measuring device based on a decoder to obtain a concrete strength measurement decoded value.
[0012] Optionally, taking each of the multiple intensity detection point ultrasonic rebound curves as a node and the detection point spatial distribution topological matrix as an edge between nodes, the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple intensity detection points are input into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected, including: performing waveform feature extraction on each of the multiple intensity detection point ultrasonic rebound curves to obtain multiple ultrasonic rebound waveform feature vectors; inputting the multiple ultrasonic rebound waveform feature vectors into a significance-driven global semantic association enhancement module to obtain multiple enhanced ultrasonic rebound waveform feature vectors; inputting the detection point spatial distribution topological matrix into a detection point spatial distribution topological feature extractor based on a convolutional neural network model to obtain a detection point spatial distribution topological feature matrix; inputting the multiple ultrasonic rebound waveform feature vectors and the detection point spatial distribution topological feature matrix into the graph neural network model to obtain the strength distribution feature matrix of the building concrete to be detected.
[0013] Optionally, waveform feature extraction is performed on each of the multiple intensity detection point ultrasonic rebound curves to obtain a plurality of ultrasonic rebound waveform feature vectors, including: passing each of the multiple intensity detection point ultrasonic rebound curves through a waveform feature extractor based on a void convolutional neural network model to obtain the plurality of ultrasonic rebound waveform feature vectors.
[0014] Optionally, the multiple ultrasonic echo waveform feature vectors are input into a significance-driven global semantic association enhancement module to obtain multiple enhanced ultrasonic echo waveform feature vectors, including: respectively calculating the mean and maximum value of each ultrasonic echo waveform feature vector in the multiple ultrasonic echo waveform feature vectors to obtain an ultrasonic echo waveform semantic global representation vector composed of multiple vector means and an ultrasonic echo waveform semantic salient representation vector composed of multiple vector maximum values; performing feature capture and expression on the ultrasonic echo waveform semantic global representation vector to obtain an ultrasonic echo waveform global semantic activation representation feature vector; performing feature capture and expression on the ultrasonic echo waveform semantic salient representation vector to obtain an ultrasonic A rebound waveform salient semantic activation representation feature vector; after fusing the ultrasonic rebound waveform global semantic activation representation feature vector and the ultrasonic rebound waveform salient semantic activation representation feature vector to obtain the ultrasonic rebound waveform global-salient semantic association activation feature vector, nonlinear activation processing is performed on the ultrasonic rebound waveform global-salient semantic association activation feature vector to obtain the ultrasonic rebound waveform global-salient semantic association weight feature vector; using the ultrasonic rebound waveform global-salient semantic association weight feature vector as a weight, positionally dot-multiplying the multiple ultrasonic rebound waveform feature vectors and adding the multiple ultrasonic rebound waveform feature vectors to obtain the multiple enhanced ultrasonic rebound waveform feature vectors.
[0015] Optionally, the ultrasonic echo waveform semantic global representation vector is feature captured and expressed to obtain an ultrasonic echo waveform global semantic activation representation feature vector, including: performing one-dimensional convolution encoding on the ultrasonic echo waveform semantic global representation vector and then processing it through a ReLU activation function to obtain an activation expression ultrasonic echo waveform semantic global representation vector; performing one-dimensional convolution encoding on the activation expression ultrasonic echo waveform semantic global representation vector again and then performing matrix multiplication with a first trainable linear transformation weight matrix to obtain the ultrasonic echo waveform global semantic activation representation feature vector.
[0016] Optionally, the ultrasonic echo waveform semantic salient representation vector is feature captured and expressed to obtain an ultrasonic echo waveform salient semantic activation representation feature vector, including: performing one-dimensional convolution encoding on the ultrasonic echo waveform semantic salient representation vector and then processing it through a ReLU activation function to obtain an activation expression ultrasonic echo waveform semantic salient representation vector; performing one-dimensional convolution encoding on the activation expression ultrasonic echo waveform semantic salient representation vector again and then performing matrix multiplication with a second trainable linear transformation weight matrix to obtain the ultrasonic echo waveform salient semantic activation representation feature vector.
[0017] Optionally, after fusing the ultrasonic echo waveform global semantic activation representation feature vector and the ultrasonic echo waveform salient semantic activation representation feature vector to obtain the ultrasonic echo waveform global-salient semantic association activation feature vector, the ultrasonic echo waveform global-salient semantic association activation feature vector is subjected to nonlinear activation processing to obtain the ultrasonic echo waveform global-salient semantic association weight feature vector, including: fusing the ultrasonic echo waveform global semantic activation representation feature vector and the ultrasonic echo waveform salient semantic activation representation feature vector to obtain the ultrasonic echo waveform global-salient semantic association activation feature vector; performing activation processing on the ultrasonic echo waveform global-salient semantic association activation feature vector using a hyperbolic tangent function and a Sigmoid function respectively to obtain an ultrasonic echo waveform global-salient feature hyperbolic tangent activation semantic representation vector and an ultrasonic echo waveform global-salient feature S-type activation semantic representation vector; performing position point multiplication on the ultrasonic echo waveform global-salient feature hyperbolic tangent activation semantic representation vector and the ultrasonic echo waveform global-salient feature S-type activation semantic representation vector to obtain the ultrasonic echo waveform global-salient semantic association weight feature vector.
[0018] Optionally, it also includes a training step: for training the waveform feature extractor based on the void convolutional neural network model, the significance-driven global semantic association enhancement module, the detection point spatial distribution topological feature extractor based on the convolutional neural network model, the graph neural network model and the decoder-based strength measurer.
[0019] Optionally, the training step includes: arranging multiple strength detection points on the building concrete to be tested, and obtaining training position information of the multiple strength detection points; using the ultrasonic rebound method to perform strength detection on the multiple strength detection points to obtain multiple training strength detection point ultrasonic rebound curves; based on the training position information of the multiple strength detection points, constructing a training detection point spatial distribution topological matrix between the multiple strength detection points, wherein the eigenvalue of each position on the non-diagonal position in the training detection point spatial distribution topological matrix is the Euclidean distance between the corresponding two strength detection points; each training strength detection point ultrasonic rebound curve in the multiple training strength detection points is ultrasonically detected. The rebound curves are respectively passed through the waveform feature extractor based on the hole convolutional neural network model to obtain multiple training ultrasonic rebound waveform feature vectors; the mean and maximum values of each training ultrasonic rebound waveform feature vector in the multiple training ultrasonic rebound waveform feature vectors are respectively calculated to obtain a training ultrasonic rebound waveform semantic global representation vector composed of multiple vector means and a training ultrasonic rebound waveform semantic salient representation vector composed of multiple vector maximum values; the training ultrasonic rebound waveform semantic global representation vector is feature captured and expressed to obtain a training ultrasonic rebound waveform global semantic activation representation feature vector; the training ultrasonic rebound waveform semantic salient representation vector is Feature capture and expression are performed to obtain a training ultrasonic echo waveform salient semantic activation representation feature vector; after fusing the training ultrasonic echo waveform global semantic activation representation feature vector and the training ultrasonic echo waveform salient semantic activation representation feature vector to obtain a training ultrasonic echo waveform global-salient semantic association activation feature vector, nonlinear activation processing is performed on the training ultrasonic echo waveform global-salient semantic association activation feature vector to obtain a training ultrasonic echo waveform global-salient semantic association weight feature vector; using the training ultrasonic echo waveform global-salient semantic association weight feature vector as a weight, the multiple training ultrasonic echo waveform feature vectors are activated. Multiply and add the multiple training ultrasonic rebound waveform feature vectors by position point to obtain multiple training enhanced ultrasonic rebound waveform feature vectors; input the training detection point spatial distribution topological matrix into the detection point spatial distribution topological feature extractor based on the convolutional neural network model to obtain the training detection point spatial distribution topological feature matrix; input the multiple training ultrasonic rebound waveform feature vectors and the training detection point spatial distribution topological feature matrix into the graph neural network model to obtain the training to-be-detected building concrete strength distribution feature matrix; pass the training to-be-detected building concrete strength distribution feature matrix through the decoder-based strength measurement device to obtain the concrete strength measurement decoding value;Based on the concrete strength measurement decoded value, the waveform feature extractor based on the hole convolutional neural network model, the saliency-driven global semantic association enhancement module, the detection point spatial distribution topological feature extractor based on the convolutional neural network model, the graph neural network model and the decoder-based strength measurer are trained. ;
[0020] In a second aspect, the present application provides a strength detection system for building concrete, the system comprising:
[0021] A detection point position information acquisition module is used to arrange multiple strength detection points on the building concrete to be detected and obtain the position information of the multiple strength detection points;
[0022] A strength detection module, used for performing strength detection at the plurality of strength detection points by using an ultrasonic rebound method to obtain ultrasonic rebound curves of the plurality of strength detection points;
[0023] A distribution topology matrix construction module is used to construct a detection point space distribution topology matrix between the multiple intensity detection points based on the position information of the multiple intensity detection points, wherein the eigenvalue of each position at the non-diagonal position in the detection point space distribution topology matrix is the Euclidean distance between the corresponding two intensity detection points;
[0024] A graph neural network module, which is used to use the ultrasonic rebound curves of each strength detection point in the multiple strength detection points as a node and the detection point spatial distribution topological matrix as an edge between nodes, and input the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple strength detection points into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected;
[0025] The decoding module is used to input the strength distribution characteristic matrix of the building concrete to be detected into a strength measuring device based on a decoder to obtain a concrete strength measurement decoding value.
[0026] By adopting the above technical solution, multiple strength detection points are arranged in the building concrete, the position information of these strength detection points is obtained, and the ultrasonic rebound method is used to perform strength detection at these strength detection points to obtain the ultrasonic rebound curve of the strength detection point. Then, the data processing and image analysis algorithm based on artificial intelligence is introduced at the back end to perform a graph-structure-based correlation analysis on the position information of these strength detection points and the ultrasonic rebound curve, so as to comprehensively understand the intrinsic strength distribution characteristics of concrete, thereby detecting and evaluating the strength of concrete. In this way, the intelligent detection of the strength of building concrete can be realized, avoiding the destructive and inefficient problems brought by traditional methods, and providing more effective means and support for the quality control of construction projects.
[0027] Other features and advantages of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] The above and other features, advantages and aspects of the embodiments of the present application will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale. In the drawings:
[0029] Figure 1 The present invention is a flow chart of a method for detecting strength of building concrete according to an exemplary embodiment.
[0030] Figure 2 The invention is a block diagram of a building concrete strength detection system according to an exemplary embodiment.
[0031] Figure 3 It is a block diagram of an electronic device according to an exemplary embodiment.
[0032] Figure 4 It is an application scenario diagram of a strength detection method for building concrete according to an exemplary embodiment. DETAILED DESCRIPTION
[0033] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.
[0034] It should be understood that the various steps described in the method implementation of the present application can be performed in different orders and / or performed in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.
[0035] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.
[0036] It should be noted that the concepts such as "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.
[0037] It should be noted that the modifications of "one" and "plurality" mentioned in the present application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".
[0038] The names of the messages or information exchanged between multiple devices in the embodiments of the present application are only used for illustrative purposes and are not used to limit the scope of these messages or information.
[0039] The specific implementation of the present application is described in detail below with reference to the accompanying drawings.
[0040] Figure 1 is a flow chart of a strength detection method for building concrete according to an exemplary embodiment. Figure 1 As shown, the method includes:
[0041] Step S101, arranging a plurality of strength detection points on the building concrete to be detected, and obtaining position information of the plurality of strength detection points;
[0042] Step S102, performing strength detection at the plurality of strength detection points by using an ultrasonic rebound method to obtain ultrasonic rebound curves of the plurality of strength detection points;
[0043] Step S103: constructing a detection point spatial distribution topology matrix between the multiple intensity detection points based on the position information of the multiple intensity detection points, wherein the eigenvalue of each position on the non-diagonal position in the detection point spatial distribution topology matrix is the Euclidean distance between the corresponding two intensity detection points;
[0044] Step S104, using each ultrasonic rebound curve of the multiple strength detection points as a node and the detection point spatial distribution topological matrix as an edge between nodes, inputting the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple strength detection points into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected;
[0045] Step S105: input the strength distribution characteristic matrix of the building concrete to be detected into a strength measuring device based on a decoder to obtain a concrete strength measurement decoded value.
[0046] In response to the above technical problems, in the technical solution of the present application, a method for strength testing of building concrete is proposed, which can perform non-destructive strength testing on concrete by utilizing the ultrasonic rebound method, that is, inferring the strength of concrete by measuring the speed of ultrasonic wave propagation in concrete and the rebound value.
[0047] However, the ultrasonic rebound method has the following limitations: affected by factors such as the surface state and moisture content of concrete, these noises will lead to inaccurate measurement results. In addition, traditional concrete strength testing methods usually ignore the spatial information between different test points and treat each test point as an independent data point for processing. This method cannot accurately describe the strength distribution characteristics inside the concrete, reduces the accuracy and reliability of concrete strength testing, and is easily interfered by external factors and produces errors.
[0048] In response to this problem, the technical concept of this application is to arrange multiple strength detection points in building concrete, obtain the position information of these strength detection points, and use the ultrasonic rebound method to perform strength detection at these strength detection points to obtain the ultrasonic rebound curve of the strength detection point. Then, an artificial intelligence-based data processing and image analysis algorithm is introduced at the back end to perform a graph-structure-based correlation analysis on the position information of these strength detection points and the ultrasonic rebound curve, so as to comprehensively understand the intrinsic strength distribution characteristics of concrete, thereby performing concrete strength detection and evaluation. In this way, intelligent detection of the strength of building concrete can be achieved, avoiding the destructive and inefficient problems brought about by traditional methods, and providing more effective means and support for construction project quality control.
[0049] Specifically, in the technical solution of the present application, first, multiple strength detection points are arranged in the building concrete to be detected, and the position information of the multiple strength detection points is obtained. Then, the ultrasonic rebound method is used to perform strength detection at the multiple strength detection points to obtain ultrasonic rebound curves of multiple strength detection points. It should be understood that the ultrasonic rebound method is a commonly used non-destructive detection method, which infers the strength of concrete by measuring the speed and rebound value of ultrasonic waves propagating in concrete. In this process, for each specific strength detection point, its corresponding ultrasonic rebound curve will be obtained, and these curves can reflect the strength information of concrete at different positions. Specifically, the ultrasonic rebound curve refers to a curve that records the rebound values at different positions after the concrete surface is hit with a rebound hammer. However, in practical applications, due to the influence of various factors, there is noise in the rebound curve, which may affect the strength test results. Therefore, it is of great significance to improve the accuracy and reliability of rebound curve analysis.
[0050] In addition, in the process of concrete strength detection, the spatial information between each detection reference point is usually ignored, which may lead to an incomplete and inaccurate understanding of the concrete strength distribution. Therefore, based on the position information of the multiple strength detection points, a detection point spatial distribution topological matrix between the multiple strength detection points is constructed, wherein the characteristic value of each position on the non-diagonal position in the detection point spatial distribution topological matrix is the Euclidean distance between the corresponding two strength detection points. By introducing the spatial distribution topological matrix and information between different strength detection points, the strength distribution characteristics inside the concrete can be better described, and the accuracy and reliability of subsequent strength detection can be improved.
[0051] In one embodiment of the present application, the ultrasonic rebound curve of each intensity detection point in the multiple intensity detection point ultrasonic rebound curves is used as a node and the detection point spatial distribution topological matrix is used as an edge between nodes, and the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple intensity detection points are input into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected, including: respectively performing waveform feature extraction on each intensity detection point ultrasonic rebound curve of the multiple intensity detection point ultrasonic rebound curves to obtain multiple ultrasonic rebound waveform feature vectors; inputting the multiple ultrasonic rebound waveform feature vectors into a significance-driven global semantic association enhancement module to obtain multiple enhanced ultrasonic rebound waveform feature vectors; inputting the detection point spatial distribution topological matrix into a detection point spatial distribution topological feature extractor based on a convolutional neural network model to obtain a detection point spatial distribution topological feature matrix; and inputting the multiple ultrasonic rebound waveform feature vectors and the detection point spatial distribution topological feature matrix into the graph neural network model to obtain the strength distribution feature matrix of the building concrete to be detected.
[0052] Furthermore, since each of the ultrasonic rebound curves of the multiple strength detection points contains the semantic information of the ultrasonic rebound curve obtained by the ultrasonic rebound method at a specific strength detection point, and the detection point spatial distribution topological matrix contains the spatial topological information of the different strength detection point brackets. The information of the two needs to be effectively integrated to provide a more comprehensive and integrated strength distribution feature of the building concrete to be detected, so as to avoid the problems of being susceptible to environmental influences and resulting in low reliability and insufficient understanding of the concrete strength distribution in the traditional concrete strength detection scheme. Based on this, in the technical solution of the present application, the ultrasonic rebound curves of each strength detection point in the multiple strength detection point ultrasonic rebound curves are used as nodes and the detection point spatial distribution topological matrix is used as the edge between nodes, and the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple strength detection points are input into the graph neural network model to obtain the strength distribution feature matrix of the building concrete to be detected.
[0053] Specifically, first, each of the multiple intensity detection point ultrasonic rebound curves is subjected to feature mining in a waveform feature extractor based on a dilated convolutional neural network model to extract waveform semantic feature information in each intensity detection point ultrasonic rebound curve, thereby obtaining multiple ultrasonic rebound waveform feature vectors. It should be understood that dilated convolution is a special convolution operation that can increase the receptive field (i.e., the range of input data that the network can observe) by introducing holes in the convolution kernel (i.e., adding zero padding to the convolution kernel) without increasing the size of the convolution kernel. This enables the network to capture a wider range of ultrasonic rebound waveform features while keeping the number of parameters relatively low.
[0054] In one embodiment of the present application, waveform feature extraction is performed on each of the multiple intensity detection point ultrasonic rebound curves to obtain a plurality of ultrasonic rebound waveform feature vectors, including: passing each of the multiple intensity detection point ultrasonic rebound curves through a waveform feature extractor based on a void convolutional neural network model to obtain the plurality of ultrasonic rebound waveform feature vectors.
[0055] Next, considering that in concrete strength detection, each ultrasonic rebound waveform feature vector contains the semantic features of the ultrasonic rebound waveform of each strength detection point extracted from the original ultrasonic rebound curve, and the ultrasonic rebound waveform semantics of different strength detection points contain some unimportant or redundant information, which may interfere with the learning and strength measurement capabilities of the model. Therefore, in order to be able to perform global relevance and importance analysis on the ultrasonic rebound waveform semantics of these different detection points, so as to identify which ones are more relevant to the subsequent strength detection task and which ones are redundant information, in the technical solution of the present application, the multiple ultrasonic rebound waveform feature vectors are further input into the significance-driven global semantic association enhancement module to obtain multiple enhanced ultrasonic rebound waveform feature vectors. Through the processing of the significance-driven global semantic association enhancement module, the global correlation relationship and importance features between the ultrasonic rebound waveform semantic features of each strength detection point can be automatically learned and captured, so as to identify the ultrasonic rebound waveform semantics expressed in each ultrasonic rebound waveform feature vector, and enhance the features related to the subsequent concrete strength distribution feature expression and strength detection task. In particular, the saliency-driven global semantic association enhancement module can adaptively adjust the enhancement strategy of different ultrasonic rebound waveform semantic features according to the specific circumstances of the data to improve the generalization ability of the model, thereby providing a more informative data basis for the subsequent analysis of the strength distribution characteristics of building concrete.
[0056] In one embodiment of the present application, the multiple ultrasonic echo waveform feature vectors are input into a significance-driven global semantic association enhancement module to obtain multiple enhanced ultrasonic echo waveform feature vectors, including: respectively calculating the mean and maximum value of each ultrasonic echo waveform feature vector in the multiple ultrasonic echo waveform feature vectors to obtain an ultrasonic echo waveform semantic global representation vector composed of multiple vector means and an ultrasonic echo waveform semantic salient representation vector composed of multiple vector maximum values; performing feature capture and expression on the ultrasonic echo waveform semantic global representation vector to obtain an ultrasonic echo waveform global semantic activation representation feature vector; performing feature capture and expression on the ultrasonic echo waveform semantic salient representation vector to obtain to the ultrasonic echo waveform salient semantic activation representation feature vector; after fusing the ultrasonic echo waveform global semantic activation representation feature vector and the ultrasonic echo waveform salient semantic activation representation feature vector to obtain the ultrasonic echo waveform global-salient semantic association activation feature vector, nonlinear activation processing is performed on the ultrasonic echo waveform global-salient semantic association activation feature vector to obtain the ultrasonic echo waveform global-salient semantic association weight feature vector; using the ultrasonic echo waveform global-salient semantic association weight feature vector as the weight, positionally dot-multiplying the multiple ultrasonic echo waveform feature vectors and adding the multiple ultrasonic echo waveform feature vectors to obtain the multiple enhanced ultrasonic echo waveform feature vectors.
[0057] Furthermore, in one embodiment of the present application, the ultrasonic echo waveform semantic global representation vector is feature captured and expressed to obtain an ultrasonic echo waveform global semantic activation representation feature vector, including: performing one-dimensional convolution encoding on the ultrasonic echo waveform semantic global representation vector and then processing it through a ReLU activation function to obtain an activation expression ultrasonic echo waveform semantic global representation vector; performing one-dimensional convolution encoding on the activation expression ultrasonic echo waveform semantic global representation vector again and then performing matrix multiplication with a first trainable linear transformation weight matrix to obtain the ultrasonic echo waveform global semantic activation representation feature vector.
[0058] Furthermore, in one embodiment of the present application, the ultrasonic echo waveform semantic salient representation vector is feature captured and expressed to obtain an ultrasonic echo waveform salient semantic activation representation feature vector, including: performing one-dimensional convolution encoding on the ultrasonic echo waveform semantic salient representation vector and then processing it through a ReLU activation function to obtain an activation expression ultrasonic echo waveform semantic salient representation vector; performing one-dimensional convolution encoding on the activation expression ultrasonic echo waveform semantic salient representation vector again and then performing matrix multiplication with a second trainable linear transformation weight matrix to obtain the ultrasonic echo waveform salient semantic activation representation feature vector.
[0059] Furthermore, in one embodiment of the present application, after fusing the ultrasonic echo waveform global semantic activation representation feature vector and the ultrasonic echo waveform salient semantic activation representation feature vector to obtain the ultrasonic echo waveform global-salient semantic association activation feature vector, the ultrasonic echo waveform global-salient semantic association activation feature vector is subjected to nonlinear activation processing to obtain the ultrasonic echo waveform global-salient semantic association weight feature vector, including: fusing the ultrasonic echo waveform global semantic activation representation feature vector and the ultrasonic echo waveform salient semantic activation representation feature vector to obtain the ultrasonic echo waveform global-salient semantic association activation feature vector; performing activation processing on the ultrasonic echo waveform global-salient semantic association activation feature vector using a hyperbolic tangent function and a Sigmoid function respectively to obtain an ultrasonic echo waveform global-salient feature hyperbolic tangent activation semantic representation vector and an ultrasonic echo waveform global-salient feature S-type activation semantic representation vector; performing position point multiplication on the ultrasonic echo waveform global-salient feature hyperbolic tangent activation semantic representation vector and the ultrasonic echo waveform global-salient feature S-type activation semantic representation vector to obtain the ultrasonic echo waveform global-salient semantic association weight feature vector.
[0060] Specifically, the plurality of ultrasonic rebound waveform feature vectors are input into the saliency-driven global semantic association enhancement module for processing using the following saliency enhancement formula to obtain the plurality of enhanced ultrasonic rebound waveform feature vectors;
[0061] The significance enhancement formula is:
[0062] V'=Avg(V)
[0063] V"=Max(V)
[0064]
[0065] V c =V⊙(tanh(C1+C2)⊙σ(C1+C2))+V
[0066] Wherein, V is the plurality of ultrasonic rebound waveform feature vectors, Avg(·) represents the global mean operation in the calculation vector, Max(·) represents the maximum value operation in the calculation vector, V' is the ultrasonic rebound waveform semantic mean representation vector, V" is the ultrasonic rebound waveform semantic maximum representation vector, Conv(·) represents one-dimensional convolutional coding, ReLU(·) represents the ReLU activation function, and denote the trainable weight matrices for linear transformation, C1 and C2 are the global semantic representation feature vector of ultrasonic rebound waveform and the prominent semantic representation feature vector of ultrasonic rebound waveform, σ(·) is the sigmoid function, tanh(·) is the tanh function, ⊙ denotes the point multiplication by position, V c are the multiple enhanced ultrasonic rebound waveform feature vectors.
[0067] Then, the detection point spatial distribution topology matrix is input into the detection point spatial distribution topology feature extractor based on the convolutional neural network model for feature extraction to extract the spatial distribution topology feature information between each intensity detection point, thereby obtaining the detection point spatial distribution topology feature matrix,
[0068] Furthermore, since the association between the semantic features of each ultrasonic rebound waveform and the spatial distribution topological feature support of each strength detection point is nonlinear, the ultrasonic rebound waveform feature vector provides detailed information about the internal characteristics of the concrete, and the spatial distribution feature matrix provides the spatial relationship between the detection points. Based on this, in order to be able to fuse and interactively analyze the features of the two, so as to learn and capture the complex nonlinear association relationship therein, so as to better perform the strength detection of concrete, in the technical solution of the present application, the multiple ultrasonic rebound waveform feature vectors and the detection point spatial distribution topological feature matrix are further input into the graph neural network model to obtain the strength distribution feature matrix of the building concrete to be detected. Specifically, the individual ultrasonic rebound waveform feature vectors are used as the feature representation of the node, and the detection point spatial distribution topological feature matrix is used as the feature representation of the edge between nodes, and the multiple ultrasonic rebound waveform feature vectors and the detection point spatial distribution topological feature matrix are input into the graph neural network model to obtain the strength distribution feature matrix of the building concrete to be detected. In particular, the graph neural network model uses learnable neural network parameters to perform graph structure data encoding on the multiple ultrasonic rebound waveform feature vectors and the detection point spatial distribution topological feature matrix to obtain the strength distribution feature matrix of the building concrete to be detected, which contains irregular detection point spatial distribution topological features and ultrasonic rebound waveform semantic feature information of each detection point. This helps to more fully understand the strength distribution of the concrete structure, thereby facilitating subsequent concrete strength detection.
[0069] Then, the strength distribution feature matrix of the building concrete to be detected is input into the strength measuring device based on the decoder to obtain the concrete strength measurement decoding value. In other words, the strength distribution feature of the building concrete to be detected expressed by the graph structure fusion feature is used for decoding regression to detect and evaluate the concrete strength. In this way, the intelligent detection of the strength of building concrete can be realized, avoiding the destructive and inefficient problems brought by the traditional method, and providing more effective means and support for the quality control of construction projects.
[0070] In one embodiment of the present application, the strength detection method of building concrete also includes a training step: for training the waveform feature extractor based on the void convolutional neural network model, the significance-driven global semantic association enhancement module, the detection point spatial distribution topological feature extractor based on the convolutional neural network model, the graph neural network model and the decoder-based strength measurer. The training step includes: arranging multiple strength detection points on the building concrete to be detected, and obtaining the training position information of the multiple strength detection points; using the ultrasonic rebound method to perform strength detection on the multiple strength detection points to obtain multiple training strength detection point ultrasonic rebound curves; based on the training position information of the multiple strength detection points, constructing a training detection point spatial distribution topological matrix between the multiple strength detection points, wherein the characteristic value of each position on the non-diagonal position in the training detection point spatial distribution topological matrix is the Euclidean distance between the corresponding two strength detection points; each training strength detection point ultrasonic rebound curve in the multiple training strength detection point ultrasonic rebound curves is respectively passed through the waveform feature extractor based on the void convolutional neural network model to obtain multiple training ultrasonic rebound waveform feature vectors; respectively calculating the mean and maximum value of each training ultrasonic rebound waveform feature vector in the multiple training ultrasonic rebound waveform feature vectors to obtain a training ultrasonic rebound waveform semantic global representation vector composed of multiple vector means and a training ultrasonic rebound waveform semantic prominence representation vector composed of multiple vector maximum values; feature capture and expression of the training ultrasonic rebound waveform semantic global representation vector to obtain a training ultrasonic rebound waveform global semantic activation representation feature vector; feature capture and expression of the training ultrasonic rebound waveform semantic salient representation vector to obtain the training ultrasonic rebound waveform salient semantic activation representation feature vector; after fusing the training ultrasonic rebound waveform global semantic activation representation feature vector and the training ultrasonic rebound waveform salient semantic activation representation feature vector to obtain the training ultrasonic rebound waveform global-salient semantic association activation feature vector, nonlinear activation processing is performed on the training ultrasonic rebound waveform global-salient semantic association activation feature vector to obtain the training ultrasonic rebound waveform global-salient semantic association weight feature vector; using the training ultrasonic rebound waveform global-salient semantic association weight feature vector as the weight, performing positional point multiplication on the multiple training ultrasonic rebound waveform feature vectors and adding the multiple training ultrasonic rebound waveform feature vectors to obtain multiple training enhanced ultrasonic rebound waveform feature vectors; inputting the training detection point spatial distribution topological matrix into the detection point spatial distribution topological feature extractor based on the convolutional neural network model to obtain the training detection point spatial distribution topological feature matrix; inputting the multiple training ultrasonic rebound waveform feature vectors and the training detection point spatial distribution topological feature matrix into the graph neural network model to obtain the training to-be-detected building concrete strength distribution feature matrix;The training to-be-detected building concrete strength distribution feature matrix is passed through the decoder-based strength measurer to obtain a concrete strength measurement decoded value; based on the concrete strength measurement decoded value, the waveform feature extractor based on the hole convolutional neural network model, the significance-driven global semantic association enhancement module, the detection point spatial distribution topological feature extractor based on the convolutional neural network model, the graph neural network model and the decoder-based strength measurer are trained. ;
[0071] In the technical solution described above, each of the multiple training enhanced ultrasonic rebound waveform feature vectors represents the image semantic features of the ultrasonic rebound curve of the training intensity detection point of each intensity detection point. The training detection point spatial distribution topological feature matrix is used to represent the spatial distribution topological features between each intensity detection point. Then, the multiple training enhanced ultrasonic rebound waveform feature vectors are used as the high-dimensional feature representation of the node and the training detection point spatial distribution topological feature matrix is used as the high-dimensional feature representation of the edge between the node and the node. When it is input into the graph neural network model for graph structure data encoding, each row vector in the training to be detected building concrete strength distribution feature matrix obtained contains the spatial distribution topological features between each intensity detection point and the image semantic features of the ultrasonic rebound curve of each training intensity detection point, which will enhance the feature representation global richness of the training to be detected building concrete strength distribution feature matrix. However, it will also cause a more significant local distribution difference between each row vector in the training to be detected building concrete strength distribution feature matrix, which will cause its harmony offset relative to the predetermined decoding value of the decoder, affecting the decoding iteration effect.
[0072] Based on this, in a preferred embodiment of the present application, the process of passing the training to-be-detected building concrete strength distribution feature matrix through a strength measuring device based on a decoder to obtain the concrete strength measurement decoded value includes:
[0073] First, the training building concrete strength distribution feature matrix to be detected is expanded into a training building concrete strength distribution feature vector to be detected, the autocorrelation matrix of the training building concrete strength distribution feature vector to be detected and its own transposition is calculated, and the inner product of the i-th and j-th row vectors of the autocorrelation matrix is calculated as the matrix value of the (i, j)-th position of the weight matrix to obtain the weight matrix, then, after matrix multiplication of the weight matrix and the autocorrelation matrix, further matrix-vector multiplication is performed with the training building concrete strength distribution feature vector to be detected to obtain a correction vector, finally, the correction vector is dot-multiplied with the training building concrete strength distribution feature vector to be detected to obtain an optimized building concrete strength distribution feature vector to be detected, and the optimized building concrete strength distribution feature vector to be detected is passed through the decoder-based strength measuring device to obtain the concrete strength measurement decoding value.
[0074] The optimization steps are specifically expressed as follows:
[0075]
[0076] Where V is the training building concrete strength distribution feature vector to be tested, V' is the optimized building concrete strength distribution feature vector to be tested, M is the autocorrelation matrix, V i and V j are the i-th and j-th row vectors of the autocorrelation matrix, respectively, m i,j is the matrix value of the (i, j)th position in the weight matrix, M' is the weight matrix, is matrix multiplication, ⊙ is point multiplication by position, V T is the transposed vector of the training building concrete strength distribution feature vector to be tested, V j T is the transposed vector of the j-th row vector of the autocorrelation matrix.
[0077] In this way, based on the self-correlation dimension of the training to-be-detected building concrete strength distribution feature vector as the object to be modulated, the correlation expansion is performed based on the spatial sub-dimensional complexity of the high-dimensional feature space of the feature distribution, thereby introducing the decomposable correlation dimension offset of the training to-be-detected building concrete strength distribution feature vector into the heterogeneous correlation embedding space, thereby enhancing its correlation self-consistency through the joint fine-tuning of the decomposable dimension set represented by the heterogeneous correlation embedding space, so as to improve the feature set of the training to-be-detected building concrete strength distribution feature vector in the predetermined decoding label harmony in the associated target decoding domain, thereby improving the decoding training speed and the accuracy of the decoding result when the to-be-detected building concrete strength distribution feature matrix is decoded by the strength measuring device based on the decoder. In this way, the intrinsic strength distribution characteristics of concrete can be understood more accurately, so as to detect and evaluate the strength of concrete, and provide more effective means and support for the quality control of construction projects.
[0078] In summary, the above scheme is adopted to arrange multiple strength detection points in building concrete, obtain the position information of these strength detection points, and use the ultrasonic rebound method to perform strength detection at these strength detection points to obtain the ultrasonic rebound curve of the strength detection point. Then, the data processing and image analysis algorithm based on artificial intelligence is introduced at the back end to perform a graph-structure-based correlation analysis on the position information of these strength detection points and the ultrasonic rebound curve, so as to comprehensively understand the intrinsic strength distribution characteristics of concrete, thereby detecting and evaluating the strength of concrete. In this way, the intelligent detection of the strength of building concrete can be realized, avoiding the destructive and inefficient problems brought by traditional methods, and providing more effective means and support for the quality control of construction projects.
[0079] Figure 2 FIG. 1 is a block diagram of a strength detection system for building concrete according to an exemplary embodiment. Figure 2 As shown, the strength detection system 200 includes:
[0080] The detection point position information acquisition module 201 is used to arrange multiple strength detection points on the building concrete to be detected and obtain the position information of the multiple strength detection points;
[0081] A strength detection module 202, configured to perform strength detection at the plurality of strength detection points by using an ultrasonic rebound method to obtain ultrasonic rebound curves of the plurality of strength detection points;
[0082] A distribution topology matrix construction module 203 is used to construct a detection point space distribution topology matrix between the multiple intensity detection points based on the position information of the multiple intensity detection points, wherein the eigenvalue of each position on the non-diagonal position in the detection point space distribution topology matrix is the Euclidean distance between the corresponding two intensity detection points;
[0083] A graph neural network module 204 is used to use each ultrasonic rebound curve of the multiple strength detection points as a node and the detection point spatial distribution topological matrix as an edge between nodes, and input the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple strength detection points into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected;
[0084] The decoding module 205 is used to input the strength distribution characteristic matrix of the building concrete to be detected into a strength measuring device based on a decoder to obtain a concrete strength measurement decoded value.
[0085] Reference below Figure 3 , which shows a schematic diagram of the structure of an electronic device 600 suitable for implementing the embodiment of the present application. The terminal device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.
[0086] like Figure 3 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.
[0087] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.
[0088] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present application are executed.
[0089] It should be noted that the computer-readable medium mentioned above in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.
[0090] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.
[0091] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.
[0092] Computer program code for performing the operations of the present application may be written in one or more programming languages or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0093] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0094] The modules involved in the embodiments described in this application can be implemented by software or hardware. The name of the module does not limit the module itself in some cases. For example, the test parameter acquisition module can also be described as a "module for acquiring device test parameters corresponding to the target device".
[0095] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.
[0096] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0097] Figure 4 FIG. 1 is an application scenario diagram of a strength detection method for building concrete according to an exemplary embodiment. Figure 4 As shown, in this application scenario, first, multiple strength detection points are arranged on the building concrete to be detected, and the position information of the multiple strength detection points is obtained (for example, Figure 4 Then, the acquired location information is input to a server (for example, Figure 4 In the S) shown in FIG. 1 , the server is capable of processing the location information based on a strength detection algorithm for building concrete to obtain a decoded value of a concrete strength measurement.
[0098] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.
[0099] In addition, although each operation is described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or to be performed in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the application. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.
[0100] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
Claims
1. A strength testing method for building concrete, characterized in that: include: Arrange multiple strength detection points on the building concrete to be detected, and obtain position information of the multiple strength detection points; Performing strength testing at the plurality of strength testing points using an ultrasonic rebound method to obtain ultrasonic rebound curves of the plurality of strength testing points; Based on the position information of the multiple intensity detection points, construct a detection point spatial distribution topological matrix between the multiple intensity detection points, wherein the eigenvalue of each position on the non-diagonal position in the detection point spatial distribution topological matrix is the Euclidean distance between the corresponding two intensity detection points; Taking each of the ultrasonic rebound curves of the multiple strength detection points as a node and the detection point spatial distribution topological matrix as an edge between nodes, the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple strength detection points are input into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected; The strength distribution characteristic matrix of the building concrete to be detected is input into a strength measuring device based on a decoder to obtain a concrete strength measurement decoded value.
2. The strength testing method for building concrete according to claim 1, characterized in that: Taking each of the ultrasonic rebound curves of the multiple strength detection points as a node and the detection point spatial distribution topological matrix as an edge between nodes, the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple strength detection points are input into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected, including: respectively extracting waveform features of each of the multiple ultrasonic rebound curves at the multiple intensity detection points to obtain multiple ultrasonic rebound waveform feature vectors; Inputting the plurality of ultrasonic rebound waveform feature vectors into a saliency-driven global semantic association enhancement module to obtain a plurality of enhanced ultrasonic rebound waveform feature vectors; Inputting the detection point spatial distribution topology matrix into a detection point spatial distribution topology feature extractor based on a convolutional neural network model to obtain a detection point spatial distribution topology feature matrix; The multiple ultrasonic rebound waveform feature vectors and the detection point spatial distribution topological feature matrix are input into the graph neural network model to obtain the strength distribution feature matrix of the building concrete to be detected.
3. The strength testing method for building concrete according to claim 2, characterized in that: The method comprises: performing waveform feature extraction on each of the multiple intensity detection point ultrasonic rebound curves respectively to obtain a plurality of ultrasonic rebound waveform feature vectors, comprising: passing each of the multiple intensity detection point ultrasonic rebound curves respectively through a waveform feature extractor based on a void convolutional neural network model to obtain the plurality of ultrasonic rebound waveform feature vectors.
4. The strength testing method for building concrete according to claim 3, characterized in that: Inputting the plurality of ultrasonic rebound waveform feature vectors into a saliency-driven global semantic association enhancement module to obtain a plurality of enhanced ultrasonic rebound waveform feature vectors, including: Calculating the mean and maximum of each ultrasonic rebound waveform feature vector in the plurality of ultrasonic rebound waveform feature vectors respectively to obtain an ultrasonic rebound waveform semantic global representation vector composed of a plurality of vector mean values and an ultrasonic rebound waveform semantic salient representation vector composed of a plurality of vector maximum values; Capturing and expressing the ultrasonic rebound waveform semantic global representation vector to obtain an ultrasonic rebound waveform global semantic activation representation feature vector; Capturing and expressing the ultrasonic rebound waveform semantic salient representation vector to obtain an ultrasonic rebound waveform salient semantic activation representation feature vector; After fusing the ultrasonic rebound waveform global semantic activation representation feature vector and the ultrasonic rebound waveform salient semantic activation representation feature vector to obtain an ultrasonic rebound waveform global-salient semantic association activation feature vector, nonlinear activation processing is performed on the ultrasonic rebound waveform global-salient semantic association activation feature vector to obtain an ultrasonic rebound waveform global-salient semantic association weight feature vector; The ultrasonic rebound waveform global-prominent semantic association weight feature vector is used as a weight, and the multiple ultrasonic rebound waveform feature vectors are multiplied by position and added to obtain the multiple enhanced ultrasonic rebound waveform feature vectors.
5. The strength testing method for building concrete according to claim 4, characterized in that: Capturing and expressing the ultrasonic rebound waveform semantic global representation vector to obtain an ultrasonic rebound waveform global semantic activation representation feature vector includes: The ultrasonic rebound waveform semantic global representation vector is subjected to one-dimensional convolution encoding and then processed by a ReLU activation function to obtain an activated expression ultrasonic rebound waveform semantic global representation vector; The activation expression ultrasonic rebound waveform semantic global representation vector is again subjected to one-dimensional convolution encoding and then matrix multiplied with the first trainable linear transformation weight matrix to obtain the ultrasonic rebound waveform global semantic activation representation feature vector.
6. The strength testing method for building concrete according to claim 5, characterized in that: Capturing and expressing the ultrasonic rebound waveform semantic salient representation vector to obtain an ultrasonic rebound waveform salient semantic activation representation feature vector includes: After performing one-dimensional convolution encoding on the ultrasonic rebound waveform semantic salient representation vector, the vector is processed by a ReLU activation function to obtain an activated expression ultrasonic rebound waveform semantic salient representation vector; The activation expression ultrasonic rebound waveform semantic salient representation vector is again subjected to one-dimensional convolution encoding and then matrix multiplied with a second trainable linear transformation weight matrix to obtain the ultrasonic rebound waveform salient semantic activation representation feature vector.
7. The strength testing method for building concrete according to claim 6, characterized in that: After fusing the ultrasonic rebound waveform global semantic activation representation feature vector and the ultrasonic rebound waveform salient semantic activation representation feature vector to obtain the ultrasonic rebound waveform global-salient semantic association activation feature vector, nonlinear activation processing is performed on the ultrasonic rebound waveform global-salient semantic association activation feature vector to obtain the ultrasonic rebound waveform global-salient semantic association weight feature vector, including: Fusion of the ultrasonic rebound waveform global semantic activation representation feature vector and the ultrasonic rebound waveform salient semantic activation representation feature vector to obtain the ultrasonic rebound waveform global-salient semantic association activation feature vector; The ultrasonic rebound waveform global-salient semantic association activation feature vector is activated by using a hyperbolic tangent function and a sigmoid function respectively to obtain an ultrasonic rebound waveform global-salient feature hyperbolic tangent activation semantic representation vector and an ultrasonic rebound waveform global-salient feature S-type activation semantic representation vector; The ultrasonic rebound waveform global-salient feature hyperbolic tangent activation semantic representation vector and the ultrasonic rebound waveform global-salient feature S-type activation semantic representation vector are multiplied point by point to obtain the ultrasonic rebound waveform global-salient semantic association weight feature vector.
8. The strength testing method for building concrete according to claim 7, characterized in that: It also includes a training step: for training the waveform feature extractor based on the void convolutional neural network model, the significance-driven global semantic association enhancement module, the detection point spatial distribution topological feature extractor based on the convolutional neural network model, the graph neural network model and the decoder-based strength measurer.
9. The strength testing method for building concrete according to claim 8, characterized in that: The training step comprises: Arrange multiple strength detection points on the building concrete to be detected, and obtain training position information of the multiple strength detection points; Performing strength detection at the plurality of strength detection points by using an ultrasonic rebound method to obtain ultrasonic rebound curves of a plurality of training strength detection points; Based on the training position information of the multiple intensity detection points, construct a training detection point spatial distribution topological matrix between the multiple intensity detection points, wherein the eigenvalue of each position on the non-diagonal position in the training detection point spatial distribution topological matrix is the Euclidean distance between the corresponding two intensity detection points; Passing each of the plurality of training intensity detection point ultrasonic rebound curves through the waveform feature extractor based on the atrous convolutional neural network model to obtain a plurality of training ultrasonic rebound waveform feature vectors; Respectively calculating the mean and maximum value of each training ultrasonic rebound waveform feature vector in the plurality of training ultrasonic rebound waveform feature vectors to obtain a training ultrasonic rebound waveform semantic global representation vector composed of a plurality of vector means and a training ultrasonic rebound waveform semantic salient representation vector composed of a plurality of vector maximum values; Capturing and expressing the training ultrasonic rebound waveform semantic global representation vector to obtain the training ultrasonic rebound waveform global semantic activation representation feature vector; Capturing and expressing the training ultrasonic echo waveform semantic salient representation vector to obtain the training ultrasonic echo waveform salient semantic activation representation feature vector; After fusing the training ultrasonic echo waveform global semantic activation representation feature vector and the training ultrasonic echo waveform salient semantic activation representation feature vector to obtain the training ultrasonic echo waveform global-salient semantic association activation feature vector, nonlinear activation processing is performed on the training ultrasonic echo waveform global-salient semantic association activation feature vector to obtain the training ultrasonic echo waveform global-salient semantic association weight feature vector; Taking the training ultrasonic rebound waveform global-salient semantic association weight feature vector as a weight, performing positional point multiplication on the multiple training ultrasonic rebound waveform feature vectors and adding the multiple training ultrasonic rebound waveform feature vectors to obtain multiple training enhanced ultrasonic rebound waveform feature vectors; Inputting the training detection point spatial distribution topology matrix into the detection point spatial distribution topology feature extractor based on the convolutional neural network model to obtain the training detection point spatial distribution topology feature matrix; Inputting the plurality of training ultrasonic rebound waveform feature vectors and the training detection point spatial distribution topological feature matrix into the graph neural network model to obtain a training to-be-detected building concrete strength distribution feature matrix; Passing the training to-be-detected building concrete strength distribution characteristic matrix through the decoder-based strength measuring device to obtain a concrete strength measurement decoded value; Based on the concrete strength measurement decoding value, the waveform feature extractor based on the void convolutional neural network model, the significance-driven global semantic association enhancement module, the detection point spatial distribution topological feature extractor based on the convolutional neural network model, the graph neural network model and the decoder-based strength measurer are trained.
10. A strength detection system for building concrete, characterized in that: include: A detection point position information acquisition module is used to arrange multiple strength detection points on the building concrete to be detected and obtain the position information of the multiple strength detection points; A strength detection module, used for performing strength detection at the plurality of strength detection points by using an ultrasonic rebound method to obtain ultrasonic rebound curves of the plurality of strength detection points; A distribution topology matrix construction module is used to construct a detection point space distribution topology matrix between the multiple intensity detection points based on the position information of the multiple intensity detection points, wherein the eigenvalue of each position at the non-diagonal position in the detection point space distribution topology matrix is the Euclidean distance between the corresponding two intensity detection points; A graph neural network module, which is used to use the ultrasonic rebound curves of each strength detection point in the multiple strength detection points as a node and the detection point spatial distribution topological matrix as an edge between nodes, and input the detection point spatial distribution topological matrix and the detection point spatial distribution topological matrix between the multiple strength detection points into a graph neural network model to obtain a strength distribution feature matrix of the building concrete to be detected; The decoding module is used to input the strength distribution characteristic matrix of the building concrete to be detected into a strength measuring device based on a decoder to obtain a concrete strength measurement decoding value.
Citation Information
Patent Citations
Intelligent control system and method for gate of urban railway AFC system
CN118505231A
Breast cancer patient rehabilitation nursing system and method based on Internet
CN118507083A
Wind power generation equipment immersive simulation practical training system based on virtual reality
CN118535021A
Mobile health medical service providing method and mobile health medical vehicle
CN118571512A
Signal test equipment and method for aircraft high-definition display
CN118604501A
