Ultrasonic testing method and system applied to joint coating testing, device and medium

Abstract

An ultrasonic testing method and system applied to joint coating testing, a device and a medium. The method comprises: acquiring first position information corresponding to a joint coating, and, on the basis of a preset first heuristic search algorithm, calculating a movement path of an ultrasonic tester corresponding to the first position information (S101); on the basis of the movement path, acquiring a first ultrasonic image corresponding to the joint coating collected by the ultrasonic tester (S102); and, on the basis of a trained identification model, acquiring joint coating identification information corresponding to the first ultrasonic image, the joint coating identification information being abnormality determination information or normality determination information (S103).

Description

Ultrasonic detection method, system, device and medium applied to joint detection

[0001] The present disclosure claims priority to Chinese Patent Application No. 202411144102.0, filed on August 20, 2024, the entire contents of which are incorporated herein by reference. TECHNICAL FIELD

[0002] The present disclosure relates to the technical field of joint ultrasonic detection, and in particular to an ultrasonic detection method, system, device and medium applied to joint detection. BACKGROUND

[0003] In the fields of industry and maintenance, it is crucial to ensure the structural integrity and safety. Joint bonding as a common method of connecting and repairing key structural components, its quality is directly related to the stability and service life of the entire structure. At present, ultrasonic detection technology is used to detect joint bonding. Ultrasonic detection technology is widely used in the fields of aviation, aerospace, shipbuilding, construction, etc. due to its high sensitivity, strong material penetration ability, and ability to detect deep defects. SUMMARY

[0004] In a first aspect, the present disclosure provides an ultrasonic detection method applied to joint detection, comprising:

[0005] obtaining first position information corresponding to a joint, and calculating a moving path of an ultrasonic detector corresponding to the first position information based on a preset first heuristic search algorithm;

[0006] obtaining a first ultrasonic image corresponding to the joint collected by the ultrasonic detector based on the moving path;

[0007] obtaining joint identification information corresponding to the first ultrasonic image based on a trained identification model, the joint identification information being abnormal information or normal information.

[0008] In some embodiments, if the joint identification information is abnormal information, the method further comprises:

[0009] obtaining a target repair scheme corresponding to the abnormal information, and repairing the joint according to the target repair scheme.

[0010] In some embodiments, the obtaining of the target repair scheme corresponding to the abnormal information comprises:

[0011] obtaining a current key feature corresponding to the abnormal information, querying the target repair scheme corresponding to the current key feature based on a standard scheme database established, and the standard scheme database comprises repair schemes corresponding to different key features.

[0012] In some embodiments, after the target repair scheme corresponding to the determination abnormal information is obtained, the method further comprises:

[0013] obtaining a second ultrasonic image corresponding to the repaired joint at the time of repair, collected by the ultrasonic detector;

[0014] determining whether the joint recognition information corresponding to the second ultrasonic image is the determination abnormal information;

[0015] If yes, revising the target repair scheme in the standard scheme database based on the second ultrasonic image and the determination abnormal information.

[0016] In some embodiments, after the joint is repaired based on the target repair scheme, the method further comprises:

[0017] If the repaired joint needs to be rechecked, obtaining a first position coordinate corresponding to the ultrasonic detector and a second position coordinate corresponding to the joint;

[0018] calculating a rechecking path corresponding to the first position coordinate and the second position coordinate based on a preset second heuristic search algorithm;

[0019] obtaining a third ultrasonic image corresponding to the repaired joint, collected by the ultrasonic detector during rechecking according to the rechecking path;

[0020] comparing the third ultrasonic image with a preset ultrasonic image corresponding to the repaired joint, and determining whether the rechecking position corresponding to the joint is accurate based on the comparison result.

[0021] In some embodiments, the obtaining of the first position information corresponding to the joint comprises:

[0022] Sa: initializing a set of particles based on a search area to obtain a particle set, the second position information of each particle representing a hypothetical position corresponding to the joint, and each particle being assigned an initial weight, the initial weights of the particles being equal, and the weight of each particle being a parameter representing the accuracy of the hypothetical position;

[0023] Sb: predicting the next position of each particle in the particle set based on a static model, and updating the second position information of the corresponding particle according to each predicted position;

[0024] Sc: For each particle, based on the updated second position information of the particle, a corresponding gusset feature matching degree of the particle is obtained, and based on the gusset feature matching degree, an updated weight corresponding to the particle is obtained, the updated weight is used to replace the initial weight as a new weight corresponding to the particle;

[0025] Sd: Resampling based on the updated weight corresponding to each particle, to obtain a new particle set;

[0026] Se: Repeating Sb to Sd based on the new particle set until a preset end condition is reached, and outputting the final first position information based on the updated second position information and the updated weight of the particles in the new particle set.

[0027] In some embodiments, the training method of the identification model comprises:

[0028] Obtaining a gusset ultrasound image data set, preprocessing the gusset ultrasound image data set, and dividing the preprocessed gusset ultrasound image data set into a training set and a validation set;

[0029] Labeling each ultrasound image data in the training set and the validation set to determine the gusset identification information corresponding to each ultrasound image data;

[0030] Training an initial model based on the training set, and verifying the trained initial model through the validation set to obtain the trained identification model.

[0031] In a second aspect, the present disclosure provides an ultrasonic detection system applied to gusset detection, comprising:

[0032] A path planning unit for obtaining first position information corresponding to a gusset, and calculating a moving path of an ultrasonic detector corresponding to the first position information based on a preset first heuristic search algorithm;

[0033] An acquisition unit for obtaining a first ultrasonic image corresponding to the gusset collected by the ultrasonic detector based on the moving path;

[0034] An abnormality identification unit for obtaining gusset identification information corresponding to the first ultrasonic image based on the trained identification model, the gusset identification information being abnormal information or normal information.

[0035] In a third aspect, the present disclosure provides an electronic device comprising a processor and a memory, the processor being coupled to the memory;

[0036] The processor is configured to execute a computer program stored in the memory, so that the electronic device performs the method according to any one of the first aspect or embodiments.

[0037] In a fourth aspect, the present disclosure provides a computer-readable storage medium, including a computer program or instructions, which, when executed on a computer, cause the computer to perform the method according to any one of the first aspect or embodiments.

[0038] In a fifth aspect, the present disclosure provides a computer program product, including computer instructions, which, when executed by a processor, implement the method according to any one of the first aspect or embodiments.

[0039] In a sixth aspect, the present disclosure provides a computer program, including computer instructions, which, when executed by a processor, implement the method according to any one of the first aspect or embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0040] FIG. 1 is a flowchart of an ultrasonic detection method applied to joint detection according to an embodiment of the present disclosure.

[0041] FIG. 2 is a flowchart of a first position information determination method according to an embodiment of the present disclosure.

[0042] FIG. 3 is a flowchart of a movement path determination method according to an embodiment of the present disclosure.

[0043] FIG. 4 is a structural block diagram of an ultrasonic detection system applied to joint detection according to an embodiment of the present disclosure.

[0044] FIG. 5 is a whole block diagram of an ultrasonic detection system applied to joint detection according to an embodiment of the present disclosure.

[0045] FIG. 6 is a structural block diagram of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0046] In related technology, an ultrasonic automatic detection system for copper and stainless steel dissimilar metal vacuum brazing water joints, the hardware is composed of a water joint automatic detection device, a single crystal water immersion focusing probe, and a contrast test block. The detection device includes an industrial computer, a motor driver and a motor controlled by the industrial computer, an ultrasonic flaw detector connected to the single crystal water immersion focusing probe and the industrial computer, respectively, a vertical motion transmission shaft driven by a motor of the detection device, a probe angle adjuster mounted on the shaft, a single crystal water immersion focusing probe mounted at the lower end of the probe angle adjuster, and a horizontal turntable mounted with a water joint to be tested driven by another motor. The position of the probe in the water joint cavity is adjusted by adjusting the probe angle adjuster, and the brazing quality of the water joint is detected by emitting ultrasonic waves.

[0047] Although the above detection system can realize automatic control, the detection system has the defect that the ultrasonic detection path is not planned in the actual joint detection process, and therefore, an automatic joint detection scheme capable of automatically planning a detection path is urgently needed.

[0048] The present disclosure will be further described in detail below with reference to the accompanying drawings.

[0049] The present disclosure provides an ultrasonic detection method applied to joint detection, which can be executed by a device, which can be a server or a terminal device. The server can be a physical server, a server cluster composed of multiple physical servers, a distributed system, or a cloud server providing cloud computing services. The terminal device can be a smartphone, a tablet computer, a desktop computer, or the like, but is not limited thereto.

[0050] As shown in FIG. 1, an ultrasonic detection method applied to joint detection is taken as an execution subject of an electronic device, and the main process of the method is described as follows (S101-S103):

[0051] S101: Obtain first position information corresponding to a joint, and calculate a movement path of an ultrasonic detector corresponding to the first position information based on a preset first heuristic search algorithm.

[0052] In this embodiment, the first position information can be a position area corresponding to the joint. It is easy to understand that the ultrasonic detector is a device that detects the joint bonding by using ultrasonic detection technology. The movement path is a path guiding the ultrasonic detector to reach the joint for detection, and the first heuristic search algorithm can be an A* heuristic search algorithm.

[0053] S102: Obtain a first ultrasonic image corresponding to the joint collected by the ultrasonic detector based on the movement path.

[0054] It is easy to understand that the ultrasonic image is a non-invasive imaging image generated by using an ultrasonic probe to emit and receive ultrasonic waves based on the reflection and scattering characteristics of ultrasonic waves at the joint, and then obtaining the internal structure and tissue information of the joint.

[0055] S103: Obtain joint identification information corresponding to the first ultrasonic image based on a trained recognition model, wherein the joint identification information is abnormal information or normal information.

[0056] In this embodiment, the recognition model can be a deep neural network model. The ultrasonic image corresponding to the gasket is input into the trained deep neural network model, and the deep neural network model can recognize the ultrasonic image and output gasket recognition information. The gasket recognition information is a parameter representing the gasket detection result. When the gasket recognition information is abnormal information, it indicates that the gasket is abnormal. When the gasket recognition information is normal information, it indicates that the gasket is normal.

[0057] The abnormal information can include at least one of an abnormal type, an abnormal distribution, and an abnormal degree. The abnormal type can include at least one of a crack, a pore, and an inclusion. The abnormal distribution can include a scattered defect, a locally concentrated defect, and a widely distributed defect. The abnormal degree can include slight, moderate, and severe. It is easy to understand that the crack means that the gasket has a crack, the pore means that the gasket has a cavity or a bubble, and the inclusion means that the gasket has a foreign matter. The scattered defect means that the abnormality is randomly distributed in the gasket. The locally concentrated defect means that the abnormality is highly concentrated in a local area of the gasket. The widely distributed defect means that the abnormality is widely distributed in the entire gasket. Slight means that the abnormality has little effect on the performance of the gasket. Moderate means that the abnormality has a certain effect on the performance of the gasket. Severe means that the abnormality has a serious effect on the performance of the gasket.

[0058] In some embodiments, the electronic device is preconfigured with a correspondence between the abnormal type, the abnormal distribution, and the abnormal degree. For example, when the abnormal type is a crack and the abnormal distribution is widely distributed, the corresponding abnormal degree can be severe. When the abnormal type is a pore and the abnormal distribution is scattered, the corresponding abnormal degree can be slight. When the abnormal type is an inclusion and the abnormal distribution is locally concentrated, the corresponding abnormal degree can be moderate.

[0059] In this embodiment, the deep learning technology is used, and the trained deep neural network model can automatically recognize the gasket recognition information corresponding to the first ultrasonic image, thereby improving the accuracy and efficiency of ultrasonic image recognition.

[0060] The electronic device can feed back the gasket recognition information to the user end, so as to facilitate the user to connect the detection situation of the current gasket. In the gasket detection process, the moving path of the ultrasonic detector is planned first. Through path planning, the ultrasonic detector can move along the optimal moving path, thereby reducing the possibility of unnecessary detours and repeated detection, and significantly improving the detection efficiency. The planning of the moving path can ensure that the ultrasonic detector can avoid obstacles and dangerous areas during the movement, thereby ensuring the safety and stability of the detection process. The planning of the moving path can ensure that the ultrasonic detector can fully cover the gasket to be detected during the detection process, thereby reducing the possibility of omission and mis-detection.

[0061] In some embodiments, the electronic device can obtain real-time detected environmental sensor data of the grommet. The environmental sensor data can include at least one of the following: laser radar point cloud, infrared ranging data, visual camera image, etc. Based on the environmental sensor data, it is determined whether there is a dynamic obstacle on the moving path. In the case that there is a dynamic obstacle on the moving path, the first heuristic search algorithm is constrained based on a time sliding window mechanism, an optimized path that avoids the dynamic obstacle is re-planned within the local path range covered by the time sliding window, and the optimized path is taken as the new path of the ultrasonic detector.

[0062] In this embodiment, the electronic device stores a positioning identification strategy for identifying the first position information of the grommet. The positioning identification strategy can include the following processes: collecting an ultrasonic image of the grommet to be detected by a high-resolution device, pre-processing the ultrasonic image to optimize the quality of the ultrasonic image, such as image denoising and image enhancement, extracting the position and shape features of the grommet from the pre-processed ultrasonic image using a convolutional neural network, and distinguishing the grommet and the surrounding structure through image segmentation to obtain a search area corresponding to the grommet to be detected; and implementing automatic and accurate determination of the first position information of the grommet based on a positioning algorithm of a probability model.

[0063] The positioning algorithm can use a Monte Carlo positioning algorithm. Through the positioning identification strategy, accurate imaging of the grommet to be detected, feature extraction corresponding to the grommet, and accurate identification of the spatial position are realized, which provides a reliable data basis for the subsequent optimization of the moving path.

[0064] As shown in FIG. 2, in this embodiment, the positioning algorithm based on the probability model automatically and accurately determines the first position information of the grommet, which can include the following processes:

[0065] Sa: based on the search area, a set of particles is initialized to obtain a particle set, the second position information of each particle represents a hypothetical position corresponding to the grommet, and each particle is assigned an initial weight, and the initial weights corresponding to each particle are equal, and the weight corresponding to each particle is a parameter representing the accuracy of the hypothetical position.

[0066] Initially, the particles are randomly distributed in the search area, and each particle is assigned an equal initial weight, indicating that all position hypotheses have the same possibility at the beginning.

[0067] Sb: based on the static model, the next position corresponding to each particle in the particle set is predicted, and the second position information of the corresponding particle is updated according to each predicted position.

[0068] The static model predicts the possible position of each particle in the particle set at the next time (or next step), and updates the second position information of the particle according to the predicted position.

[0069] Sc: For each particle, the corresponding feature matching degree of the particle is obtained based on the updated second position information of the particle, and the corresponding update weight of the particle is obtained based on the feature matching degree, which is used to replace the initial weight as the new weight of the particle.

[0070] After updating the second position information of the particle, the feature matching degree of each particle can be obtained by calculating the matching degree of the second position information of the particle and the feature of the mouth (such as the edge, texture, etc. in the ultrasound image), and then the corresponding particle is re- assigned an update weight according to the feature matching degree. Exemplarily, the particle with high feature matching degree is assigned a higher update weight, indicating that the second position information of the particle is more likely to be the real position of the mouth.

[0071] Sd: Resample based on the update weight of each particle to obtain a new particle set.

[0072] As time goes on, the update weight of some particles will become very low, and the update weight of some other particles will become very high. In order to avoid wasting computing resources on particles with low update weight, resampling is needed. In the resampling process, particles with high update weight can be selected multiple times, while particles with low update weight can be ignored or selected only once. In this way, the new particle set will be more concentrated around the real position of the mouth.

[0073] Se: Repeat Sb to Sd based on the new particle set until a preset end condition is reached, and output the final first position information based on the updated second position information and the update weight of the particles in the new particle set.

[0074] Thus, the feature extraction and accurate identification of the spatial position corresponding to the detected mouth are realized, which provides a reliable data basis for the subsequent optimization of the moving path.

[0075] The preset end condition can be that the number of repeated execution reaches a preset iteration number, or that the updated second position information of a preset number of particles in the new particle set reaches position estimation convergence.

[0076] Due to the randomness of the particles and the dynamics of the system, the second position information of the particles in the particle set will change in each update iteration. However, when the position estimation converges, the distribution of the particle set can stably reflect the real position of the gasket, that is, a preset number of particles in the particle set can be concentrated near the real position of the gasket, and such distribution no longer changes significantly.

[0077] In this embodiment, the weighted average position can be calculated according to the updated second position information of all particles and the update weight, and the weighted average position can be taken as the first position information of the gasket, wherein the particles with higher update weight have greater contribution to the final first position information.

[0078] As shown in FIG. 3, in this embodiment, taking the first heuristic search algorithm A* heuristic search algorithm as an example, S101 can further include the following processing:

[0079] S1011: discretize the position region of the gasket into a series of nodes, each node representing a potential monitoring point, create an open list and a closed list, and add the starting node to the open list.

[0080] In the open list (K i ,K j ) and the closed list (G i ,G j ), i, j = 0...N.

[0081] The node is the basic unit of the search algorithm. Two lists are created to manage the nodes in the search process, the open list can contain nodes to be evaluated, and the closed list can contain nodes that have been evaluated. The starting node (initial position of the ultrasonic detector) is added to the open list as the starting point of the search.

[0082] S1012: construct a reachability model between nodes according to the movement ability of the ultrasonic detector.

[0083] S1013: define an evaluation function, which can be expressed as:

[0084] g(n) represents the actual distance from the starting node (K0, K0) to the current node , h(n) represents the predicted distance from the current node to the target node , e(n) represents the energy consumption through the target node , and a, b and g represent weight coefficients for adjusting the influence of each part on the total evaluation.

[0085] S1014: perform a loop search.

[0086] S1014 can comprise repeating Sf, Sg and Sh when the open list is not empty.

[0087] Sf: selecting a node with the lowest value from the open list as the current node

[0088] Sg: moving the current node to the closed list;

[0089] Sh: evaluating the neighboring nodes of the current node .

[0090] Sh can comprise:

[0091] if the neighboring node is in the closed list or is determined to be unreachable based on the reachability model, skip;

[0092] otherwise, calculating a path from the start node (K0, K0) to the neighboring node through the current node ;

[0093] if the neighboring node is not in the open list, adding it to the open list;

[0094] if the neighboring node is in the open list but the new path is better, updating the information of the current node ;

[0095] recording the neighboring nodes of the current node for subsequent path reconstruction.

[0096] S1015: moving path reconstruction and performing simulated annealing optimization on the calculated moving path.

[0097] when the target node is added to the open list, it indicates that a path is found, and the complete moving path can be reconstructed by backtracking from the target node to the start node (K0, K0) through the neighboring node information, and the calculated moving path is subjected to simulated annealing optimization. The moving path is converted into control instructions to drive the ultrasonic detector to detect along the moving path, and when all potential monitoring points of the girth joint are detected, the girth joint detection is completed.

[0098] In this embodiment, S1012 can comprise the following processing:

[0099] ​The environment of the location area of the patch is converted into a discrete grid structure, the movement ability of the ultrasonic detector is determined according to the physical characteristics of the ultrasonic detector, and the size and shape of the grid structure are determined according to the movement ability of the ultrasonic detector, including the movement direction, the movement distance and the turning limit.

[0100] The patch location area is converted into a discrete grid structure, which facilitates modeling and calculation.

[0101] The boundaries of the explicitly detected area are determined, and the obstacles in the non-passable area are marked in the grid structure, including pipelines, equipment and other fixed structures.

[0102] By setting a specific value at the position of the obstacle in the grid structure, the obstacle coordinates can be marked, and marking the obstacles in the non-passable area in the grid structure can ensure that the generated movement path will not pass through these areas.

[0103] An r x r matrix is created, where r represents the total number of nodes in the environment, for each pair of nodes (p, q), p, q = 1... r, if node q is within the movement range of node p, and there is no obstacle between nodes q, mark the corresponding position in the matrix as reachable.

[0104] The creation of the matrix can be used to record the reachability between nodes, and if there is no obstacle between two nodes and the distance is within the movement range of the probe, they are considered to be reachable.

[0105] According to the reachability information between nodes, a graph T = (V, E) is constructed, where V is the node set, representing each node corresponding to a specific position in the physical space; E is the edge set, representing the connection relationship between nodes. For each pair of nodes in the graph, if the distance between the nodes is less than or equal to the movement distance of the probe, and there is no obstacle on the connection between the nodes, an edge is added between the nodes.

[0106] Based on the reachability information between nodes, a graph is constructed to represent the connection relationship between nodes. Each edge in the graph represents a possible movement path between nodes.

[0107] The energy consumption and time consumption between nodes are calculated, and the cost information is included in the critical matrix, and a cost value is assigned to each reachable node.

[0108] A cost value is assigned to each reachable node, which can include energy consumption, time consumption and other factors, which are used to evaluate the pros and cons of the path. The critical matrix is used to store the reachability information between nodes. When checking whether the nodes are reachable, the entries in the critical matrix can be used. When checking whether the nodes are reachable, the corresponding entry in the critical matrix can be found according to the index (or identifier) of the node.

[0109] The movement path of the ultrasonic detector is calculated by the A* heuristic search algorithm, the entire location area of the joint coating is efficiently and comprehensively covered, the detection efficiency and coverage rate are improved, and the accuracy and reliability of the detection process are ensured.

[0110] In this embodiment, the training method of the recognition model can include the following processes:

[0111] An ultrasonic image data set of the joint coating is obtained, the ultrasonic image data set of the joint coating is preprocessed, and the preprocessed ultrasonic image data set of the joint coating is divided into a training set and a validation set. Each ultrasonic image data in the training set and the validation set is labeled to determine the joint coating recognition information corresponding to each ultrasonic image data. The initial model is trained based on the training set, and the trained initial model is verified by the validation set to obtain the trained recognition model.

[0112] Therefore, the trained recognition model can automatically identify the joint coating recognition information corresponding to the first ultrasonic image, improving the accuracy and efficiency of ultrasonic image recognition.

[0113] The deep neural network model is trained by a large amount of ultrasonic image data of the joint coating to enable the deep neural network model to automatically learn and identify the feature patterns of the ultrasonic image data of the joint coating. The preprocessing of the joint coating ultrasonic image data set includes size adjustment, pixel value normalization, contrast enhancement, and other processes to reduce the computational burden during model training and improve recognition accuracy.

[0114] After preprocessing the joint coating ultrasonic image data set, the ultrasonic image data in the ultrasonic image data set needs to be labeled to clearly indicate the joint coating recognition information corresponding to each ultrasonic image data.

[0115] The initial model is trained using the labeled training set to extract ultrasonic image features from the input ultrasonic image data and map them to the correct joint coating recognition information. The weights in the network are adjusted using the backpropagation algorithm, the cross-entropy loss function is used to measure the classification performance index, the RMSprop algorithm is used to optimize the parameters of the deep neural network model, and the formula of the RMSprop algorithm can be expressed as:

[0116] v t denotes the exponentially weighted moving average of the squared gradient, ω denotes the decay factor, which is usually set to 0.9, g t denotes the gradient at time step t, θ t denotes the parameter model at time step t, η denotes the learning rate, and ε denotes a constant to avoid division by zero.

[0117] The performance of the deep neural network model is evaluated using each ultrasound image data in the verification set, and the deep neural network model structure, hyperparameters and training strategy are adjusted to improve the generalization ability of the deep neural network model.

[0118] In the embodiment, if the joint identification information is determination abnormal information, the method further comprises: obtaining a target repair scheme corresponding to the determination abnormal information to repair the joint according to the target repair scheme.

[0119] The repair scheme is a scheme for repairing the joint. When the joint has an abnormality, the joint can be repaired according to the target repair scheme corresponding to the determination abnormal information, thereby improving the standardization degree of the repair work of the joint and the repair quality of the joint.

[0120] The target repair scheme corresponding to the determination abnormal information can include the following processing: obtaining a current key feature corresponding to the determination abnormal information, querying the target repair scheme corresponding to the current key feature based on the established standard scheme database, and the standard scheme database includes repair schemes corresponding to different key features.

[0121] Through the standard scheme database, the target repair scheme corresponding to the determination abnormal information can be queried, the accuracy of the target repair scheme is improved, and the repair effect of the joint is ensured.

[0122] In the embodiment, establishing the standard scheme database can include:

[0123] Determine the data types, query types and user groups that the standard scheme database needs to support, and identify the data sources;

[0124] Collect and organize historical repair cases, including repair schemes, material types used and repair effects, obtain industry standards and specification documents of joint bonding quality;

[0125] Data cleaning is performed on the collected historical repair cases, key features are extracted from complex data, and data is converted into a unified format for storage and query. Data cleaning includes removing inconsistent, incorrect and incomplete data, and key features can include material type, abnormal type, abnormal distribution, abnormal degree and repair scheme;

[0126] Design an entity relationship model of the standard scheme database, including entities, attributes and relationships, set the storage structure of the data to a relational database, set a hash index to improve query efficiency, and set the access permission of the data according to the user role to read only, read and write, and manage.

[0127] The data types can include structured data, semi-structured data and unstructured data, the structured data can include tables, numbers and texts, the semi-structured data can include XML and JSON, and the unstructured data can include images, audios and videos.

[0128] The query types can include simple queries, complex queries, full-text searches and real-time queries. The user groups can include detection engineers, maintenance personnel, R&D personnel and regulatory personnel. The data sources can include internal data sources and external data sources.

[0129] By the correspondence between the key features and the repair schemes in the standard scheme database, the target repair scheme corresponding to the current key feature can be queried.

[0130] In some embodiments, the target repair scheme corresponding to the determination abnormal information can be obtained by: obtaining the current key feature corresponding to the determination abnormal information, and obtaining the target repair scheme corresponding to the current key feature based on the trained neural network model. As an example, a conditional generative adversarial network model can be used, the input is the current key feature corresponding to the abnormal information, and the output is the repair scheme. The training data can include a historical abnormal sample library containing features of different repair joints and corresponding successful repair scheme records.

[0131] In the embodiment, after the target repair scheme corresponding to the determination abnormal information is obtained, the second ultrasonic image corresponding to the repair joint at the time when the repair is completed is obtained by the ultrasonic detector; it is judged whether the repair joint identification information corresponding to the second ultrasonic image is the determination abnormal information; if yes, the target repair scheme in the standard scheme database is modified based on the second ultrasonic image and the determination abnormal information.

[0132] In the repair process, the ultrasonic detection equipment is used to collect detection data, and according to the actual ultrasonic detection result, it is determined whether the repair quality of the repair joint reaches the predetermined standard in the industry standard and specification document of the repair joint bonding quality. If the predetermined standard is reached, the repair joint identification information corresponding to the second ultrasonic image is the determination normal information, and if the predetermined standard is not reached, the repair joint identification information corresponding to the second ultrasonic image is the determination abnormal information.

[0133] When the repair joint identification information corresponding to the second ultrasonic image is the determination abnormal information, the repaired repair joint has defects, and the detected second ultrasonic image and the determination abnormal information can be fed back to the standard scheme database for rewriting of the standard scheme database, so as to improve the reliability and effectiveness of the overall repair work.

[0134] In the embodiment, after the patch is repaired based on the target repair scheme, the method further includes the following processing: if the repaired patch needs to be rechecked, obtaining a first position coordinate corresponding to the ultrasonic detector and a second position coordinate corresponding to the patch; based on a preset second heuristic search algorithm, calculating a rechecking path corresponding to the first position coordinate and the second position coordinate; obtaining a third ultrasonic image corresponding to the repaired patch collected by the ultrasonic detector in the rechecking along the rechecking path; comparing the third ultrasonic image with a preset ultrasonic image corresponding to the repaired patch, and based on the comparison result, judging whether the rechecking position corresponding to the patch is accurate.

[0135] According to the second position coordinate of the repaired patch and the first position coordinate of the ultrasonic detector, a shortest and obstacle-free optimal rechecking path can be calculated by using a fast path planning algorithm, so as to ensure that the ultrasonic detector can quickly and accurately move to the position of the repaired patch along the rechecking path, and efficient rechecking is realized.

[0136] The second heuristic search algorithm can also be an A* heuristic search algorithm, and based on the preset second heuristic search algorithm, the rechecking path corresponding to the first position coordinate and the second position coordinate can include:

[0137] Obtaining the first position coordinate (x c ,y c ) and all second position coordinates (x i ,y i ) that need to be rechecked, and taking all the second position coordinates that need to be rechecked as target coordinates;

[0138] An A* heuristic search algorithm is used to calculate the shortest path, and a node network is constructed, wherein each node represents a position point, the edges between the nodes represent the feasibility of movement, and the weight of the edge represents the movement cost;

[0139] An evaluation function is defined, which can be expressed as: f(n)=g(n)+h(n)

[0140] g(n) is the actual distance from the starting point to the current node, Manhattan distance is used as the movement cost, so g(n)=d(start,n), h(n) is the estimated distance from the current node to the target node, based on heuristic estimation, Euclidean distance is defined, so x n ,y n represent the horizontal coordinate and the vertical coordinate of the current node n in the two-dimensional space respectively, and x goal ,y goal represent the horizontal coordinate and the vertical coordinate of the target node in the two-dimensional space respectively;

[0141] An A* heuristic search algorithm is performed, and a node with the lowest f(n) value is selected as the next visited node until the target node is reached, thereby generating the reinspection path.

[0142] The entire search area is abstracted as a network of nodes, each of which represents a location point (which can be the start node, the target node, or any point on the path). The edges between the nodes represent the feasibility of moving from one node (location) to another. The weight of the edge represents the cost (or expense) of moving from one node to another. In this embodiment, the Manhattan distance is used to calculate the actual movement cost.

[0143] The Manhattan distance represents the distance from one point to another when movement is only possible along grid lines. The Euclidean distance provides an estimate of the straight-line distance between two points.

[0144] The electronic device is configured with an image-guided strategy, which includes acquiring a real-time third ultrasound image of the repaired joint gap position by an ultrasound detector, and performing preprocessing such as denoising and enhancement on the third ultrasound image to improve the clarity of the third ultrasound image.

[0145] The electronic device stores a preset ultrasound image corresponding to the repaired joint gap, which is used as a standard ultrasound image. The Canny edge detection algorithm can be used to identify the outline of the joint gap in the third ultrasound image, and the third ultrasound image can be compared with the preset ultrasound image to verify the accuracy of the reinspection position corresponding to the joint gap using the structural similarity index.

[0146] Through the image-guided strategy, the ultrasound detector can be efficiently and accurately moved to the position of the joint gap that needs to be rechecked after repair during the reinspection process, and high-precision reinspection verification can be achieved.

[0147] Based on the same technical concept, the present disclosure also provides an ultrasonic detection system applied to joint gap detection, as shown in FIG. 4. The ultrasonic detection system 200 applied to joint gap detection can include:

[0148] A path planning unit 201 is configured to acquire first position information corresponding to a joint gap, and calculate a movement path of an ultrasound detector corresponding to the first position information based on a preset first heuristic search algorithm.

[0149] An acquisition unit 202 is configured to acquire a first ultrasound image of the joint gap collected by the ultrasound detector based on the movement path.

[0150] An anomaly identification unit 203 is configured to acquire joint gap identification information corresponding to the first ultrasound image based on a trained identification model, wherein the joint gap identification information is determination abnormal information or determination normal information.

[0151] As shown in FIG. 4 and FIG. 5, in some embodiments, the path planning unit 201 comprises a target positioning unit and a path optimization unit.

[0152] The target positioning unit can comprise:

[0153] An initial sub-module is configured to initialize a set of particles based on the search area, to obtain a particle set, the second position information of each particle in the particle set representing a hypothetical position corresponding to the joint, and to assign an initial weight to each particle, the initial weight of each particle being equal, and the weight corresponding to each particle being a parameter representing the accuracy of the hypothetical position.

[0154] A prediction updating sub-module is configured to predict the position of each particle in the particle set at the next step based on a static model, and to update the second position information of each particle according to the predicted position.

[0155] A weight updating sub-module is configured to, for each particle, obtain the joint feature matching degree corresponding to the particle based on the updated second position information of the particle, and obtain the updated weight corresponding to the particle based on the joint feature matching degree, the updated weight being used to replace the initial weight as the new weight corresponding to the particle.

[0156] A resampling sub-module is configured to resample each particle based on the updated weight corresponding to the particle, to obtain a new particle set.

[0157] An iteration sub-module is configured to repeat the processing of the prediction updating sub-module to the resampling sub-module based on the new particle set until a preset ending condition is reached, and to output the final first position information based on the updated second position information and the updated weight of each particle in the new particle set.

[0158] In some embodiments, the first heuristic search algorithm can be an A* heuristic search algorithm, and the path optimization unit is configured to calculate the movement path based on the A* heuristic search algorithm.

[0159] In some embodiments, the anomaly identification unit 203 comprises an anomaly recognition unit, and the recognition model can be a deep neural network model. The anomaly recognition unit is configured to obtain the joint recognition information based on the trained deep neural network model.

[0160] The training device of the recognition model comprises:

[0161] An acquisition processing module is configured to obtain a joint ultrasound image dataset, to pre-process the joint ultrasound image dataset, and to divide the pre-processed joint ultrasound image dataset into a training set and a validation set.

[0162] The labeling module is configured to label each of the ultrasound image data in the training set and the validation set to determine corresponding joint recognition information of each of the ultrasound image data.

[0163] The training and verification module is configured to train an initial model based on the training set and verify the trained initial model based on the validation set to obtain the trained recognition model.

[0164] In some embodiments, if the joint recognition information is abnormal judgment information, the system further comprises a repair strategy unit and a quality monitoring unit.

[0165] The repair strategy unit is configured to obtain a target repair scheme corresponding to the abnormal judgment information to repair the joint based on the target repair scheme.

[0166] The repair strategy unit further comprises:

[0167] The query submodule is configured to obtain a current key feature corresponding to the abnormal judgment information, and query the target repair scheme corresponding to the current key feature based on the established standard scheme database, wherein the standard scheme database comprises repair schemes corresponding to different key features.

[0168] The repair strategy unit further comprises:

[0169] The first acquisition submodule is configured to acquire a second ultrasound image corresponding to the repaired joint collected by the ultrasonic detector when the repair is completed.

[0170] The quality monitoring unit is configured to determine whether the joint recognition information corresponding to the second ultrasound image is the abnormal judgment information, and if so, the standardized correction unit is entered.

[0171] The standardized correction unit is configured to correct the target repair scheme in the standard scheme database based on the second ultrasound image and the abnormal judgment information.

[0172] In some embodiments, after the joint is repaired based on the target repair scheme, the system further comprises a re-inspection positioning unit, and the re-inspection positioning unit comprises a re-inspection planning unit and a re-inspection execution unit.

[0173] The re-inspection planning unit comprises:

[0174] The second acquisition submodule is configured to acquire a first position coordinate corresponding to the ultrasonic detector and a second position coordinate corresponding to the joint when the repaired joint needs to be re-inspected.

[0175] The calculation submodule is configured to calculate a re-inspection path corresponding to the first position coordinate and the second position coordinate based on a preset second heuristic search algorithm.

[0176] The reinspection execution unit comprises:

[0177] The third acquisition sub-module is configured to acquire a third ultrasonic image corresponding to the repaired gusset joint collected by the ultrasonic detector during reinspection along the reinspection path.

[0178] The comparison and judgment sub-module is configured to compare the third ultrasonic image with a preset ultrasonic image corresponding to the repaired gusset joint, and judge whether the reinspection position corresponding to the gusset joint is accurate based on the comparison result.

[0179] In one example, the modules in any of the above systems can be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASICs), or one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.

[0180] For another example, when the modules in the system can be implemented in the form of a processing element scheduler, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor that can invoke programs. For another example, these modules can be integrated together to implement in the form of a system-on-a-chip (SOC).

[0181] In the present disclosure, various objects such as messages / information / devices / network elements / systems / devices / actions / operations / processes / concepts, etc. are named. It can be understood that these exemplary names do not constitute a limitation on the relevant objects, and the assigned names can be changed according to factors such as scene, context or usage habits. The technical meaning of the technical terms in the present disclosure should be mainly determined from the function and technical effect embodied / implemented in the technical scheme.

[0182] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the example working processes of the above-described system, device and module can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0183] Those of skill in the art would understand that the modules and algorithms described in connection with the examples disclosed herein can be implemented in electronic hardware, computer software, or combinations of both. The disclosure encompasses both hardware and software implementations of examples described herein. The implementation decisions are driven by cost constraints, consideration of manufacturing technology, and / or other constraints as would be understood by those of ordinary skill in the art. The disclosure therefore does not exclusively describe only hardware implementations or software implementations.

[0184] Based on the same technical concept, the disclosure also provides an electronic device, as shown in FIG. 6, which includes a processor 301 and a memory 302, and can further include one or more of an information input / output (I / O) interface 303, a communication component 304, and a communication bus 305.

[0185] The processor 301 is configured to control the overall operation of the electronic device 300 to complete the ultrasonic detection method for joint detection described above; the memory 302 is configured to store various types of data to support the operation of the electronic device 300, which can include, for example, instructions for operating any application or method on the electronic device 300, and application-related data. The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as one or more of a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk, or an optical disk.

[0186] The I / O interface 303 provides an interface between the processor 301 and other interface modules, which can be a keyboard, a mouse, a button, etc. The buttons can be virtual buttons or physical buttons. The communication component 304 is configured to test wired or wireless communication between the electronic device 300 and other devices. The wireless communication, for example, Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination of one or more of them, and thus the communication component 304 can include a Wi-Fi component, a Bluetooth component, and an NFC component.

[0187] The communication bus 305 can include a path for transmitting information between the components. The communication bus 305 can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 305 can be divided into an address bus, a data bus, a control bus, etc.

[0188] The electronic device 300 can be implemented with one or more ASICs (Application Specific Integrated Circuits), DSPs (Digital Signal Processors), DSPDs (Digital Signal Processing Devices), PLDs (Programmable Logic Devices), FPGAs (Field Programmable Gate Arrays), controllers, microcontrollers, microprocessors, or other electronic elements for executing the ultrasonic testing method for a joint detection according to the embodiments.

[0189] The electronic device 300 can include, but is not limited to, a mobile terminal such as a digital broadcast receiver, a PDA (Personal Digital Assistant), a PMP (Portable Multimedia Player), etc., and a stationary terminal such as a digital TV (Television), a desktop computer, etc., and can also be a server, etc.

[0190] Based on the same technical concept, the disclosure also provides a computer readable storage medium (including a non-transitory computer readable storage medium) having a computer program stored thereon, the computer program being executed by a processor to implement the ultrasonic detection method for joint coating detection.

[0191] The computer readable storage medium can include a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0192] The terms "comprising", "containing" or any other variant thereof are intended to cover a non-exclusive inclusion, so that a process, method, article or apparatus that comprises a list of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or apparatus.

[0193] In addition, the terms "first", "second", and the like are used only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second", and the like can explicitly or implicitly include at least one of the features. In the description of the disclosure, the meaning of "a plurality of" is at least two, such as two, three, and the like, unless otherwise explicitly limited.

[0194] In the description of the present specification, the description referring to the terms "one / some embodiments", "some embodiments", "examples", "specific examples", or "some examples" and the like means that some features, structures, materials or characteristics described in connection with the embodiments or examples are included in at least one embodiment or example of the present disclosure. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, some features, structures, materials or characteristics described can be combined in any suitable manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine different embodiments or examples described in the present specification and the features of different embodiments or examples, without contradiction.

[0195] In the description of the present disclosure, unless otherwise stated, " / " means "or", for example, A / B can mean A or B. It can be understood that the functions, steps, etc. shown in the present disclosure can occur in an order different from that shown in the present disclosure, and there can be other functions, steps, etc. between the two adjacent functions, steps, etc. shown in the present disclosure.

[0196] Although the embodiments of the present disclosure have been shown and described above, it is understood that the above-described embodiments are exemplary, and it is not construed that the present disclosure is limited to the above-described embodiments, and a person of ordinary skill in the art can make changes, modifications, replacements, and variations to the above-described embodiments within the scope of the present disclosure.

Claims

1. An ultrasonic detection method applied to joint detection, comprising: obtaining first position information corresponding to a joint, and calculating a moving path of an ultrasonic detector corresponding to the first position information based on a preset first heuristic search algorithm; obtaining a first ultrasonic image corresponding to the joint collected by the ultrasonic detector based on the moving path; obtaining joint identification information corresponding to the first ultrasonic image based on a trained identification model, the joint identification information being abnormal information or normal information; 2. The ultrasonic testing method for detecting a patch according to claim 1, wherein if the joint identification information is abnormal information, the method further comprises: obtaining a target repair scheme corresponding to the abnormal information to repair the joint according to the target repair scheme.

3. The ultrasonic testing method for detecting a patch according to claim 2, wherein The method further comprises: obtaining a target repair scheme corresponding to the abnormal information, comprising:

4. The ultrasonic testing method for a splicing detection according to claim 3, wherein, obtaining a current key feature corresponding to the abnormal information, querying the target repair scheme corresponding to the current key feature based on a standard scheme database established, and the standard scheme database comprising repair schemes corresponding to different key features. After obtaining the target repair scheme corresponding to the abnormal information, the method further comprises: obtaining a second ultrasonic image corresponding to the joint collected by the ultrasonic detector when the repair is completed; determining whether the joint identification information corresponding to the second ultrasonic image is the abnormal information; 5. An ultrasonic testing method for patch detection according to any one of claims 2 to 4, wherein, if yes, modifying the target repair scheme in the standard scheme database based on the second ultrasonic image and the abnormal information. After the joint is repaired based on the target repair scheme, the method further comprises: if the repaired joint needs to be rechecked, obtaining a first position coordinate corresponding to the ultrasonic detector and a second position coordinate corresponding to the joint; calculating a rechecking path corresponding to the first position coordinate and the second position coordinate based on a preset second heuristic search algorithm; obtaining a third ultrasonic image corresponding to the repaired joint collected by the ultrasonic detector during rechecking according to the rechecking path; 6. An ultrasonic testing method for patch detection according to any one of claims 1 to 5, wherein, comparing the third ultrasonic image with a preset ultrasonic image corresponding to the repaired joint, and determining whether the rechecking position corresponding to the joint is accurate based on the comparison result. The method further comprises: Sa: initializing a group of particles based on a search area to obtain a particle set, the second position information of each particle representing a hypothetical position corresponding to the joint, and each particle being assigned an initial weight, the initial weight of each particle being equal, and the weight of each particle being a parameter representing the accuracy of the hypothetical position; Sb: predicting the next position of each particle in the particle set based on a static model, and updating the second position information of each particle according to the predicted position; Sc: for each particle, obtaining a joint feature matching degree corresponding to the particle based on the updated second position information of the particle, and obtaining an updated weight corresponding to the particle based on the joint feature matching degree, the updated weight being used to replace the initial weight as a new weight corresponding to the particle; Sd: resample the particles based on the updated weights of the respective particles to obtain a new set of particles; Se: repeat Sb to Sd based on the new set of particles until a preset ending condition is reached, and output the final first position information based on the updated second position information and the updated weights of the particles in the new set of particles.

7. An ultrasonic testing method for patch detection according to any one of claims 1 to 6, wherein, The training method of the recognition model comprises: obtaining a gasket ultrasonic image dataset, preprocessing the gasket ultrasonic image dataset, and dividing the preprocessed gasket ultrasonic image dataset into a training set and a validation set; annotating each ultrasonic image data in the training set and the validation set to determine the gasket recognition information corresponding to each ultrasonic image data; training an initial model based on the training set and verifying the trained initial model through the validation set to obtain the trained recognition model.

8. An ultrasonic detection system applied to gasket detection, comprising: a path planning unit configured to obtain first position information corresponding to a gasket, and calculate a moving path of an ultrasonic detector corresponding to the first position information based on a preset first heuristic search algorithm; an acquisition unit configured to obtain a first ultrasonic image corresponding to the gasket collected by the ultrasonic detector based on the moving path; an abnormality identification unit configured to obtain gasket recognition information corresponding to the first ultrasonic image based on a trained recognition model, the gasket recognition information being abnormality determination information or normality determination information.

9. An electronic device comprising a processor and a memory, wherein, The processor is coupled to the memory; The processor is configured to execute a computer program stored in the memory, so that the electronic device executes the method according to any one of claims 1 to 7.

10. A computer readable storage medium comprising a computer program or instructions, wherein, When the computer program or instructions run on the computer, the computer executes the method according to any one of claims 1 to 7.

11. A computer program product comprising computer instructions which, when executed by a processor, implement the method according to any one of claims 1 to 7.

12. A computer program comprising computer instructions which, when executed by a processor, implement the method according to any one of claims 1 to 7.

Images