A large equipment abnormal component cooperative detection method based on multiple robots

Abstract

The application belongs to the technical field of intelligent equipment fault detection, and discloses a large-scale equipment abnormal component cooperative detection method based on multiple robots. The large-scale complex intelligent equipment abnormal component detection method based on multiple robots and cloud edge cooperative architecture utilizes clustering, A* algorithm, shortest route distribution and Kalman filter fusion algorithm to realize rapid inspection and accurate detection of abnormal components by multiple robots. Through uniform distributed inspection by multiple robots, establishment of a global three-dimensional coordinate system, secondary inspection of abnormal points, clustering analysis screening of abnormal points, task list distribution, observable path division fitting, optimal detection scheme construction, and finally fusion of multiple-angle multi-modal abnormal data, accurate detection results are obtained. Through uniform distribution of multiple robots at equal intervals around the large-scale complex intelligent equipment for inspection, the detection efficiency is improved, and the coverage range is wider, and abnormal information can be fed back in time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent equipment fault detection, and in particular to a large equipment abnormal component cooperative detection method based on multiple robots. BACKGROUND

[0002] Large intelligent equipment usually runs in an extreme environment such as high temperature and high pressure, which is difficult for personnel to approach. In order to ensure its efficient and stable operation, an industrial inspection robot plays an indispensable role. Industrial large and complex intelligent equipment is prone to failure and extremely dangerous. In order to ensure its safe, reliable, continuous and stable operation, the operation prediction and accurate operation and maintenance of such equipment need to be realized through an inspection robot. The detection of abnormal components of large and complex intelligent equipment is an important part of promoting industrial intelligent operation and maintenance, which helps to improve factory production efficiency, product quality, and reduce factory safety risks and labor costs. However, efficient and accurate detection of large and complex intelligent equipment components is still a difficult problem to be solved.

[0003] Cloud-edge collaboration is the close cooperation of cloud computing and edge computing. Through the cooperative work of the cloud and the edge, the optimal allocation and efficient use of resources are realized. The cloud-edge collaboration framework includes three main capabilities: resource collaboration, data collaboration and service collaboration. Resource collaboration involves the sharing and scheduling of computing power, storage and other resources of the cloud and edge; data collaboration focuses on the optimization of data collection, processing, analysis and storage processes; service collaboration emphasizes the seamless integration and unified management of cloud and edge services. Based on the cloud-edge collaboration architecture, the cooperative detection of abnormal components of complex equipment is quite critical.

[0004] However, at present, due to the fact that industrial large and complex intelligent equipment is prone to failure and extremely dangerous, manual inspection is time-consuming and cannot achieve real-time monitoring. Moreover, when the large and complex intelligent equipment is in an extreme environment, manual inspection is dangerous; the coverage of a single robot is low, and the time for a single cycle of detection is long, which cannot achieve real-time monitoring. SUMMARY

[0005] The present application aims to provide a large equipment abnormal component cooperative detection method based on multiple robots, which can improve the problem that a single robot cannot detect abnormalities in time and the problem of low efficiency of multiple robot cooperative detection in the existing intelligent equipment detection technology.

[0006] This invention proposes a collaborative detection method for abnormal components in large equipment based on multiple robots. This method improves the efficiency of abnormal component detection by having multiple robots collaboratively inspect the area around the large equipment. Simultaneously, based on a cloud-edge collaborative architecture, this method uploads a large amount of multimodal abnormal data collected by multiple robots (at the edge) from different perspectives to the cloud. The cloud then performs multimodal and multi-perspective abnormal data fusion detection, improving the accuracy of abnormal detection. Building upon this, after an anomaly occurs, this invention utilizes robots to find and mark the optimal observable location points of the abnormal components around the large equipment. All observable location points are integrated and divided to obtain the optimal observable path. Finally, tasks are assigned and paths are planned for the robots to travel along different observable paths to collect abnormal data, improving both the accuracy of anomaly detection and the system's operational efficiency.

[0007] The technical solution adopted in this invention is as follows: A collaborative detection method for abnormal components of large equipment based on multiple robots. The method sets the number of robots to n (n≥2) according to the size and complexity of the large equipment, and numbers the robots sequentially from 1 to n. The robots are evenly distributed around the large equipment for inspection. Each robot is equipped with three types of sensors: a visible light camera, a thermal infrared camera, and an audio-visual unit. During the inspection process, the robot scans the equipment to acquire visible light images. vis Thermal infrared image I tir Harmony and Acoustic Image I ac The system is based on a cloud-edge collaborative architecture, with the robot acting as an edge device to collect I... vis I tir and I ac After receiving the three types of image data, the image data is uploaded in real time to a cloud platform with corresponding image detection and analysis functions for processing. The cloud platform can perform anomaly detection based on image processing on the received image data to determine whether large equipment has malfunctioned. The collaborative detection of abnormal components of multiple robots includes two stages: the first stage is the inspection stage, and the second stage is the collaborative detection stage.

[0008] Furthermore, the inspection phase includes: cruise detection, where the robot inspects the equipment along a designated inspection direction, and establishes a three-dimensional XYZ coordinate system based on the two-dimensional XY horizontal plane where the large equipment base is located; wherein, from the top-down view of the large equipment, the counterclockwise inspection direction is defined as the positive direction, and the clockwise inspection direction as the negative direction; abnormal data collection, during which, when the cloud receives the image data collected by a single sensor of the i-th robot at time t, and the image detection and analysis function detects an abnormality in a component, the i-th robot uploads the image at time t. vis I tir and I acThe three types of image data are fused using Kalman filtering to obtain a composite image, which is then used for image detection and analysis to determine whether a component has malfunctioned. Once a component malfunction is confirmed, the cloud-based system calculates and marks the observed malfunction location coordinates P on the global XYZ three-dimensional coordinate system. it (x it ,y it ,z it ), calculate and mark the position coordinates R of the i-th robot in the two-dimensional XY plane coordinate system. it (x i ′ t ,y i ′ t The mapping relationship is recorded as f:R→P. When the image data uploaded by robot i at time t is found to be normal in the cloud, no position coordinates are calculated or marked. Observable points are calculated, and reliable anomaly points are selected in the cloud using a clustering algorithm. The observed anomaly position coordinates P are then recorded. it Perform clustering and group P in the global XYZ 3D coordinate system. it Divided into core point P Δ Three types of points are identified: boundary point P′ and noise point P″, and several clusters are obtained. and each cluster center of gravity The core point P Δ P represents the cluster it Furthermore, the number of outlier coordinates within the spherical neighborhood ε of this point with radius eps is greater than Pts. Cor Given the radius eps and the minimum number of points Pts in the neighborhood of the core point, Cor All are initial settings; boundary point P′ represents P in the cluster excluding the core point. it Noise point P″ represents P that is not in the cluster. it Different clusters Used to indicate the different abnormal components detected in this round of inspection, and their respective centers of gravity. Considered as the corresponding cluster Identify the anomaly in the component and re-mark it as P. anom Meanwhile, the cloud will transmit different anomaly points P anom The anomalies are randomly placed into the task list. In the subsequent collaborative detection phase, anomalies are retrieved from the task list from front to back for detection. If an anomaly P is found... anom Cluster The coordinates of the outlier point P are included. it The number is less than the set minimum number of points Pt required to satisfy collaborative detection. sClus Then this type of outlier P anom Do not add to the task list; the cloud will then determine the task based on each cluster. The coordinates of the anomaly location P itWith inverse mapping f -1 :P→R, derive the coordinates of P in the two-dimensional XY plane. it The corresponding robot position coordinates R it Then the coordinates of these robot positions are outlier points P. anom The observable locations are then remarked with the robot's position coordinates as observable location points Q. k (x k ,y k ); Observable path solution, which uses an observable path solution algorithm in the cloud to solve for each outlier point P. anom observable paths l1, l2, ..., l j ..., let θ be the angle between a point in a two-dimensional plane and the positive x-axis after connecting that point and the origin; the observable path solving algorithm first calculates the observable position points Q in the two-dimensional XY plane coordinate system. k The corresponding angle θ (0≤θ<2π) is determined, and each observable position point Q is assigned a corresponding angle θ (0≤θ<2π). k Sort the elements according to the size of their angles θ, with smaller angles first and larger angles last. If the (x+1)th Q in the sorting... k Angle θ χ+1 Subtract the x-th Q k Angle θ χ If the angle difference between adjacent observable points along the same path is less than or equal to the maximum angle difference Δθ, then the (x+1)th Q will be... k and the χth Q k Considered to be on the same observable path j In the middle, otherwise the (x+1)th Q k and the χth Q k In different observable paths, different observable paths are divided and fitted to obtain observable paths.

[0009] Furthermore, in the abnormal data acquisition step, the moment when the i-th robot first detects the abnormality is recorded as time 0, and the position coordinates of the i-th robot at time 0 are R. i0 (x i ′0,y i ′0), will be a point with the origin as its endpoint and passing through point R. i0 (x i ′0,y i The ray at '0' is set as the reference position. When the i-th robot reaches this ray α times, that is, after a certain robot first detects the anomaly, all robots continue to patrol around the large equipment α times, completing the anomaly position coordinate P. it (x it ,y it ,z it ) and robot position coordinates R it (x i ′ t,y i ′ t Once the abnormal data collection is completed (marked and uploaded), the abnormal data collection is finished; after completing the α-lap, a preliminary set of the robot's position coordinates is now available in the cloud. and the set of abnormal location coordinates And the mapping relationship f:R→P between set R and set P.

[0010] Furthermore, the collaborative detection phase includes: calculating the optimal detection scheme, which is performed in the cloud. This first requires identifying anomaly points P from the task list. anom We take objects from front to back for testing, and let P be the object we take out for testing. a ′ nom Let the object to be detected be P. a ′ nom There are a total of m observable paths, and l observable paths at the same time. j The observable points with the smallest and largest angles are set as endpoints, and the observable point with the smallest angle is denoted as... The observable point with the largest angle is denoted as Using the A* algorithm, find the path from the current position of robot i to path l. j endpoints and The shortest route distances are denoted as follows: and The distance cost from the robot's current position to each path endpoint is calculated using the A* algorithm. This yields the distances from each robot to the observable points with the minimum and maximum angles along each path. Cost matrix C1 is constructed using the distances from each robot to the observable points with the minimum angles along each path, and cost matrix C2 is constructed using the distances from each robot to the observable points with the minimum angles along each path. The final cost matrix C = g(C1, C2), where g represents C[p][q] = min(C1[p][q], C2[p][q]), meaning the element in the p-th row and q-th column of the final cost matrix C is equal to the minimum value of the elements in the p-th row and q-th column of C1 and C2. After obtaining the cost matrix, the optimal detection scheme for collaborative detection is solved using the shortest route allocation algorithm, minimizing the total path length for all robots. The optimal detection scheme is executed, assigning each robot to travel along the optimal route. Upon reaching an observable path, abnormal data is collected from abnormal components, and the collected data is uploaded to the cloud in real time. The cloud uses Kalman filtering to fuse the collected multimodal and multi-view abnormal data to obtain the abnormal point P. a ′ nom The final test results; completion of the current test target P a ′ nom After completing the detection task, check if there are any other tasks P to be detected in the task list. anomIf an anomaly exists, the newest anomaly point at the top of the list is selected as the target for the next round of detection; anomaly point P in the list... anom Once all tests are completed, the collaborative testing phase ends.

[0011] Furthermore, the shortest route allocation algorithm includes the following steps:

[0012] S1: Subtract the minimum value in each row of the cost matrix, and then subtract the minimum value in each column, thus ensuring that each row and each column has at least one zero.

[0013] S2: Cover all zeros in the matrix with as few row and column labels as possible. One row label will cover the entire row of elements in the matrix, and one column label will cover the entire column of elements in the matrix. If the sum of the number of row and column labels is less than min(n,m), proceed to step S3; if the number of row and column labels is equal to min(n,m), the current optimal solution is found, and step S4 is executed.

[0014] S3: Find the minimum value ξ among the elements not covered by row and column labels, subtract ξ from all elements not covered by row and column labels, add ξ to the elements covered by row and column labels twice, remove the row and column labels, and repeat step S2.

[0015] S4: After obtaining the current optimal solution, different outlier points P are detected. anom The relationship between the number of robots n and the number of observable paths m varies, and the system operation schemes also differ. By comparing the sizes of n and m, different schemes can be selected for operation.

[0016] S5: Add the total length of the current detection route of each robot and the distance from the end of the current detection route to the endpoint of the remaining path as the distance cost for multiple robots to reach the remaining observable path. At the same time, in the same way as the above method for constructing the cost matrix, reconstruct a new cost matrix, and repeat step S1 to solve the optimal detection scheme for collaborative detection.

[0017] Furthermore, step S4 selects different schemes by comparing the sizes of n and m, specifically including the following:

[0018] When n > m, the current optimal solution is taken as the optimal detection scheme for collaborative detection, and m robots are dispatched to m paths to detect the current target P. a ′ nom Collaborative detection is performed, and the remaining nm robots continue to inspect and collect abnormal data from large equipment;

[0019] When n = m, the current optimal solution is taken as the optimal detection scheme for collaborative detection, and n robots are dispatched to n paths to detect the current target P. a ′ nom Conduct collaborative detection;

[0020] When n < m, the current optimal solution is used as a partial solution for the overall shortest path plan of the current collaborative detection. n robots are dispatched to n paths respectively to detect the current detection target P a ′ nom for collaborative detection. The current detection route of the robot is recorded as the route to the specified path plus the route for detection on the specified path. Assume that the i-th robot is dispatched to the observable path l j , and it first passes through the observable point with the smallest angle The detection end point is the observable point with the largest angle Calculate the total length s of the current detection route of the i-th robot, that is, from the current position to the detection end point using the A* algorithm i and the shortest route distance from the detection end point of the current route to the end point of the remaining path and are respectively denoted as and the distances to, and step S5 is executed

[0021] Furthermore, the distances of each robot to the observable points with the smallest and largest angles on each path are used. A cost matrix C1 is constructed with the distances of each robot to the observable points with the smallest angles on each path, and a cost matrix C2 is constructed with the distances of each robot to the observable points with the largest angles on each path:

[0022]

[0023]

[0024] where is the shortest route distance from the current position of the i-th robot to the end point j of the path l , is the shortest route distance from the current position of the i-th robot to the end point j of the path l .

[0025] Furthermore, during the abnormal data collection, when a robot detects multiple abnormal points at time t, the positions of the robot and the abnormal points are marked respectively; if the i-th robot detects three abnormal points at time t, the abnormal position coordinates are marked as P it 1 (x it ,y it ,z it ), P it 2 (x it ,yit ,z it ) and P it 3 (x it ,y it ,z it ), and the robot's position is R. it 1 (x i ′ t ,y i ′ t ), R it 2 (x i ′ t ,y i ′ t ) and R it 3 (x i ′ t ,y i ′ t ), and record the one-to-one correspondence.

[0026] Furthermore, in the calculation of observable points, the core point P with a distance less than eps is included. Δ The boundary point P′ and the boundary point P′ are clustered in the same initial cluster. In the middle, according to the core point P Δ Calculate the initial cluster using the weights of boundary point P′. center of gravity Core point P Δ The weights are set to τ times the boundary point P′; this yields the initial cluster centroid. Then, select clusters Center and weight The core points P of the first v% with relatively close Euclidean distance Δ And the boundary point P′, the core point P after filtering Δ Together with boundary point P′, they form a cluster representing the anomalous component. And calculate clusters center of gravity

[0027] Furthermore, the current optimal solution in step S2 is the adjusted cost matrix, which satisfies that the number of row and column labels is equal to min(n,m). Specific solutions include the following:

[0028] If n≤m, then each row of the cost matrix has at least one zero element, and at least one zero element can be selected from each row to ensure that the selected zero elements are in different columns. The row and column where the selected zero element is located represent the observable path endpoint that the robot should go to in the corresponding column.

[0029] If n>m, then each column of the cost matrix has at least one zero element, and at least one zero element can be selected from each column to ensure that the selected zero elements are all in different rows. The row and column where the selected zero element is located represent the observable path endpoint that the robot should go to in the corresponding column.

[0030] Furthermore, the element mapping relationship between sets R and P is bijective, denoted as f: R→P, and has an inverse mapping f. -1 :P→R.

[0031] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0032] 1. By evenly distributing multiple robots around large equipment at equal intervals, the equipment is inspected in two stages: routine inspection and collaborative detection. This avoids the problem that a single robot may not be able to detect anomalies in time, improves detection efficiency, covers a wider range, and improves the system's response efficiency, enabling multi-robot collaborative detection.

[0033] 2. Based on the size and complexity of the on-site equipment, the number of robots is determined. Each robot is equipped with three types of sensors: a visible light camera, a thermal infrared camera, and an acoustic imager, acquiring visible light images, thermal infrared images, and acoustic images, respectively. Simultaneously, based on a cloud-edge collaborative architecture, the robots, acting as edge devices, collect the three types of image data and upload them in real time to the cloud, which has corresponding image detection and analysis capabilities for processing. The cloud can perform anomaly detection based on image processing to determine whether large equipment has malfunctioned. The number of robots is set according to the equipment situation to quickly determine the fault condition, obtain the optimal observable path, and then take improvement measures. This improves both the accuracy and efficiency of anomaly detection.

[0034] 3. By clustering the abnormal location coordinates, the collected abnormal location coordinates are filtered to obtain more reliable abnormal location coordinates, and more reliable observable location points are deduced from them; and by using the observable path solving algorithm, the angle of the above-observable location points on the two-dimensional horizontal plane is calculated, and then the difference in the angle values ​​of adjacent points is compared with the maximum difference in the angles of adjacent observable points on the same path, and they are divided into different observable paths, thus obtaining the observable path of each abnormal component, which improves the accuracy of detection.

[0035] 4. The optimal detection scheme of the system is calculated by using the A* algorithm and the shortest route allocation algorithm. Each robot is assigned to its corresponding observable path for collaborative detection according to the optimal detection scheme. The scheme satisfies the requirement of the shortest overall running distance of multiple robots, thereby further improving the overall operating efficiency of the system.

[0036] 5. By having multiple robots travel to their respective observable paths to conduct collaborative detection, multi-view and multi-modal data are collected. These multi-view and multi-modal abnormal image data are then fused and detected in the cloud using Kalman filtering, which improves the accuracy of the system in detecting equipment anomalies. Attached Figure Description

[0037] Figure 1 This is a flowchart of the overall method of the present invention;

[0038] Figure 2 This is a schematic diagram of the cloud-edge collaborative architecture of the present invention;

[0039] Figure 3 This is a schematic diagram of the multi-robot inspection of the present invention;

[0040] Figure 4 This is a flowchart of the inspection phase method of the present invention;

[0041] Figure 5 This is a global XYZ three-dimensional coordinate system diagram of the present invention;

[0042] Figure 6 This is a top-view diagram illustrating the robot inspection layout of the present invention.

[0043] Figure 7 This is a flowchart of the clustering algorithm of the present invention;

[0044] Figure 8 This is a flowchart of the observable path solving algorithm of the present invention;

[0045] Figure 9 This is a flowchart of the shortest route allocation algorithm of the present invention;

[0046] Figure 10 This is a flowchart of the collaborative detection stage of the present invention;

[0047] Figure 11 This is a mapping diagram between the abnormal position coordinates and the robot position coordinates of the present invention;

[0048] Figure 12 This is a schematic diagram of the observable path solving algorithm of the present invention;

[0049] Figure 13 This is a schematic diagram of the path endpoints and shortest route of the present invention;

[0050] Figure 14 This is the distance cost table from each robot to each path endpoint of the present invention;

[0051] Figure 15 This is a schematic diagram illustrating two types of distances for the robot to reach the remaining path according to the present invention;

[0052] Figure 16This is a cost table of distances from each robot to the remaining path breakpoints in the new detection route of this invention;

[0053] Figure 17 This is a schematic diagram of an embodiment of the present invention. Figure 1 ;

[0054] Figure 18 This is a schematic diagram of an embodiment of the present invention. Figure 2 ;

[0055] Figure 19 This is a schematic diagram of an embodiment of the present invention. Figure 3 ;

[0056] Figure 20 This is a schematic diagram of an embodiment of the present invention. Figure 4 . Detailed Implementation

[0057] The present invention will now be described in detail with reference to the accompanying drawings.

[0058] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0059] like Figures 1-3 As shown, a collaborative detection method for abnormal components of large equipment based on multiple robots is presented. The system is based on a cloud-edge collaborative architecture and utilizes a combination of clustering algorithms, observable path solving algorithms, A* algorithms, shortest path allocation algorithms, and Kalman filter fusion algorithms. This enables multiple robots to quickly detect abnormal components during inspections, ensuring accurate and reliable detection results. The technical solution is as follows:

[0060] First, by evenly distributing multiple robots around large equipment for inspection, and uploading the collected image data to the cloud in real time for image detection and analysis, the system's ability to detect anomalies is improved.

[0061] Secondly, using the two-dimensional XY horizontal plane where the large equipment base is located as a reference, a global XYZ three-dimensional coordinate system is established. The robot's position coordinates are marked on the two-dimensional XY horizontal plane to solve for and fit the observable path. The observed anomaly position coordinates are marked on the global XYZ three-dimensional coordinate system to solve for the locations of abnormal components and points. Furthermore, there is a one-to-one correspondence between the robot's position coordinates and the anomaly position coordinates.

[0062] Then, in order to more accurately locate the anomaly and determine the observable path, after the anomaly is first detected in the cloud, the robot continues to patrol around the large equipment α times to collect more anomaly location coordinates and robot position coordinates.

[0063] Subsequently, the cloud uses a clustering algorithm to group the collected and labeled abnormal location coordinates into multiple clusters, remove noise points, and classify the points in the clusters into core points and boundary points for calculating the cluster centroid. Then, reliable abnormal location coordinates in the clusters are further filtered. The filtered clusters represent abnormal components, and the filtered cluster centroids represent abnormal points, thus finally obtaining the abnormal components and abnormal points.

[0064] Meanwhile, the cloud randomly places different anomalies into a task list. During the collaborative detection phase, task points are retrieved from this list in that order for detection. Furthermore, if the number of anomaly coordinates within the cluster containing an anomaly is less than a set value (Pts), the detection process will proceed as follows: Clus At that time, do not add it to the task list.

[0065] After initially obtaining the abnormal components and abnormal locations, the cloud uses the one-to-one correspondence between the abnormal location coordinates and the robot's position coordinates to find the observable location points of each abnormal component. Then, it calculates the angle difference between each adjacent observable location point on the two-dimensional XY horizontal plane and compares it with the maximum angle difference Δθ between adjacent observable points on the same path, thus dividing and fitting different observable paths.

[0066] After obtaining the observable paths of each abnormal component, the cloud uses the A* algorithm to calculate the distance cost from the robot's current position to each path endpoint, and takes the minimum distance cost from each robot to each path to construct the cost matrix. The cost matrix is ​​then solved using the shortest route allocation algorithm to obtain the optimal detection scheme of the system.

[0067] Finally, each robot goes to the designated path to collect abnormal data of abnormal parts. The cloud uses Kalman filtering to fuse the multi-angle and multi-modal abnormal data collected by each robot to obtain a more accurate final detection result.

[0068] like Figure 4 As shown, the inspection phase includes four steps: patrol detection, abnormal data collection, observable point calculation, and observable path calculation.

[0069] Cruise detection: Using the two-dimensional XY horizontal plane where the large equipment base is located as a reference, establish a system as follows... Figure 5 The diagram shows a global XYZ three-dimensional coordinate system, where the cylinder in the diagram represents the circumscribed cylinder of a large device.

[0070] like Figure 6 As shown, the patrol inspection is performed by the robot following a specified inspection direction, with counterclockwise inspection being the positive direction and clockwise inspection being the negative direction, to inspect the equipment.

[0071] Abnormal Data Acquisition: When the cloud receives image data collected by a single sensor of robot i at time t, and detects an abnormality in a component through image detection and analysis, it will record the data uploaded by robot i at the timestamp t. vis I tir and I ac Three types of image data are fused using Kalman filtering to obtain a composite image, which is then used for image detection and analysis to determine if a component has malfunctioned. Once a component malfunction is confirmed, the cloud-based system calculates and marks the observed malfunction location coordinates P on the global XYZ three-dimensional coordinate system. it (x it ,y it ,z it ), calculate and mark the position coordinates R of the i-th robot in the two-dimensional XY plane coordinate system. it (x i ′ t ,y i ′ t ), and record the one-to-one correspondence.

[0072] When the cloud-based image detection and analysis function detects multiple anomalies in the image data of robot i at time t, it also marks the anomaly location coordinates on the global XYZ three-dimensional coordinate system and the position coordinates of robot i on the two-dimensional XY plane coordinate system, recording the one-to-one correspondence. For example, if robot i detects three anomalies at time t, they are marked as P... it 1 (x it ,y it ,z it ) and R it 1 (x i ′ t ,y i ′ t ), P it 2 (x it ,y it ,z it ) and R it 2 (x i ′ t ,y i ′ t ), P it 3 (x it ,y it ,z it ) and R it 3 (x i ′ t ,y i ′t ).

[0073] If the image data uploaded by robot i at time t is found to be normal in the cloud, then no position coordinates are calculated or marked.

[0074] After the cloud processes all the data uploaded by the robots through image detection, analysis, and location recording, the recorded abnormal location coordinates, robot position coordinates, and corresponding relationships are as follows: Figure 11 As shown.

[0075] Figure 11 This indicates that time 0 is the moment the system first detects an anomaly in the large equipment, and the i-th robot is the first to detect, resolve, and mark the anomaly; at time t1, robot 1 detects three anomalies; at time t2, robot 2 does not detect any anomalies. The position coordinates of robot i at time 0 are R. i0 (x i ′0,y i ′0), will be a point with the origin as its endpoint and passing through point R. i0 (x i ′0,y i The ray at '0' is set as the reference position. When the i-th robot reaches this ray α times, that is, after a certain robot first detects the anomaly, all robots continue to patrol around the large equipment α times, completing the anomaly position coordinate P. it (x it ,y it ,z it ) and robot position coordinates R it (x i ′ t ,y i ′ t Once the abnormal data is marked and uploaded, the abnormal data collection is complete.

[0076] After completing the alpha circle, they are now in the clouds. Figure 11 The initial set of robot position coordinates has been obtained. and the set of abnormal location coordinates Let f: R → P be a bijective mapping between elements in sets R and P, with an inverse mapping f. -1 :P→R.

[0077] Observable point solution: Observable point solution involves using a clustering algorithm in the cloud to select reliable outlier locations and assigning the coordinates P of the observed outlier locations. it Perform clustering and group P in the global XYZ 3D coordinate system. it Divided into core point P □ Three types of points are identified: boundary point P′ and noise point P″, and several clusters are obtained. and each cluster center of gravity The core point P □ P represents the cluster it Furthermore, the number of outlier coordinates within the spherical neighborhood ε of this point with radius eps is greater than Pts. Cor Given the radius eps and the minimum number of points Pts in the neighborhood of the core point, Cor All are initial settings; boundary point P′ represents P in the cluster excluding the core point. it Noise point P″ represents P that is not in the cluster. it Different clusters Used to indicate the different abnormal components detected in this round of inspection, and their respective centers of gravity. Considered as the corresponding cluster Identify the anomaly in the component and re-mark it as P. anom .in, Figure 7 This is a flowchart of the clustering algorithm.

[0078] At the same time, the cloud will transmit different anomaly points P anom Anomalies are randomly added to the task list in sequence. During subsequent collaborative detection, anomalies are retrieved from the task list from front to back for detection. If an anomaly P exists... anom Cluster The coordinates of the outlier point P are included. it The number is less than the set minimum number of points Pts required to satisfy collaborative detection. Clus Then this type of outlier P anom Do not add to the task list.

[0079] The cloud then determines the clusters The coordinates of the anomaly location P it With inverse mapping f -1 :P→R, derive the coordinates of P in the two-dimensional XY plane. it The corresponding robot position coordinates R it Then the coordinates of these robot positions are outlier points P. anom The observable locations are then remarked with the robot's position coordinates as observable location points Q. k (x k ,y k ).

[0080] like Figure 8 As shown, observable path solving: Each anomaly point P is solved in the cloud using an observable path solving algorithm. anom observable paths l1, l2, ..., l j ... It is agreed that the angle θ of a point in a two-dimensional plane is the angle between that point and the positive x-axis when connected to the origin. The observable path solving algorithm first calculates the observable positions Q on the two-dimensional XY plane coordinate system. kThe corresponding angle θ (0≤θ<2π) is determined, and each observable position point Q is assigned a corresponding angle θ (0≤θ<2π). k Sort the elements according to the size of their angles θ, with smaller angles first and larger angles last. If the (x+1)th Q in the sorting... k Angle θ χ+1 Subtract the x-th Q k Angle θ χ If the angle difference between adjacent observable points along the same path is less than or equal to the maximum angle difference Δθ, then the (x+1)th Q will be... k and the χth Q k Considered to be on the same observable path j In the middle, otherwise the (x+1)th Q k and the χth Q k In different observable paths.

[0081] The observable path is finally obtained through the observable path solving algorithm, such as Figure 12 As shown. Only the solved l1 and l2 are represented. j Two observable paths are shown; the remaining paths are omitted from the diagram. These are the observable positions at the x-1, x-th, and x+1th positions after being sorted according to the magnitude of angle θ, as shown in the figure. and The corresponding angle θ χ With θ χ-1 The difference is greater than Δθ. and The corresponding angle θ χ+1 With θ χ The difference is less than Δθ; the shaded area represents the occluded region, that is, the corresponding path is an unobservable path.

[0082] like Figure 10 As shown, the collaborative detection phase includes two steps: calculating the optimal detection scheme and executing the optimal detection scheme.

[0083] The optimal detection scheme is calculated in the cloud. First, the anomaly point P needs to be selected from the task list. anom We extract objects from the front and test them from the back. Let P be the object being tested. a ′ nom Let the object to be detected be P. a ′ nom There are a total of m observable paths. Simultaneously, there are also observable paths l. j The observable points with the smallest and largest angles are defined as endpoints, and are denoted as and respectively. and

[0084] The A* algorithm is a heuristic search algorithm used to find the shortest path from a start point to an end point in a graph. It selects the optimal path by combining the actual cost of the path with a heuristic estimate. In cooperative detection tasks, the constraints of the A* algorithm include: the path cannot enter areas containing large equipment; and the standard heuristic function, Manhattan distance, is used. With these constraints, the A* algorithm can efficiently calculate the shortest path distance from the robot's position to the path endpoint.

[0085] The A* algorithm is used to find the path from the current position of robot i to path l. j endpoints and The shortest route distances are denoted as follows: and like Figure 13 As shown.

[0086] The distance cost from the robot's current position to each path endpoint is calculated using the A* algorithm, such as... Figure 14 As shown.

[0087] This allows us to obtain the distances from each robot to the observable points with the minimum and maximum path angles. We then construct cost matrix C1 using the distances from each robot to the observable points with the minimum path angles, and cost matrix C2 using the distances from each robot to the observable points with the maximum path angles.

[0088]

[0089]

[0090] The final cost matrix C = g(C1, C2), where g represents C[p][q] = min(C1[p][q], C2[p][q]), that is, the element in the p-th row and q-th column of the final cost matrix C is equal to the minimum value of the elements in the p-th row and q-th column of C1 and C2.

[0091] After obtaining the cost matrix, the optimal detection scheme for collaborative detection is solved using the shortest path allocation algorithm, so that the total path of all robots is minimized.

[0092] like Figure 9 As shown, the shortest route allocation algorithm generally consists of the following five steps:

[0093] The first step is to subtract the minimum value within each row of the cost matrix, and then subtract the minimum value within each column, thus ensuring that each row and each column has at least one zero.

[0094] Step 2: Cover all the zeros in the matrix with as few row markers and column markers as possible. A row marker will cover all the elements in a row of the matrix, and a column marker will cover all the elements in a column of the matrix. If the sum of the number of row and column markers is less than min(n, m), then proceed to Step 3; if the sum of the number of row and column markers is equal to min(n, m), then find the current optimal solution (the current optimal solution means: taking min(n, m) zeros from the matrix, and these zeros each occupy different rows and columns of the matrix. The solution is that the robots represented by the rows where these zeros are located go to the path endpoints represented by the columns where they are located, and then go to the other endpoint of the same path), and proceed to Step 4;

[0095] Step 3: Find the minimum value ξ among the elements not covered by the row and column markers. Subtract ξ from all the elements not covered by the row and column markers, add ξ to the elements covered twice by the row and column markers, remove the row and column markers, and then re - execute Step 2;

[0096] Step 4: After obtaining the current optimal solution, since the relationship between the number of robots n and the number of observable paths m is different when detecting different abnormal points, and the system operation plans are also different, the following discussions are carried out:

[0097] ① When n > m, then take the current optimal solution as the optimal detection plan for collaborative detection, and dispatch m robots to m paths respectively to perform collaborative detection on the current detection target P a ′ nom , and the remaining n - m continue to patrol and collect abnormal data of large - scale equipment;

[0098] ② When n = m, then take the current optimal solution as the optimal detection plan for collaborative detection, and dispatch n robots to n paths respectively to perform collaborative detection on the current detection target P a ′ nom ;

[0099] ③ When n < m, then take the current optimal solution as part of the solution with the shortest total path for the current collaborative detection, and dispatch n robots to n paths respectively to perform collaborative detection on the current detection target P a ′ nom , and record the current detection route of the robot as the route to the specified path plus the route for detection on the specified path. Assume that the i - th robot is dispatched to the observable path l j , and first passes through The detection end - point is Calculate the total length s of the current detection route of the i - th robot, that is, from the current position to the detection end - point[[ID=3,6]] i and the detection end - point of the current route to reach the remaining path endpoint and The shortest route distances are denoted as follows: and The distance, such as Figure 15 As shown, proceed to step five.

[0100] The fifth step is to add the total length of the current detection route and the distance from the current detection route endpoint to the remaining path endpoint as the distance cost of the new multi-robot detection route. Figure 16 As shown, similar to the method for constructing the cost matrix described above, a new cost matrix is ​​reconstructed, and the first step is executed again to solve for the optimal detection scheme for collaborative detection.

[0101] The current optimal solution in the second step is the adjusted cost matrix, which satisfies that the number of row and column labels is equal to min(n,m), specifically including the following:

[0102] If n≤m, then each row of the cost matrix has at least one zero element, and at least one zero element is selected from each row to ensure that the selected zero elements are in different columns. The row and column where the selected zero element is located represent the observable path endpoint that the robot should go to in the corresponding column.

[0103] If n>m, then each column of the cost matrix has at least one zero element, and at least one zero element is selected from each column to ensure that the selected zero elements are all in different rows. The row and column where the selected zero element is located represent the observable path endpoint that the robot should go to in the corresponding column.

[0104] The optimal detection scheme for system collaborative detection is obtained through the shortest path allocation algorithm described above.

[0105] In executing the optimal detection plan, each robot is assigned to travel along the optimal detection route. After reaching the observable path, it collects abnormal data on abnormal parts and uploads the collected data to the cloud in real time. The cloud uses Kalman filtering to fuse multimodal abnormal data from different angles to obtain more accurate detection results.

[0106] The Kalman filter algorithm is a recursive estimation algorithm used in dynamic systems to provide an optimal estimate of the system state by combining a system model and measurement data. It works through two main steps: a prediction step predicts the state at the next moment based on the system model, and an update step corrects the prediction based on actual measurements, thus achieving an accurate estimate of the system state even in the presence of noise.

[0107] Complete the current detection target P a ′ nom After completing the detection task, check if there are any other tasks P to be detected in the task list. anom If an anomaly exists, the new anomaly that appears first is selected as the target for the next round of detection.

[0108] The collaborative detection phase ends once all anomalies in the list have been detected.

[0109] In this embodiment, the number of robots is set to n=4, the number of times all robots continue to inspect the large equipment after the first anomaly is detected is α=3, the given radius is eps=15cm, and the minimum number of points in the neighborhood of the core point is Pts. Cor =10. Minimum number of points required for collaborative detection (Pts) Clus =5000, the ratio of core point weight to boundary point weight is τ=2, the cluster screening ratio is v=95%, and the maximum difference in angle between adjacent observable points on the same path is Δθ=1°.

[0110] Reference Figure 17 The application scenario of this invention is the large equipment shown in Figure 2. In the figure, numbers 2-5 are four inspection robots placed around the large equipment. They are evenly distributed around the large equipment and patrol around it in the direction indicated by arrows 6-9.

[0111] During multi-robot inspection, the collected I data is transmitted in real time. vis I tir and I ac Three types of image data are uploaded to the cloud. After the cloud detects that a certain robot has an anomaly at a certain moment, all robots continue to patrol around the large equipment three times. During the patrol, at the locations where abnormal parts can be observed, the robot's position coordinates, the abnormal position coordinates, and the correspondence between the two coordinates are marked.

[0112] After three laps are completed, the cloud uses a clustering algorithm based on a given radius eps=15 and the minimum number of points Pts in the neighborhood of the core point. Cor =10 Classify the abnormal location coordinates into core points, boundary points, and noise points. At the same time, based on the ratio of the weight of the core point to the weight of the boundary point τ=2 and the clustering screening ratio v=95%, reliable abnormal point locations are screened to obtain the locations of abnormal components and abnormal points. Based on this, more reliable observable location points are obtained by inversely deducing the correspondence between the robot's position coordinates and the abnormal position coordinates.

[0113] Reference Figure 18 Using the aforementioned observable locations, the angles of each observable location on the two-dimensional XY horizontal plane are calculated in the cloud (as agreed above, the angle θ of a point in the two-dimensional plane is the angle between that point and the origin and the positive x-axis), as shown in the observable locations figure. The angle is θ χ =90°, simultaneously observable location points Adjacent observable locations The difference in angle values ​​is greater than the maximum difference in angle between adjacent observable points on the same path, Δθ = 1°. Therefore, they are divided into different paths, resulting in five observable paths: l1, l2, l3, l4, and l5.

[0114] Reference Figure 19 Using the A* algorithm and the shortest path assignment algorithm, in the cloud, the first three steps of the shortest path assignment algorithm are used to solve the current optimal solution for system collaborative detection and assign robots to the corresponding paths. The current optimal solution is: robot 1 goes to the observable path l2, robot 2 goes to l3, robot 3 goes to l5, and robot 4 goes to l1.

[0115] Reference Figure 20 In the fourth step of the shortest path assignment algorithm, since the number of observable paths is greater than the number of robots, the algorithm also needs to calculate the distance cost of the remaining paths. Then, in the fifth step, a new cost matrix is ​​calculated, and the algorithm returns to the first step for iterative calculation to solve for a new round of detection schemes. This process continues until the optimal detection scheme is obtained and the robot assignment information is updated. In this example, the optimal detection scheme is as follows: Figure 20 As shown: Robot 1 goes to the observable path l2, Robot 2 goes to l3 and then to l4, Robot 3 goes to l5, and Robot 4 goes to l1;

[0116] Each robot proceeds to its corresponding path to collect information on abnormal components and uploads it to the cloud in real time. The cloud then uses Kalman filtering to fuse the multimodal and multi-view abnormal data to obtain the final detection result.

[0117] This invention provides a fast and accurate anomaly detection method for collaborative detection of abnormal components in large equipment by comprehensively applying clustering algorithms, observable path solving algorithms, A* algorithms, shortest path allocation algorithms, and Kalman filter fusion algorithms. Simultaneously, it utilizes the correspondence between robot positions and simultaneously observed fault point positions to solve for optimal observable locations, and then transforms these observable locations into observable paths using the observable path solving algorithm. Furthermore, it combines A* algorithms and shortest path allocation algorithms to address the problem of how to allocate observable paths when the number of robots and the number of observable paths have different relative sizes.

[0118] By clustering the coordinates of abnormal locations, the collected coordinates of abnormal locations are filtered to obtain more reliable coordinates of abnormal locations, and from this, more reliable observable locations are deduced.

[0119] The observable path solving algorithm calculates the angle of the observable location point on the two-dimensional XY horizontal plane. Then, the difference in angle values ​​between adjacent points is compared with the maximum difference in angles Δθ between adjacent observable points on the same path, and the points are assigned to different observable paths, thus obtaining the observable paths of each abnormal component.

[0120] The optimal detection scheme of the system is calculated by using the A* algorithm and the shortest route allocation algorithm. The scheme determines that each robot should go to its corresponding observable path for collaborative detection, and the scheme satisfies the requirement of the shortest overall running distance of multiple robots, thereby further improving the overall operating efficiency of the system.

[0121] Finally, multiple robots were used to conduct collaborative detection along their respective observable paths, collecting multi-view and multi-modal data. These multi-view and multi-modal abnormal image data were then fused and detected in the cloud using Kalman filtering, which improved the accuracy of the system in detecting equipment anomalies.

[0122] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A collaborative detection method for abnormal components of large equipment based on multiple robots, characterized in that, The method sets the number of robots based on the size and complexity of the large equipment. , and the robot from 1 to Number them sequentially and distribute the robots evenly around the large equipment for inspection; Each robot is equipped with three types of sensors: a visible light camera, a thermal infrared camera, and an audio-visual system. During the inspection process, the robot scans the equipment to acquire visible light images. Thermal infrared images Harmony and acoustic images ; The system is based on a cloud-edge collaborative architecture, with the robot acting as the edge device to collect data. , and After receiving the three types of image data, the image data is uploaded to the cloud with corresponding image detection and analysis functions in real time for processing. The cloud can perform anomaly detection based on image processing on the received image data to determine whether the large equipment has an anomaly. The multi-robot abnormal component collaborative detection includes two stages: the first stage is the inspection stage, and the second stage is the collaborative detection stage. The collaborative detection phase includes: Calculating the optimal detection scheme is done in the cloud. First, anomalies need to be identified from the task list. The items are taken out from front to back for testing, and the currently taken-out test object is recorded as follows: Let the object of detection be... There are a total of observable paths 1000 lines, 2000 lines, 3000 lines, 4000 lines, 50 The observable points with the smallest and largest angles are set as endpoints, and the observable point with the smallest angle is denoted as... The observable point with the largest angle is denoted as ; Find the first... using the A* algorithm Robot's current position to path endpoints and The shortest route distances are denoted as follows: and The distance cost from the robot's current position to each path endpoint is calculated using the A* algorithm. This allows us to obtain the distances from each robot to the observable points with the minimum and maximum angles on each path. A cost matrix is ​​then constructed using the distances from each robot to the observable points with the minimum angles on each path. A cost matrix is ​​constructed based on the distance from each robot to the observable point with the largest path angle. ; Final cost matrix ,in express That is, the final cost matrix The Line number Column element equals and No. Line number The minimum value of a column element; After obtaining the cost matrix, the optimal detection scheme for collaborative detection is solved by the shortest path allocation algorithm, so that the total running path of all robots is minimized; The optimal detection plan is executed, and each robot is assigned to travel along the route of the optimal detection plan. After reaching the observable path, abnormal data is collected on the abnormal parts. At the same time, the collected data is uploaded to the cloud in real time. The cloud uses Kalman filtering to fuse the abnormal data of multimodal and multi-viewpoints to obtain the final detection result of the abnormal point. Complete the current detection objective After completing the detection task, check if there are any remaining tasks to be detected in the task list. If an anomaly exists, the new anomaly that appears first is selected as the target for the next round of detection. Anomalies in the list Once all tests are completed, the collaborative testing phase ends.

2. The collaborative detection method for abnormal components of large equipment based on multiple robots according to claim 1, characterized in that, The inspection phase includes: Cruise inspection involves a robot circling the equipment according to a designated inspection direction, and inspecting the equipment within a two-dimensional space where the large equipment base is located. Establish a reference plane. Three-dimensional coordinate system; Among them, the counterclockwise inspection direction is set as the positive direction, and the clockwise inspection direction is set as the negative direction; Abnormal data collection, when the cloud receives the first Robot No. 1 After collecting image data from a single sensor, if the image detection and analysis function detects any component malfunction, then... The time stamp below this moment Uploaded by robot # , and The three types of image data are fused using Kalman filtering to obtain a composite image, which is then used for image detection and analysis to determine if a component is malfunctioning. Once a component is determined to be malfunctioning, the cloud performs a global... Solve and mark the coordinates of the observed anomaly locations in a three-dimensional coordinate system. In two dimensions Solve and label the first in a planar coordinate system The location coordinates of robot number And record the one-to-one correspondence, denoted as the mapping relationship. ; and when the first Robot No. 1 If the uploaded image data is found to be normal in the cloud, no location coordinates will be calculated or marked. Observable point calculation involves using clustering algorithms in the cloud to filter reliable outlier locations and then recording the coordinates of the observed outliers. Perform clustering, and globally General in three-dimensional coordinate system Divided into core points Boundary points and noise points Three types of points, and several clusters were obtained. and each cluster center of gravity The core point Indicates in the cluster And this point is Sphere neighborhood with radius The number of abnormal location coordinates within is greater than Given radius Minimum number of points in the neighborhood of the core point All are initial settings; boundary points This refers to the cluster excluding the core. Noise point Indicates not in the cluster Different clusters Used to indicate the different abnormal components detected in this round of inspection, and their respective centers of gravity. Considered as the corresponding cluster Identify the abnormal points in the component and re-mark them. ; At the same time, the cloud will detect different anomalies. The anomalies are randomly placed into the task list. In the subsequent collaborative detection phase, anomalies are retrieved from the task list from front to back for detection. If anomalies are found... Cluster Includes the coordinates of the outlier locations The number is less than the minimum number of points required to satisfy collaborative detection. Then this type of anomaly Do not add to task list; The cloud then determines the clusters Anomaly location coordinates With inverse mapping Derivation of two dimensions In a planar coordinate system Corresponding robot position coordinates Then the coordinates of these robot positions are outliers. The observable locations are then remarked with the robot's position coordinates as observable location points. ; Observable path solving involves using an observable path solving algorithm in the cloud to identify each outlier. observable path Angle set at a point in a two-dimensional plane Connect this point to the origin and then... The angle formed by the positive axis; the observable path solution algorithm first in two dimensions Calculate each observable location point in a planar coordinate system Corresponding angle , and each observable location point According to angle Sort the angles by size, with smaller angles first and larger angles last. If the sorting is in the order of... indivual Angle Subtract the first indivual Angle Less than or equal to the maximum angle difference between adjacent observable points on the same path. Then the first indivual and the indivual Considered to be on the same observable path In the middle, otherwise the first indivual and the indivual In different observable paths, different observable paths are divided and fitted to obtain observable paths.

3. The collaborative detection method for abnormal components of large equipment based on multiple robots according to claim 2, characterized in that, In the abnormal data collection step, the first... The moment when robot No. 1 first detected the anomaly is recorded as time 0, and the coordinates of time 0 are: The origin will be the endpoint and pass through the point The ray is set as the reference position, when the ray is... The robot then reached the ray. After that, once a certain robot first detected the anomaly, all robots continued to patrol around the large equipment. Circle, complete the coordinates of the abnormal location. Robot position coordinates Once the abnormal data is marked and uploaded, the abnormal data collection is complete; Finish After circling around, a preliminary set of the robot's position coordinates was obtained in the cloud. and the set of abnormal location coordinates and sets and set mapping relationship .

4. The collaborative detection method for abnormal components of large equipment based on multiple robots according to claim 1, characterized in that, The shortest route allocation algorithm includes the following steps: S1: Subtract the minimum value in each row of the cost matrix, and then subtract the minimum value in each column, thus ensuring that each row and each column has at least one zero. S2: Cover all zeros in the matrix using as few row and column tags as possible. One row tag covers an entire row of elements, and one column tag covers an entire column of elements. If the sum of the number of row and column tags is less than 1, then the matrix is ​​considered to have zero zeros. Then proceed to step S3; if the number of row and column markers is equal to If the current optimal solution is found, proceed to step S4. S3: Find the minimum value among the elements not covered by row and column markers. Subtract all elements not covered by row and column markers Elements that are covered twice by row and column markers plus Remove the row and column markers and repeat step S2; S4: After obtaining the current optimal solution, different outliers are detected. Number of robots With the number of observable paths Different sizes result in different system operation schemes. and Depending on the size, different schemes can be selected for operation; S5: Add the total length of the current detection route of each robot and the distance from the end of the current detection route to the endpoint of the remaining path as the distance cost for multiple robots to reach the remaining observable path. At the same time, in the same way as the above method for constructing the cost matrix, reconstruct a new cost matrix, and repeat step S1 to solve the optimal detection scheme for collaborative detection.

5. A collaborative detection method for abnormal components of large equipment based on multiple robots according to claim 4, characterized in that, Step S4 involves comparison. and The size depends on the chosen scheme, specifically including the following: when When this happens, the current optimal solution is taken as the optimal detection scheme for collaborative detection, and the appropriate detection methods are dispatched. The robots went to The path to the current detection target Perform collaborative detection, the remaining... One robot continues to inspect and collect abnormal data from large equipment; when When this happens, the current optimal solution is taken as the optimal detection scheme for collaborative detection, and the appropriate detection methods are dispatched. The robots went to The path to the current detection target Conduct collaborative detection; when When this happens, the current optimal solution is taken as part of the shortest path solution for the current collaborative detection, and dispatched. The robots went to The path to the current detection target Perform collaborative detection, and denote the robot's current detection route as the sum of the route to the specified path and the route detected along the specified path. Assume the first... Robot No. 1 was dispatched to the observable path. And first pass through the observable point with the smallest angle. The detection endpoint is the observable point with the largest angle. Calculate the first digit using the A* algorithm. The current detection route for robot number 1 is from its current position to the detection endpoint. Total length and current route detection endpoint Remaining path endpoints and The shortest route distances are denoted as follows: and Measure the distance and proceed to step S5.

6. The collaborative detection method for abnormal components of large equipment based on multiple robots according to claim 1, characterized in that, The initial cost matrix of the distances from each robot to the observable points with the minimum and maximum path angles. and They are respectively: in, For the first Robot's current position to path endpoints The shortest route distance, For the first Robot's current position to path endpoints The shortest route distance.

7. A collaborative detection method for abnormal components of large equipment based on multiple robots according to claim 3, characterized in that, When the abnormal data is collected, when the robot is Multiple abnormal location coordinates were detected at any time. At that time, mark the robot's position and the coordinates of abnormal positions respectively. ; If the first Robot No. 1 When three abnormal location coordinates are detected at any given time, the abnormal location coordinates are marked as follows: , and and the robot's location , and And record the one-to-one correspondence.

8. A collaborative detection method for abnormal components of large equipment based on multiple robots according to claim 2, characterized in that, In the calculation of the observable points, those with a distance less than [a certain value] will be considered. core point and boundary points Clustered in the same initial cluster In the middle, according to the core point and boundary points initial cluster weight calculation center of gravity The core point Weights set to boundary points of Times; Obtain the initial cluster centroid Then, select clusters Center and weight The closer the Euclidean distance is to the front core point and boundary points The core points after screening and boundary points Clusters that constitute abnormal components and calculate clusters center of gravity .

9. A collaborative detection method for abnormal components of large equipment based on multiple robots according to claim 4, characterized in that, The current optimal solution in step S2 is the adjusted cost matrix, which satisfies that the number of row and column labels is equal to the number of rows and columns. Specifically, it includes the following: like If at this time, each row of the cost matrix has at least one zero element, and at least one zero element is selected from each row to ensure that the selected zero elements are in different columns. The row and column where the selected zero element is located represent the observable path endpoint that the robot should go to in the corresponding column. like If at this point, each column of the cost matrix has at least one zero element, and at least one zero element is selected from each column to ensure that the selected zero elements are all in different rows, then the row and column where the selected zero element is located represent the observable path endpoint that the robot should go to in the corresponding column.

Images