Robot operational situation awareness and early warning system based on multi-source data fusion

Abstract

This invention discloses a robot operation situation awareness and early warning system based on multi-source data fusion, belonging to the field of robot operation anomaly early warning technology. It includes a historical operation data acquisition module, a trajectory space acquisition module, a step cycle acquisition module, a reference area acquisition module, a situation dependency coefficient acquisition module, a dependency node determination module, and an anomaly node marking module. The system quantifies the motion and spatial changes of each joint in different cycles using position migration coefficients and spatial migration coefficients. It calculates the situation dependency coefficient by analyzing the migration coefficients between the operation node and its sub-nodes, which intuitively reflects the correlation strength between each sub-node and the operation node. When an operation node is anomaly marked as a primary anomaly node, the system simultaneously checks its dependent nodes, marking dependent nodes that deviate from the reference area as collaborative anomaly nodes. This not only enables timely detection of anomalies but also facilitates tracing the root cause of the anomaly.

Description

Technical Field

[0001] This invention belongs to the field of robot operation anomaly early warning technology, specifically a robot operation situation perception and early warning system based on multi-source data fusion. Background Technology

[0002] With the rapid development of industrial automation and intelligent manufacturing, robotics has become the core of modern production systems. In particular, robotic arms are widely used in high-precision and highly repetitive manufacturing tasks due to their high efficiency and stability.

[0003] However, during long-term operation, the performance of each joint of the robot's manipulator will gradually deteriorate due to wear, fatigue and other factors, causing movement deviations or malfunctions in each joint of the manipulator. This will affect the production quality and efficiency of the robot in the process of manufacturing the same product, and may even lead to sudden failures, resulting in production stoppages and economic losses.

[0004] Meanwhile, during the manufacturing process of the same product by the robotic arm, the end effector, as the core component that directly performs the manufacturing operation, typically relies on simple threshold alarms for the end effector during robot anomaly monitoring. However, since the end effector's manufacturing actions depend on the coordinated operation of other related joints, simply issuing threshold alarms for the end effector cannot accurately pinpoint which coordinated joint's movement deviation caused the end effector's anomaly. This makes it difficult to quickly locate the root joint causing the anomaly, bringing great difficulties to fault diagnosis and maintenance, and failing to effectively ensure the safety and stability of robot operation. Based on this, a robot operation situation awareness and early warning system based on multi-source data fusion is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a robot operation situation awareness and early warning system based on multi-source data fusion, which solves the technical problem that it is impossible to accurately locate which coordinating joint's movement deviation caused the operator's abnormality during the process of monitoring robot anomalies.

[0006] A robot operation situation awareness and early warning system based on multi-source data fusion includes:

[0007] The historical operation data acquisition module calibrates the joint nodes corresponding to each joint of the robot arm and acquires the historical motion trajectory and production cycle of each joint node when the robot arm performs multiple identical product manufacturing tasks.

[0008] The trajectory space acquisition module draws the trajectory space based on the overall length of the manipulator;

[0009] The step cycle acquisition module obtains the baseline cycle based on each production cycle and evenly divides the baseline cycle into multiple step cycles.

[0010] The reference region acquisition module combines and analyzes each historical motion trajectory with the trajectory space to obtain the reference regions corresponding to each joint node in different step cycles.

[0011] The situation dependence coefficient acquisition module takes the rightmost joint node of the robot arm as the operation node and marks the remaining nodes as sub-nodes. It obtains the center point coordinates of each subspace and analyzes the center point coordinates and reference area volume of the subspace contained in the reference area of ​​each joint node in different step cycles to obtain the situation dependence coefficient between each sub-node and the operation node.

[0012] The dependency node determination module determines the dependency nodes corresponding to the operation nodes based on the situation dependency coefficient.

[0013] The abnormal node marking module compares and analyzes the real-time trajectory points of the robot arm's operating nodes and dependent nodes in the trajectory space with the comparison areas corresponding to the operating nodes and their dependent nodes, and marks the main abnormal nodes and cooperative abnormal nodes.

[0014] As a further aspect of the present invention, the specific method for drawing the trajectory space is as follows:

[0015] Mark the overall length of the manipulator as arm length L. Draw a sphere Q with arm length L as the radius. Divide the internal space of the sphere into multiple subspaces Ei. Use Ei as the spatial label of each subspace, where i represents a different subspace, i=1, 2, ..., a1, where a1 represents the total number of subspaces, a1 is a positive integer, and a1≥6.

[0016] As a further aspect of the present invention, the specific method for uniformly dividing the reference period into multiple stepped periods is as follows:

[0017] The average of the maximum and minimum values ​​in the production cycle corresponding to the robot manipulator when performing multiple identical product manufacturing tasks is used as the reference cycle for the robot manipulator when performing the corresponding product manufacturing tasks. The reference cycle is evenly divided at the same time interval to obtain multiple stepped cycles Jj, where j represents different stepped cycles.

[0018] As a further aspect of the present invention, the specific method for obtaining the reference regions corresponding to each joint node in different step cycles is as follows:

[0019] S1: Select any one of the different step cycles as the analysis period;

[0020] S2: Select any one of the joint nodes of the robot arm as the analysis node.

[0021] S3: Align the center point of the left end face of the first joint of the robot arm with the center point of the trajectory space as the alignment point. At the same time, place the historical motion trajectories of the analysis node in the analysis cycle when performing multiple identical product manufacturing tasks in the trajectory space. Obtain the spatial labels of the subspaces involved by each historical motion trajectory of the analysis node in the trajectory space. Extract the spatial labels of the subspaces involved by each historical motion trajectory from the trajectory space and associate them with the corresponding analysis node as the activity area of ​​the analysis node in the analysis cycle when performing multiple manufacturing products. Combine the subspaces involved by each activity area as the reference area of ​​the analysis node in the analysis cycle. At the same time, associate the spatial labels of the common subspaces with the analysis node as the reference area of ​​the analysis node in the analysis cycle.

[0022] S4: Repeat steps S2-S3, using the same analysis method to analyze the remaining joint nodes, and obtain the reference area corresponding to each joint node within the analysis cycle;

[0023] S5: Repeat steps S2-S4, using the same analysis method to analyze the remaining step cycles, and obtain the reference area corresponding to each node in different step cycles.

[0024] As a further aspect of the present invention, the specific method for obtaining the situational dependency coefficients between each sub-node and the operation node is as follows:

[0025] S01: Designate the rightmost joint node of the robot arm as the operation node and mark the remaining nodes as sub-nodes.

[0026] S02: Select any step cycle from different step cycles as the target cycle;

[0027] Obtain the coordinates of the center points of each subspace within the reference region corresponding to the operation node within the target period. Use the average of the x and y coordinates of each center point as the punctuation coordinates of the operation node within the reference region within the target period. Calculate the volume V of the trajectory space. Obtain the sub-volume F of a single subspace using F=V / a1. Obtain the number n of subspaces within the reference region corresponding to the operation node within the target period. Use the product of the number n and the sub-volume F as the volume of the reference region corresponding to the operation node within the target period.

[0028] S03: Using the same analysis method as in step S02, analyze the operation node in the remaining step cycle to obtain the punctuation coordinates of the operation node in each step cycle; at the same time, using the same analysis method as in step S02, analyze the subspaces contained in the reference regions corresponding to the operation node in the remaining step cycle to obtain the reference region volume corresponding to the reference region of the operation node in each step cycle;

[0029] S04: Take each two adjacent step cycles as a calibration cycle, and then obtain multiple calibration cycles. Analyze the vertical and horizontal coordinates of the calibration points in each two adjacent step cycles to obtain the position migration coefficient of the operation node in each calibration cycle. At the same time, take the absolute value of the difference between the reference area volumes corresponding to the reference areas of the operation node in each two adjacent step cycles as the spatial migration coefficient of the operation node in each calibration cycle.

[0030] S05: Select a joint node from each sub-node as the target node, and use the same method as steps S02-S04 to analyze the center point coordinates of each subspace contained in the reference area of ​​the target node within the target period, thereby obtaining the position migration coefficient and spatial migration coefficient of the target node within each calibration period.

[0031] The position migration coefficient and spatial migration coefficient of the operating node in each calibration period are compared and analyzed with those of the target node in each calibration period. Based on the analysis results, the situational dependence coefficient between the target node and the operating node is obtained.

[0032] S06: Use the same analysis method as in step S05 to analyze the remaining sub-nodes, and then obtain the situational dependence coefficients between each sub-node and the operation node.

[0033] As a further aspect of the present invention, the specific method for obtaining the position migration coefficients of the operating node within each calibration period is as follows:

[0034] The ratio between the absolute difference of the ordinate and the absolute difference of the abscissa in the coordinates of each two adjacent step cycles is used as the position migration coefficient of the operating node between each two adjacent step cycles, thereby obtaining the position migration coefficient of the operating node in each calibration cycle.

[0035] As a further aspect of the present invention, the specific method for obtaining the situational dependency coefficient between the target node and the operating node is as follows:

[0036] From the position migration coefficients and spatial migration coefficients of the operating node and target node in each calibration period, obtain the position migration coefficients and spatial migration coefficients of the operating node and target node within the same calibration period. When both the position migration coefficients and spatial migration coefficients of the operating node and target node within the same calibration period are not 0, the corresponding calibration period is marked as a two-level transformation period. When both the position migration coefficients and spatial migration coefficients of the operating node and target node within the corresponding calibration period are not 0 and the spatial migration coefficient is 0, the corresponding calibration period is marked as a single-level transformation period. When all values ​​are non-zero and the position migration coefficient is zero, the corresponding calibration period is marked as a single-layer transformation period. Otherwise, no processing is performed. The position migration coefficient and spatial migration coefficient of the operating node and the target node in each calibration period are compared and analyzed one by one. At the same time, the number of double-layer transformation periods and single-layer transformation periods corresponding to each calibration period are obtained. The sum of the products of the number of double-layer transformation periods and single-layer transformation periods with β1 and β2 respectively is used as the situational dependence coefficient between the target node and the operating node. β1 and β2 are preset coefficients that satisfy 1=β1+β2 and β1≥β2.

[0037] As a further aspect of the present invention, the specific method for determining the dependent nodes corresponding to the operation nodes is as follows:

[0038] Obtain the mean value of the situation dependency coefficients corresponding to each sub-node and the operation node, as well as the absolute value of the difference between the maximum and minimum values ​​of the situation dependency coefficients. Use the difference between the mean value and the absolute value of the difference as the boundary value. Sub-nodes with situation dependency coefficients greater than the boundary value are determined as dependent nodes corresponding to the operation nodes. Otherwise, no processing is performed.

[0039] As a further aspect of the present invention, the specific method for determining and marking the primary abnormal node and the cooperating abnormal node is as follows:

[0040] The system acquires real-time trajectory points for the robot's manipulator nodes and their dependent nodes. Based on the acquisition time of the real-time trajectory points, it obtains the real-time ladder cycle corresponding to the acquisition time from different ladder cycles. From the reference areas corresponding to each node in different ladder cycles, it obtains the reference areas corresponding to the manipulator nodes and their dependent nodes in the real-time ladder cycle and uses these as the comparison areas for the manipulator nodes and their dependent nodes in the real-time ladder cycle. When the real-time trajectory point of the manipulator node is within its corresponding comparison area in the real-time ladder cycle, no processing is performed. When the real-time trajectory point of the manipulator node is not within its corresponding comparison area in the real-time ladder cycle, the manipulator node is marked as a primary abnormal node. At the same time, the real-time trajectory points of each dependent node are compared with their corresponding comparison areas in the real-time ladder cycle one by one. Dependent nodes whose real-time trajectory points are not within their corresponding comparison areas are marked as cooperative abnormal nodes. Otherwise, no processing is performed. Both primary abnormal nodes and cooperative abnormal nodes are output simultaneously.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] (1) In this invention, by combining historical motion trajectories with trajectory space, cross-analysis of multiple historical motion trajectories is performed to extract the reference area of ​​each joint in different step cycles, and the safe range of motion of the robot joint under normal operation is represented by the reference area;

[0043] (2) In this invention, the motion and spatial changes of each joint in different cycles are quantified by position migration coefficient and spatial migration coefficient. The situation dependence coefficient is calculated by analyzing the migration coefficient between the operation node and the sub-node. The situation dependence coefficient directly reflects the correlation strength between each sub-node and the operation node.

[0044] (3) In this invention, when an operation node is marked as a primary abnormal node, the system will simultaneously check its dependent nodes and mark the dependent nodes that deviate from the reference area as collaborative abnormal nodes. This not only enables timely detection of abnormalities but also facilitates tracing the root cause of the abnormality. It solves the problem of not being able to accurately determine the abnormal associated nodes in the prior art, making it easier for staff to quickly locate faults and take targeted maintenance measures. This achieves comprehensive and accurate perception and effective early warning of the robot's operating status. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the system framework structure of the present invention;

[0046] Figure 2 This is a schematic diagram of the framework structure of the situation dependence coefficient acquisition module of the present invention;

[0047] Figure 3 This is a schematic diagram of the method of marking the corresponding calibration period as a two-layer transformation period or a single-layer transformation period according to the present invention. Detailed Implementation

[0048] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Example 1: Please refer to Figures 1-3 This application provides a robot operation situation awareness and early warning system based on multi-source data fusion, including:

[0050] The historical operation data acquisition module obtains the center point of the right end face of each joint of the robot arm. The center points of each joint are labeled as joint nodes Dd from left to right according to the joint sequence. Dd is used as the node number of each joint node, where d represents a different joint node (d = 1, 2, ..., e), and e represents the total number of joint nodes (e is a positive integer, e ≥ 2). The module acquires historical operation data for each joint node of the robot arm when performing multiple identical product manufacturing tasks. This historical operation data includes the historical motion trajectory of each joint node and the production cycle corresponding to each product manufacturing task performed by the robot arm.

[0051] It should be noted that when a robot performs multiple identical manufacturing tasks, the motion path of each joint can be monitored and recorded in real time by sensors installed on the robot. This allows us to obtain the coordinate sequence of each joint node at different points in time. These data points are then linked together to form the historical motion trajectory of the joint. Simultaneously, by recording the time at the beginning of each product manufacturing task and stopping the timer at the end of the task, the production cycle of a single product can be obtained. By recording the time for the production tasks of multiple products in this way, a series of production cycle data can be obtained. All of the above are existing and mature technologies, so they will not be elaborated on here.

[0052] By integrating existing mature sensor and time recording technologies, the historical motion trajectory and production cycle of each joint of the robot can be obtained efficiently and accurately, providing reliable basic data for subsequent analysis and ensuring the effectiveness of the entire system.

[0053] The trajectory space acquisition module obtains the overall length of the manipulator, draws a sphere based on the overall length of the manipulator, and evenly divides the internal space of the sphere into multiple subspaces to obtain the trajectory space. The specific method is as follows:

[0054] Obtain the overall length of the manipulator when all joints are in the extended position, i.e., the overall length of the manipulator when all joints are in the same direction, and label it as arm length L. Draw a sphere Q with arm length L as the radius, and divide the internal space of the sphere into multiple subspaces Ei. At the same time, use Ei as the spatial label of each subspace, where i represents different subspaces, i=1,2,...,a1, where a1 represents the total number of subspaces, a1 is a positive integer, and a1≥6;

[0055] It should be noted that the overall length of the manipulator is the total length of each joint of the manipulator, which can be obtained by measurement or from the robot manipulator's instruction manual and the corresponding system. These are all existing and mature technologies, so they will not be elaborated on here.

[0056] By plotting the robot's working area as a spherical trajectory space and dividing it into uniform subspaces, a unified reference system is provided for subsequent trajectory analysis. This spatial quantization method allows trajectory data from different tasks and periods to be compared and analyzed in the same coordinate system, greatly simplifying the complexity of data processing.

[0057] The stepped cycle acquisition module obtains the baseline cycle for the robot manipulator when performing multiple identical product manufacturing tasks, based on the production cycles corresponding to each task. This baseline cycle is then divided into multiple stepped cycles. The specific method is as follows:

[0058] The production cycles corresponding to the robot manipulator when performing multiple identical product manufacturing tasks are obtained. The average of the maximum and minimum values ​​in each production cycle is taken as the reference cycle corresponding to the robot manipulator when performing the corresponding product manufacturing task. The reference cycle is evenly divided at the same time interval to obtain multiple stepped cycles Jj, that is, the time interval between each stepped cycle is the same, where j refers to different stepped cycles, j=1, 2, ..., a2, where a2 refers to the total number of stepped cycles, a2 is a positive integer, a2≥3;

[0059] By averaging and dividing historical production cycles at equal intervals, discretizing continuous time data into multiple stepped cycles can effectively eliminate random errors in a single task, highlight the macroscopic trend of robot operation cycle changes over time, and provide a stable time dimension for subsequent reference area division.

[0060] The reference region acquisition module combines and analyzes the historical motion trajectories and trajectory spaces of each joint node of the robot arm when performing multiple identical product manufacturing tasks to obtain the reference regions corresponding to each joint node in different step cycles. Specifically, the method is as follows:

[0061] S1: Select any one of the different step cycles as the analysis period;

[0062] S2: Select any one of the joint nodes of the robot arm as the analysis node.

[0063] S3: Align the center point of the leftmost end face of the first joint of the robot arm with the center of the sphere in the trajectory space as the alignment point. Simultaneously, place the historical motion trajectories of the analysis node during multiple identical product manufacturing tasks within the trajectory space. Obtain the spatial labels of the subspaces involved in each historical motion trajectory of the analysis node within the trajectory space. Extract the spatial labels of the subspaces involved in each historical motion trajectory from the trajectory space and associate them with the corresponding analysis node. This serves as the activity area of ​​the analysis node during the analysis period when manufacturing multiple products. Combine the subspaces contained in each activity area as the reference area corresponding to the analysis node within the analysis period. Simultaneously, associate the spatial labels of the commonly involved subspaces with the analysis node as the reference area G11 corresponding to the analysis node within the analysis period.

[0064] S4: Repeat steps S2-S3, using the same analysis method to analyze the remaining joint nodes, and obtain the reference region G1d corresponding to each joint node within the analysis period;

[0065] S5: Repeat steps S2-S4, using the same analysis method to analyze the remaining step cycles, and obtain the reference region Gjd corresponding to each node in different step cycles;

[0066] By combining historical motion trajectories with trajectory space and performing cross-analysis of multiple historical motion trajectories, reference regions for each joint within different step cycles are extracted. These reference regions represent the safe range of motion of the robot joints under normal operating conditions, providing a benchmark for subsequent real-time monitoring and anomaly detection.

[0067] The situation dependency coefficient acquisition module takes the rightmost joint node of the robot arm as the operating node (the last joint node counted from left to right), and marks the remaining nodes as sub-nodes. Each two adjacent step cycles constitute a calibration cycle. The center point of the left end face of the first joint of the robot arm is aligned with the center point of the trajectory space to form the alignment point. A three-dimensional coordinate system is then drawn in the trajectory space with the alignment point as the origin to obtain the center point coordinates of the subspace. The module analyzes the center point coordinates of the subspaces contained within the reference regions of each joint node in different step cycles and the corresponding reference region volumes to obtain the situation dependency coefficients between each sub-node and the operating node. The specific method is as follows:

[0068] S01: Take the rightmost joint node among all the joint nodes of the robot arm as the operation node, and mark the remaining nodes as sub-nodes.

[0069] S02: Select any step cycle from different step cycles as the target cycle;

[0070] Obtain the center point coordinates of each subspace contained in the reference area corresponding to the operation node within the target period. Use the average of the x-coordinate and y-coordinate of each center point coordinate as the punctuation coordinates B1(C1x, C1y) of the operation node in the reference area within the target period, where C1x is the average of the x-coordinates of each center point coordinate and C1y is the average of the y-coordinates of each center point coordinate.

[0071] via V=3 / 4πL 3 The volume V of the trajectory space is calculated. The sub-volume F corresponding to a single sub-space is calculated using F=V / a1. The number n of sub-spaces contained in the reference region corresponding to the operation node within the target period is obtained. The product of the number of sub-spaces n and the sub-volume F is taken as the reference region volume H1 corresponding to the reference region of the operation node within the target period.

[0072] S03: Using the same analysis method as in step S02, analyze the coordinates of the center points of each subspace contained in the reference region within each remaining step cycle, and then obtain the coordinates Bj(Cjx, Cjy) of the operation node in each step cycle:

[0073] Simultaneously, the same analysis method as in step S02 is used to analyze each subspace contained in the reference region corresponding to the operation node in the remaining step cycle, to obtain the reference region volume Hj corresponding to the reference region of the operation node in each step cycle;

[0074] S04: Take each two adjacent step cycles as a calibration cycle q, and then obtain multiple calibration cycles. Take the ratio between the absolute value of the difference between the vertical coordinates and the absolute value of the difference between the horizontal coordinates of the calibration points in each two adjacent step cycles as the position migration coefficient of the operating node between each two adjacent step cycles, and then obtain the position migration coefficient Qq of the operating node in each calibration cycle, where q refers to different calibration cycles.

[0075] The absolute value of the difference between the reference region volumes corresponding to the reference regions of the operating node in every two adjacent step cycles is used as the spatial migration coefficient Kq of the operating node in each calibration cycle.

[0076] S05: Select a joint node from each sub-node as the target node, obtain the center point coordinates of each subspace contained in the reference area corresponding to the target node within the target period, and analyze it in the same way as steps S02-S04 to obtain the position migration coefficient and spatial migration coefficient of the target node within each calibration period.

[0077] The position and spatial migration coefficients of the operator node in each calibration period are compared and analyzed with those of the target node in each calibration period. Based on the analysis results, the situational dependency coefficient Y1 between the target node and the operator node is obtained. The specific method is as follows:

[0078] From the position and spatial migration coefficients of the operating node and target node in each calibration period, obtain the position and spatial migration coefficients of the operating node and target node within the same calibration period. When both the position and spatial migration coefficients of the operating node and target node within the same calibration period are not 0, the corresponding calibration period is marked as a two-level transformation period. When both the position and spatial migration coefficients of the operating node and target node within the corresponding calibration period are not 0 and both are 0, the corresponding calibration period is marked as a single-level transformation period. When both the spatial and position migration coefficients of the operating node and target node within the corresponding calibration period are not 0 and both are 0, the corresponding calibration period is marked as a single-level transformation period. The period is marked as a single-layer transformation cycle, and no processing is performed in other cases. The position migration coefficient and spatial migration coefficient of the operation node and the target node in each calibration cycle are compared and analyzed one by one in each calibration cycle to complete the marking of each calibration cycle. The number of double-layer transformation cycles and single-layer transformation cycles corresponding to each calibration cycle are obtained. The sum of the products of the number of double-layer transformation cycles and single-layer transformation cycles with β1 and β2 respectively is used as the situation dependence coefficient Y1 between the target node and the operation node. That is, situation dependence coefficient Y1 = number of double-layer transformation cycles × β1 + number of single-layer transformation cycles × β2, where β1 and β2 are preset coefficients, satisfying 1 = β1 + β2 and β1 ≥ β2;

[0079] S06: Analyze the remaining sub-nodes using the same analysis method as in step S05, and obtain the situation dependence coefficient Yf between each sub-node and the operation node, where f refers to different sub-nodes;

[0080] By introducing position migration coefficient and spatial migration coefficient, the motion and spatial changes of each joint in different cycles are quantified. By comparing and analyzing the migration coefficients between the operation node and the sub-nodes, the situation dependence coefficient is calculated. The situation dependence coefficient intuitively reflects the motion correlation strength between each sub-node and the operation node, providing a key basis for subsequent anomaly tracing.

[0081] The dependency node determination module determines the dependency nodes corresponding to the operation nodes based on the situational dependency coefficients between each sub-node and the operation node. Specifically, the method is as follows:

[0082] Obtain the mean Yp of the situation dependence coefficient Yf between each sub-node and the operation node, and the absolute value J of the difference between the maximum and minimum values ​​of the situation dependence coefficient Yf. Use the difference between the mean Yp and the absolute value J as the boundary value M. Determine the sub-nodes whose situation dependence coefficient Yf is greater than the boundary value M as the dependent nodes corresponding to the operation nodes. Otherwise, do not perform any processing, thus completing the determination of the dependent nodes corresponding to the operation nodes.

[0083] By dynamically setting the limit value using the mean and range of the situation dependence coefficient, this adaptive judgment method avoids the limitations of fixed thresholds and can intelligently identify the dependent joints most strongly associated with the operation node, ensuring the accuracy of the early warning.

[0084] The abnormal node marking module acquires the real-time trajectory points of the joint nodes corresponding to the robot manipulator's operation nodes and their dependent nodes in the trajectory space during the robot manipulator's product manufacturing task. Based on the acquisition time of the real-time trajectory points, it obtains the real-time step cycle corresponding to the acquisition time of the real-time trajectory points from different step cycles. From the reference areas corresponding to each joint node in different step cycles, it obtains the reference areas corresponding to the operation nodes and their dependent nodes in the real-time step cycle and uses them as the comparison areas corresponding to the operation nodes and their dependent nodes in the real-time step cycle. When the real-time trajectory point of the operation node is located in the comparison area corresponding to its real-time step cycle, no processing is performed. When the real-time trajectory point of the operation node is not located in the comparison area corresponding to its real-time step cycle, the operation node is marked as a primary abnormal node. At the same time, the real-time trajectory points of each dependent node are compared with the comparison areas corresponding to their real-time step cycles one by one. Dependent nodes whose real-time trajectory points are not located in their corresponding comparison areas are marked as cooperative abnormal nodes. Otherwise, no processing is performed. The primary abnormal node and cooperative abnormal node are output simultaneously.

[0085] When the real-time trajectory point of an operating node deviates from its reference area, the system immediately marks it as a primary abnormal node. Simultaneously, it checks the real-time trajectory of its dependent joints and marks the dependent joints that deviate from the reference area as collaborative abnormal nodes. This linkage marking mechanism of primary and collaborative abnormalities not only enables timely detection of problems but also allows for tracing the root cause of the problem. This facilitates rapid fault location by staff and the implementation of targeted maintenance measures, effectively preventing situations such as decreased product quality, reduced production efficiency, equipment failure, and safety accidents caused by untimely handling of abnormalities. It achieves comprehensive and accurate perception and effective early warning of the robot's operational status.

[0086] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0087] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A robot operation situation awareness and early warning system based on multi-source data fusion, characterized in that, include: The historical operation data acquisition module calibrates the joint nodes corresponding to each joint of the robot arm and acquires the historical motion trajectory and production cycle of each joint node when the robot arm performs multiple identical product manufacturing tasks. The trajectory space acquisition module draws a sphere based on the overall length of the manipulator, and evenly divides the internal space of the sphere into multiple subspaces to obtain the trajectory space; The step cycle acquisition module obtains the baseline cycle based on each production cycle and divides the baseline cycle evenly into multiple step cycles. The reference area acquisition module combines and analyzes each historical motion trajectory with the trajectory space to obtain the reference areas corresponding to each joint node in different step cycles. The situation dependence coefficient acquisition module takes the rightmost joint node of the robot arm as the operation node and marks the remaining nodes as sub-nodes. At the same time, it obtains the center point coordinates of each subspace and analyzes the center point coordinates and reference area volume of the subspace contained in the reference area of ​​each joint node in different step cycles to obtain the situation dependence coefficient between each sub-node and the operation node. The dependency node determination module determines the dependency nodes corresponding to the operation nodes based on the situation dependency coefficient. The abnormal node marking module compares and analyzes the real-time trajectory points of the robot arm's operating nodes and dependent nodes in the trajectory space with the corresponding comparison areas of the operating nodes and their dependent nodes, and marks the main abnormal nodes and cooperative abnormal nodes.

2. The robot operation situation awareness and early warning system based on multi-source data fusion according to claim 1, characterized in that, The specific method for obtaining the trajectory space is as follows: Mark the overall length of the manipulator as arm length L. Draw a sphere Q with arm length L as the radius. Divide the internal space of the sphere into multiple subspaces Ei. Use Ei as the spatial label of each subspace, where i represents a different subspace, i = 1, 2, ..., a1, where a1 represents the total number of subspaces, a1 is a positive integer, and a1 ≥ 6.

3. The robot operation situation awareness and early warning system based on multi-source data fusion according to claim 2, characterized in that, The specific method for uniformly dividing the reference period into multiple stepped periods is as follows: The production cycle corresponding to the robot arm performing multiple identical product manufacturing tasks is obtained. The average of the maximum and minimum values ​​in each production cycle is used as the base cycle. The base cycle is evenly divided into multiple stepped cycles Jj at the same time interval, where j represents different stepped cycles.

4. The robot operation situation awareness and early warning system based on multi-source data fusion according to claim 3, characterized in that, The specific method for obtaining the reference regions corresponding to each joint node in different step cycles is as follows: S1: Select any one of the different step cycles as the analysis period; S2: Select any one of the joint nodes of the robot arm as the analysis node. S3: Align the center point of the leftmost end face of the robot arm with the center of the sphere in the trajectory space. Simultaneously, place the historical motion trajectories of the analysis node during the analysis cycle when performing multiple identical product manufacturing tasks within the trajectory space. Obtain the spatial labels of the subspaces involved in each historical motion trajectory of the analysis node within the analysis cycle. Extract the spatial labels of the subspaces involved in each historical motion segment from the trajectory space and associate them with the corresponding analysis node. Use these labels as the activity areas of the analysis node during the analysis cycle when performing multiple manufacturing products. Combine the subspaces contained in each activity area as the reference area corresponding to the analysis node within the analysis cycle. S4: Repeat steps S2-S3 and use the same analysis method to analyze the remaining joint nodes to obtain the reference area corresponding to each joint node within the analysis period. S5: Repeat steps S2-S4, using the same analysis method to analyze the remaining step cycles, and obtain the reference area corresponding to each node in different step cycles.

5. The robot operation situation awareness and early warning system based on multi-source data fusion according to claim 4, characterized in that, The specific method for obtaining the situational dependency coefficients between each sub-node and the operation node is as follows: S01: Designate the rightmost joint node of the robot arm as the operation node and mark the remaining nodes as sub-nodes. S02: Select any step cycle from different step cycles as the target cycle; Obtain the center point coordinates of each subspace contained within the reference region of the operating node within the target period. Take the average of the x and y coordinates of each center point and use it as the x and y coordinates of the corresponding punctuation point within the reference region of the operating node within the target period. Calculate the sub-volume F corresponding to a single subspace using F=V / a1, where V is the volume of the trajectory space. Obtain the number of subspaces contained within the reference region of the operating node within the target period. Use the product of the number of subspaces and the sub-volume F as the reference region volume corresponding to the operating node within the target period. S03: Using the same analysis method as in step S02 for obtaining the coordinates of the reference area corresponding to the operation node within the target period, analyze each remaining step period to obtain the coordinates of the reference area corresponding to the operation node in each step period. Simultaneously, using the same analysis method as in step S02 for obtaining the volume of the reference area corresponding to the operation node within the target period, analyze the number of subspaces contained in the reference area within each remaining step period to obtain the volume of the reference area corresponding to the operation node in each step period. S04: Take each two adjacent step cycles as a calibration cycle, and then obtain multiple calibration cycles. Analyze the vertical and horizontal coordinates of the calibration points in each two adjacent step cycles to obtain the position migration coefficient of the operation node in each calibration cycle. At the same time, take the absolute value of the difference between the reference area volumes of the reference areas of the operation node in each two adjacent step cycles as the spatial migration coefficient of the operation node in each calibration cycle. S05: Select a joint node from each sub-node as the target node, and use the same analysis method as in steps S02-S04 to analyze the center point coordinates of each subspace contained in the reference area of ​​the target node within the target period, thereby obtaining the position migration coefficient and spatial migration coefficient of the target node within each calibration period. The position migration coefficient and spatial migration coefficient of the operating node in each calibration period are compared and analyzed with those of the target node in each calibration period. Based on the analysis results, the situational dependence coefficient between the target node and the operating node is obtained. S06: Analyze the remaining sub-nodes using the same analysis method as in step S05 to obtain the situational dependence coefficients between each sub-node and the operation node.

6. The robot operation situation awareness and early warning system based on multi-source data fusion according to claim 5, characterized in that, The specific method for obtaining the position migration coefficients of the operating node in each calibration period is as follows: The ratio between the absolute difference of the ordinate and the absolute difference of the abscissa in the coordinates of each two adjacent step cycles is used as the position migration coefficient of the operation node in each calibration cycle.

7. The robot operation situation awareness and early warning system based on multi-source data fusion according to claim 6, characterized in that, The specific method for obtaining the situational dependency coefficient between the target node and the operator node is as follows: From the position and spatial migration coefficients of the operating node and target node in each calibration period, obtain the position and spatial migration coefficients of the operating node and target node in the same calibration period. When both the position and spatial migration coefficients of the operating node and target node in the same calibration period are not 0, the corresponding calibration period is marked as a two-level transformation period. When both the position and spatial migration coefficients of the operating node and target node in the corresponding calibration period are not 0 and both are 0, the corresponding calibration period is marked as a single-level transformation period. When all migration coefficients are non-zero and the position migration coefficient is zero, the corresponding calibration period is marked as a single-layer transformation period. Otherwise, no processing is performed. The position migration coefficients and spatial migration coefficients of the operating node and the target node in each calibration period are compared and analyzed one by one. At the same time, the number of double-layer transformation periods and single-layer transformation periods corresponding to each calibration period are obtained. The sum of the products of the number of double-layer transformation periods and single-layer transformation periods with the preset coefficients β1 and β2 is used as the situational dependence coefficient between the target node and the operating node, where 1 = β1 + β2 and β1 ≥ β2.

8. The robot operation situation awareness and early warning system based on multi-source data fusion according to claim 5, characterized in that, The specific method for determining the dependent nodes corresponding to the operation node is as follows: Obtain the mean value of the situation dependency coefficients corresponding to each sub-node and the operation node, as well as the absolute value of the difference between the maximum and minimum values ​​of the situation dependency coefficients. Use the difference between the mean value and the absolute value of the difference as the boundary value. Sub-nodes with situation dependency coefficients greater than the boundary value are determined as dependent nodes corresponding to the operation nodes. Otherwise, no processing is performed.

9. The robot operation situation awareness and early warning system based on multi-source data fusion according to claim 8, characterized in that, The specific method for determining and marking primary and secondary anomalous nodes is as follows: The system acquires real-time trajectory points for the robot's manipulator nodes and their dependent nodes. Based on the acquisition time of the real-time trajectory points, it obtains the real-time ladder cycle corresponding to the acquisition time from different ladder cycles. From the reference areas corresponding to each node in different ladder cycles, it obtains the reference areas corresponding to the manipulator nodes and their dependent nodes in the real-time ladder cycle and uses these as the comparison areas for the manipulator nodes and their dependent nodes in the real-time ladder cycle. When the real-time trajectory point of the manipulator node is within its corresponding comparison area in the real-time ladder cycle, no processing is performed. When the real-time trajectory point of the manipulator node is not within its corresponding comparison area in the real-time ladder cycle, the manipulator node is marked as a primary abnormal node. At the same time, the real-time trajectory points of each dependent node are compared with their corresponding comparison areas in the real-time ladder cycle one by one. Dependent nodes whose real-time trajectory points are not within their corresponding comparison areas are marked as cooperative abnormal nodes. Otherwise, no processing is performed. Both primary abnormal nodes and cooperative abnormal nodes are output simultaneously.

Images