An engineering robot failure prediction method and system

Abstract

The present application relates to the technical field of data anomaly detection, and proposes an engineering robot fault prediction method and system, comprising: collecting pressure data of each component of the engineering robot in multiple handling operations; analyzing the stability of the pressure data of each component based on density clustering; analyzing the incoherence of each handling operation according to the pressure change time and the interval time; comprehensively determining the influence degree of each component in each operation according to the coordination of the change time, the incoherence and the stability; finally, combining the number of handling operations and the historical pressure data, training the autoregressive model and comparing the predicted value with the actual value to realize fault prediction. The present application solves the problem that the prior art cannot accurately identify the fault of multiple components working together.

Description

Technical Field

[0001] This invention relates to the field of data anomaly detection technology, specifically to a method and system for predicting faults in engineering robots. Background Technology

[0002] Engineering robots are intelligent robots specifically designed for automated material handling, including grasping, moving, and placing, in industrial, construction, and logistics environments. They are typically equipped with end effectors such as robotic arms, vacuum suction cups, or grippers, and integrate sensors, navigation, and control systems to autonomously or semi-autonomously perform repetitive, high-intensity material handling tasks. For example, they enable automated workpiece handling and positioning in machine tool loading and unloading, stamping production lines, or assembly lines. However, engineering robots often operate in high-temperature, high-load, or human-robot mixed environments, and sudden malfunctions can lead to equipment damage, production interruptions, and even personnel injuries. By monitoring parameters such as vibration, temperature, and the condition of the ropes in real time, and identifying trends in component performance degradation, unplanned downtime can be effectively avoided, thereby ensuring the continuous execution of critical tasks.

[0003] However, existing fault monitoring for engineering robots primarily focuses on independent health monitoring of individual components (such as robotic arm joints and drive motors) through threshold judgment or trend analysis. While these methods are effective at detecting performance degradation in individual components, they struggle to effectively capture and diagnose systemic faults caused by the malfunction of multiple components working in tandem. For example, a slight deviation in the timing of the gripper's and the moving platform's movements may lead to handling failure, even if the pressure data of both the gripper and the platform remain within normal ranges. When different parts fail to coordinate properly, resulting in misalignment, existing fault monitoring methods often cannot accurately identify the fault state of such engineering robots. Summary of the Invention

[0004] This invention provides a method and system for predicting faults in engineering robots, to solve the problem of inaccurately identifying faults in which different parts cannot cooperate to complete tasks. The specific technical solution adopted is as follows:

[0005] In a first aspect, one embodiment of the present invention provides a method for predicting faults in engineering robots, the method comprising the following steps:

[0006] The pressure data of all components of the engineering robot are collected at different times during the second preset number of handling operations, until the data collection stops at the current time.

[0007] Based on the similarity between all pressure data collected during all handling operations of the same component, a cluster of the same component is obtained. Based on the differences between pressure data contained in the cluster of the same component and the number of pressure data, as well as the number of clusters of the same component, the stability of all pressure data collected during all handling operations of the same component is determined.

[0008] Based on the differences between all adjacent pressure data collected from the same component during the handling process, the pressure change interval time at the start time, change time, and change time is determined. Based on the differences in the pressure change interval at the corresponding change time of the same component in different handling processes, as well as the differences in stability, the discontinuity of each handling operation is determined.

[0009] Based on the number of identical change moments corresponding to different components of the engineering robot during the handling process, the duration of consecutive occurrence of identical change moments, the discontinuity of the engineering robot during the handling process, and the stability of all pressure data collected from the components of the engineering robot during the handling process, the degree of influence of each component of the engineering robot during the handling process is determined.

[0010] By combining all the data collected by the engineering robot, fault prediction can be achieved.

[0011] Furthermore, the method for obtaining the clusters of the same component is as follows:

[0012] A Cartesian coordinate system is established with the acquisition time as the x-axis and the pressure data as the y-axis. The pressure data of the same component at all acquisition times during the second preset number of handling operations are marked as points in the Cartesian coordinate system. Density clustering is performed on all points in the Cartesian coordinate system to obtain the cluster of the same component.

[0013] Furthermore, the method for determining the stability of all pressure data collected during all handling processes of the same component based on the differences and quantity of pressure data within clusters of the same component, as well as the number of clusters of the same component, includes the following specific methods:

[0014] The pressure data of any component at any time during any handling operation is recorded as the target pressure data.

[0015] The number of all pressure data contained in the component corresponding to the target pressure data during the same handling operation is recorded as the first number of target pressure data.

[0016] The number of clusters of components corresponding to the target pressure data is denoted as the second number of the target pressure data.

[0017] The stability of the target pressure data is determined based on the number of points contained in the cluster where the target pressure data is located, the first number of target pressure data points, and the second number of target pressure data points.

[0018] Furthermore, the method for determining the stability of the target pressure data based on the number of points contained in the cluster where the target pressure data is located, the first number of target pressure data points, and the second number of target pressure data points includes the following specific methods:

[0019] The ratio of the number of points contained in the cluster containing the target pressure data to the first number of the target pressure data is denoted as the first ratio of the target pressure data. The normalized value of the ratio of the first ratio of the target pressure data to the second number is denoted as the stability of the target pressure data.

[0020] Furthermore, the method for determining the pressure change interval at the start time, the change time, and the change time based on the differences between all adjacent pressure data collected from the same component during the handling process includes:

[0021] All pressure data collected for the same component during the same handling process are compared with the previous adjacent pressure data. When the pressure data is not equal to the previous adjacent pressure data, the collection time corresponding to the pressure data is marked as the change time.

[0022] The first data acquisition moment during the same handling process of the same component is marked as the start moment;

[0023] The time interval between the moment of change and the start time of the same component during the same handling process is denoted as the pressure change interval time at the moment of change.

[0024] Furthermore, the formula for calculating the discontinuity of the transport operation is:

[0025]

[0026] In the formula, Indicates the first to be evaluated The number of transport operations; C represents the number of parts contained in the engineering robot; Z represents the number of all transport operations; c is the transport operation index, traversing from 1 to Z transport operations; a is the part index, traversing from 1 to C parts; N represents the minimum number of change moments for the a-th part during the c-th and x-th transport operations; i is the change moment index, traversing from 1 to N. Indicates the first The component in the first During the second handling operation, the first Stability of pressure data at each changing moment; Indicates the first The component in the first During the second handling operation, the first Stability of pressure data at each changing moment; and These represent the time intervals between pressure changes at corresponding moments. This represents the normalization function.

[0027] Furthermore, the method for determining the degree of impact of each component of the engineering robot during the handling process is as follows:

[0028] Obtain any one of the components of the engineering robot and denote it as component o;

[0029] The sum of the stability of all pressure data collected by component o in the xth handling operation is used as the first factor, and the ratio of the discontinuity of the xth handling operation is used as the first factor. The first factor is multiplied by the exponential decay factor to obtain the product result. The normalization function is applied to the product result to obtain the degree of influence of component o in the xth handling operation.

[0030] The exponential decay factor is obtained in the following way:

[0031] For each other component a in the engineering robot, calculate the duration of the continuous identical change moments between component o and component a in the xth handling operation, and record it as the first time length; obtain the number of identical change moments between component o and component a in the xth handling operation, and record it as the first change quantity; and record the product of the first time length and the first change quantity as the cooperative influence value corresponding to component a.

[0032] Summing the synergistic impact values ​​corresponding to all components a, we obtain the total synergistic impact.

[0033] Using the natural constant as the base and the negative value of the sum of the synergistic effects as the exponential function, the value of the exponential function is calculated, which is the exponential decay factor.

[0034] Furthermore, the specific methods for combining all the data collected by the engineering robot to achieve fault prediction include:

[0035] The normalized value of the product of the number of handling tasks completed by the engineering robot at the current acquisition time and the mean of all pressure data collected for the handling tasks at the current acquisition time is denoted as the adjustment weight at the current acquisition time.

[0036] An autoregressive model is trained based on the pressure data from the first preset number of consecutive handling processes preceding the current handling operation and the pressure data collection time, to obtain the collection time-pressure curve. The sum of the impact of the robot's components on all handling operations is recorded as the first weight of the robot's components. The product of the adjustment weight at the current collection time and the first weight of the robot's components is used as the autocorrelation coefficient of the pressure data collected at the component positions of the robot in the autoregressive model. The independent variable of the time-pressure curve is the collection time, and the dependent variable is the pressure data of each component.

[0037] Based on the difference between the pressure data of each component in the time-pressure curve and the pressure data at the current time of acquisition, fault prediction of engineering robots can be achieved.

[0038] Furthermore, the method for predicting engineering robot faults based on the difference between the pressure data values ​​of each component in the time-pressure curve and the pressure data at the current time of acquisition includes the following specific methods:

[0039] The pressure data of each component in the time-pressure curve is taken as the standard value corresponding to the pressure data.

[0040] The engineering robot is deemed to have malfunctioned when the absolute value of the difference between the standard value corresponding to the pressure data at the current acquisition time and the pressure data at the current acquisition time is greater than or equal to the fault threshold.

[0041] If the absolute value of the difference between the standard value corresponding to the pressure data at the current acquisition time and the pressure data at the current acquisition time is less than the fault threshold, the engineering robot is determined to be without fault.

[0042] Secondly, embodiments of the present invention also provide an engineering robot fault prediction system, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0043] The beneficial effects of this invention are:

[0044] This application takes a handling robot in engineering robots as an example to predict the failure of engineering robots. First, it analyzes the different stress conditions of different components during the object handling process, noting that the wear and tear of different components varies greatly, and components with greater wear and tear are more likely to fail. Based on the uniformity and repeatability of the pressure experienced by the same component at different data collection times, the degree of wear on the component at each pressure data collection time is evaluated, and the stability of all pressure data collected for the same component throughout all handling processes is obtained. Considering that abnormal pressure failures of a component will affect the continuity of time in different stages, the continuity of pressure data for each component is analyzed and evaluated to determine the discontinuity of each handling operation. Furthermore, it analyzes the influence of all other components on a particular component based on the number of identical change times corresponding to different components during the handling operation, the duration of consecutive identical change times, the discontinuity of the handling operation, and the stability of all pressure data collected for each component during the handling operation, thus obtaining the degree of influence of each component of the engineering robot during the handling operation. Finally, by combining all the data collected from the engineering robot, failure prediction is achieved, solving the problem of not being able to accurately identify failures where different parts cannot cooperate to complete the work. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart illustrating a method for predicting faults in an engineering robot according to an embodiment of the present invention.

[0047] Figure 2 This is a flowchart illustrating the stability acquisition process provided in one embodiment of the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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] Please see Figure 1The diagram illustrates a flowchart of a fault prediction method for engineering robots according to an embodiment of the present invention. The method includes the following steps:

[0050] Step S001: Collect pressure data of all components of the engineering robot at different collection times during the second preset number of handling operations, until data collection stops at the current collection time.

[0051] This embodiment selects a handling robot from engineering robots as an example to perform fault prediction for engineering robots.

[0052] Material handling robots are primarily used for moving objects. Different components experience varying pressures from the objects being moved, making them susceptible to wear and tear during the handling process. Therefore, pressure sensors are installed on all components of the material handling robot. Starting from the moment the robot begins operation, pressure data is collected every second using these sensors, and the timing of each data collection is recorded until the robot completes its handling task and stops collecting data.

[0053] Preferably, in one embodiment of this application, pressure data and pressure data collection times are collected for the current transport operation and the first preset number of transport operations preceding the current transport operation. In this embodiment, the first preset number is 100. The sum of the first preset number and the number 1 is recorded as the second preset number. That is, pressure data and pressure data collection times are collected for the current transport operation and the 100 consecutive transport operations preceding the current transport operation, for a total of 101 pressure data and pressure data collection times during the transport operations. 101 is the second preset number.

[0054] It is understandable that during the current data collection time, the transport robot may not have completed its transport work at that time. Therefore, during the current data collection time, the transport work will continue until the current data collection time stops.

[0055] At this point, pressure data for all components of the handling robot at different times during the second preset number of handling operations have been obtained.

[0056] Step S002: Based on the similarity between all pressure data collected during all handling processes of the same component, obtain the cluster of the same component. Based on the differences between the pressure data contained in the cluster of the same component and the number of pressure data, as well as the number of clusters of the same component, determine the stability of all pressure data collected during all handling processes of the same component.

[0057] When a transport robot is transporting objects, the objects it needs to move are heavy. During the transport process, different parts are subjected to different forces, and the wear and tear on different parts varies greatly. Parts with greater wear and tear are more likely to fail.

[0058] When a component of a handling robot exhibits a singular behavior and experiences relatively stable pressure variations during the handling process, that component is less likely to suffer wear and tear and malfunction. Therefore, the degree of wear and tear on a component can be evaluated based on the pressure it experiences at different data collection points.

[0059] A Cartesian coordinate system is established with the acquisition time as the x-axis and the pressure data as the y-axis. The pressure data of the same component at all acquisition times during the second preset number of handling operations are marked as points in the Cartesian coordinate system. Density clustering is performed on all points in the Cartesian coordinate system to obtain the clusters of the same component. The number of clusters of the same component and the number of points contained in each cluster are counted.

[0060] In this embodiment, the DBSCAN (density-based spatial clustering with noise) algorithm is preferably used for density clustering. Its core parameters (such as neighborhood radius Eps and minimum number of points MinPts) can be set according to the data distribution of the specific application scenario. It is understood that those skilled in the art, after understanding the purpose of this invention (i.e., finding dense regions of pressure data in the time-pressure coordinate system), can also use other density-based clustering algorithms (such as OPTICS) to achieve the same goal; this application does not impose any special limitations.

[0061] Based on the differences and number of pressure data within the clusters containing all pressure data collected from the same component during all handling operations, and the clusters of the same component, the stability of all pressure data collected from the same component during all handling operations is determined.

[0062] Preferably, as an embodiment of this application, the pressure data of any component at any time during any handling operation is recorded as the target pressure data; the number of all pressure data of the component corresponding to the target pressure data during the same handling operation is recorded as the first quantity of the target pressure data; the number of clusters of the component corresponding to the target pressure data is recorded as the second quantity of the target pressure data; the ratio of the number of points contained in the cluster of the target pressure data to the first quantity of the target pressure data is recorded as the first ratio of the target pressure data; and the normalized value of the ratio of the first ratio of the target pressure data to the second quantity is recorded as the stability of the target pressure data.

[0063] When the number of points contained in the cluster containing the target pressure data is greater than the number of points in the target pressure data, the target pressure data is closer to the values ​​of other pressure data generated by the same component in the same handling process, and the less likely the target pressure data is to be abnormal. At the same time, when the number of clusters of the component corresponding to the target pressure data is smaller, the values ​​of the pressure data collected by the component corresponding to the target pressure data during the handling process are closer, and the behavior of the component corresponding to the target pressure data during the handling process is more singular and more repetitive. In this case, the stability of the target pressure data is greater.

[0064] The same method can be used to obtain the stability of pressure data for all components at all collection points during all handling operations. The flowchart for obtaining stability is as follows: Figure 2 As shown.

[0065] This establishes the stability of pressure data for all components at all acquisition points during all handling operations.

[0066] Step S003: Based on the differences between all adjacent pressure data collected from the same component during the handling process, determine the pressure change interval time at the start time, change time, and change time. Based on the differences in the pressure change interval at the corresponding change time of the same component in different handling processes, as well as the differences in stability, determine the discontinuity of each handling operation.

[0067] When a handling robot performs a handling operation, it requires the cooperation of various transmission systems. First, the transmission system provides power, and the robot grasps the object. After grasping, another transmission system moves the object to its designated location. Then, other transmission systems place the object in the predetermined position, completing the handling operation. The robot then repeats the process for the next item handling activity. In other words, the coordination between different transmission components is fixed in each item handling process, and the different actions are sequential in time. When a component experiences an abnormal pressure failure, such as a delayed pressure response, it will affect the sequentiality of the different stages. Therefore, it is essential to analyze and evaluate the consistency of pressure data for each component.

[0068] All pressure data collected for the same component during the same handling process are compared with the previous adjacent pressure data. When the pressure data is not equal to the previous adjacent pressure data, the collection time corresponding to the pressure data is marked as the change time. The first collection time of the same component during the same handling process is marked as the start time. The time interval between the change time and the start time of the same component during the same handling process is recorded as the pressure change interval time of the change time.

[0069] It is understandable that the pressure data of the first acquisition moment of the same component during the same handling operation does not have the previous adjacent pressure data. Therefore, the first acquisition moment during the handling operation is not discussed when marking the change moment.

[0070] Based on the differences in the pressure change intervals of the same component in different handling processes, and the differences in the stability of the pressure data at the corresponding change times, the discontinuity of each handling operation is determined.

[0071]

[0072] In the formula, Indicates the first The inconsistency of the secondary handling work; Represents the normalization function; Indicates the first The component in the first During the second handling operation, the first Stability of pressure data at each changing moment; Indicates the first The component in the first During the second handling operation, the first Stability of pressure data at each changing moment; Indicates the first The component in the first During the second handling operation, the first The time interval between pressure changes at each moment of change; Indicates the first The component in the first During the second handling operation, the first The time interval between pressure changes at each moment of change; Indicates the first The component in the first Second and third The minimum number of times of change during a single handling operation; This represents the total number of all handling processes, i.e., the second preset number. In this embodiment, the value of the second preset number is 101. This indicates the number of parts contained in the transport robot.

[0073] When the difference between the pressure change intervals of the same component in different handling processes is greater, and the difference between the stability of the pressure data at the corresponding change times is greater, the component is more likely to malfunction or fail. The more complex the behavior of the component is in different handling processes, the greater the possibility of the handling robot failing in this handling process. At this time, the discontinuity of the handling work is greater.

[0074] This completes the understanding of the discontinuity in all the handling tasks performed by the handling robot.

[0075] Step S004: Based on the number of identical change moments corresponding to different parts of the engineering robot during the handling process, the duration of consecutive occurrence of identical change moments, the discontinuity of the engineering robot during the handling process, and the stability of all pressure data collected by the parts of the engineering robot during the handling process, determine the degree of influence of each part of the engineering robot during the handling process.

[0076] The behavior of a handling robot during the handling process affects subsequent actions. For example, when a handling robot grasps and moves an object, due to the object's inertia, some components of the robot remain under significant pressure even after the movement stops; this pressure does not immediately decrease upon cessation. Therefore, the degree of influence of all other components on a particular component can be analyzed based on the behavior of all other components of the handling robot.

[0077] Based on the number of identical change moments corresponding to different components of the handling robot during the handling process, the duration of consecutive occurrence of identical change moments, the inconsistency of the handling robot during the handling process, and the stability of all pressure data collected from the components of the handling robot during the handling process, the degree of influence of each component of the handling robot during the handling process is determined.

[0078]

[0079] In the formula, Components representing a transport robot In the The degree of impact during the secondary handling operation; Represents the normalization function; Indicates the first The inconsistency of the secondary handling work; Indicates components In the The sum of the stability of all pressure data collected during this transport operation; Represents an exponential function with the natural constant as its base; Indicates components and the The component in the first The number of identical changes during each transport operation; Indicates components and the The component in the first The duration of consecutive identical changes during a single handling operation; This indicates the number of parts contained in the transport robot.

[0080] It is understandable that, for example, if a component and the The component in the first During the second handling operation, if the pressure data acquisition times for the 3rd, 4th, 5th, 8th, and 11th pressure data points are all changing, and the acquisition times for other pressure data points are not changing, the component... and the The component in the first The number of identical change moments in this handling operation is 5. Among these 5 identical change moments, the 3rd, 4th, and 5th pressure data acquisition moments are consecutive change moments. The length of the time period formed by the acquisition moments from the 3rd to the 5th pressure data is the component's... and the The component in the first The duration of consecutive identical changes during a single transport operation.

[0081] Because the effects of inertia and other factors are relatively short-lived, the smaller the number of identical changes occurring at different parts of the handling robot during handling, and the shorter the duration of consecutive occurrences of the same changes, the greater the likelihood that the pressure data changes of the analyzed part are caused by the influence of other parts. At the same time, the greater the stability of all pressure data collected by the analyzed part during handling, and the smaller the discontinuity of the handling operation, the greater the degree of influence the analyzed part is on the handling operation, that is, the greater the degree of influence the analyzed part is on other parts during handling.

[0082] This allows us to determine the degree of impact of each component of the handling robot during the handling process.

[0083] Step S005: Combine all the data collected by the engineering robot to achieve fault prediction.

[0084] All pressure data collected by the handling robot during its handling operations directly reflects its working status. A significant difference between the pressure data collected during handling and historical data suggests a higher likelihood of a component malfunctioning. However, the handling robot experiences wear and tear on its components during operation, causing a gradual deviation in the relationship between the pressure data collected and the timing of these collections. Therefore, the relationship between the collection time and pressure data needs to be adjusted based on the specific circumstances and frequency of the handling operations.

[0085] The normalized value of the product of the number of handling tasks completed by the handling robot at the current data collection time and the mean of all pressure data collected for the handling tasks at the current data collection time is denoted as the adjustment weight for the current data collection time.

[0086] An Autoregressive Model is trained based on the pressure data from a first preset number of consecutive handling operations preceding the current handling operation and the pressure data collection time. The sum of the impacts on the robot's components across all handling operations is designated as the first weight of the robot's components. The product of the adjusted weight at the current collection time and the first weight of the robot's components is used as the autocorrelation coefficient of the pressure data collected at the component locations in the autoregressive model. A collection time-pressure curve is obtained, where the independent variable of the time-pressure curve is the collection time, and the dependent variable is the pressure data of each component.

[0087] The pressure data values ​​of each component in the time-pressure curve are used as the standard values ​​corresponding to the pressure data. Based on the time-pressure curve, the standard values ​​corresponding to the pressure data of each component at the current time of acquisition are obtained. When the absolute value of the difference between the standard value and the pressure data is greater than or equal to the fault threshold, the handling robot is determined to have malfunctioned; when the absolute value of the difference between the standard value and the pressure data is less than the fault threshold, the handling robot is determined not to have malfunctioned.

[0088] The fault threshold is a preset constant value, and in this embodiment, the fault threshold is set to 2.

[0089] This achieves the goal of predicting engineering robot malfunctions.

[0090] Based on the same inventive concept as the above method, this embodiment of the invention also provides an engineering robot fault prediction system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described engineering robot fault prediction methods.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An engineered robot failure prediction method, characterized by, The method includes the following steps: Collect pressure data of all components of the engineering robot at different times during the second preset number of handling operations; Obtain clusters of the same component, and determine the stability of all pressure data collected for the same component during all handling operations based on the differences between pressure data contained in the clusters of the same component, the number of pressure data, and the number of clusters of the same component. Based on the differences between all adjacent pressure data collected from the same component during the handling process, the pressure change interval time at the start time, change time, and change time is determined. Based on the differences in the pressure change interval at the corresponding change time of the same component in different handling processes, as well as the differences in stability, the discontinuity of each handling operation is determined. Based on the number of identical change moments corresponding to different components of the engineering robot during the handling process, the duration of consecutive occurrence of identical change moments, the discontinuity of the engineering robot during the handling process, and the stability of all pressure data collected from the components of the engineering robot during the handling process, the degree of influence of each component of the engineering robot during the handling process is determined. By combining all the data collected by the engineering robot, fault prediction can be achieved.

2. The method of claim 1, wherein, The method for obtaining the clusters of the same component is as follows: A Cartesian coordinate system is established with the acquisition time as the x-axis and the pressure data as the y-axis. The pressure data of the same component at all acquisition times during the second preset number of handling operations are marked as points in the Cartesian coordinate system. Density clustering is performed on all points in the Cartesian coordinate system to obtain the cluster of the same component.

3. The method of claim 1, wherein, The method for determining the stability of all pressure data collected during all handling processes of the same component based on the differences and quantity of pressure data within clusters of the same component, as well as the number of clusters of the same component, includes the following specific methods: The pressure data of any component at any time during any handling operation is recorded as the target pressure data. The number of all pressure data contained in the component corresponding to the target pressure data during the same handling operation is recorded as the first number of target pressure data. The number of clusters of components corresponding to the target pressure data is denoted as the second number of the target pressure data. The stability of the target pressure data is determined based on the number of points contained in the cluster where the target pressure data is located, the first number of target pressure data points, and the second number of target pressure data points.

4. The method of claim 3, wherein, The method for determining the stability of the target pressure data based on the number of points contained in the cluster where the target pressure data is located, the first number of target pressure data points, and the second number of target pressure data points includes the following specific methods: The ratio of the number of points contained in the cluster containing the target pressure data to the first number of the target pressure data is denoted as the first ratio of the target pressure data. The normalized value of the ratio of the first ratio of the target pressure data to the second number is denoted as the stability of the target pressure data.

5. The method of claim 1, wherein, The method for determining the pressure change interval at the start time, the change time, and the change time based on the differences between all adjacent pressure data collected from the same component during the handling process includes the following: All pressure data collected for the same component during the same handling process are compared with the previous adjacent pressure data. When the pressure data is not equal to the previous adjacent pressure data, the collection time corresponding to the pressure data is marked as the change time. The first data acquisition moment during the same handling process of the same component is marked as the start moment; The time interval between the moment of change and the start time of the same component during the same handling process is denoted as the pressure change interval time at the moment of change.

6. The method of claim 1, wherein, The formula for calculating the discontinuity of the transport operation is: In the formula, Indicates the first to be evaluated The number of transport operations; C represents the number of parts contained in the engineering robot; Z represents the number of all transport operations; c is the transport operation index, traversing from 1 to Z transport operations; a is the part index, traversing from 1 to C parts; N represents the minimum number of change moments for the a-th part during the c-th and x-th transport operations; i is the change moment index, traversing from 1 to N. Indicates the first The component in the first During the second handling operation, the first Stability of pressure data at each changing moment; Indicates the first The component in the first During the second handling operation, the first Stability of pressure data at each changing moment; and These represent the time intervals between pressure changes at corresponding moments. This represents the normalization function.

7. The method of claim 1, wherein, The method for determining the degree of impact on each component of the engineering robot during the handling process is as follows: Obtain any one of the components of the engineering robot and denote it as component o; The sum of the stability of all pressure data collected by component o in the xth handling operation is used as the first factor, and the ratio of the discontinuity of the xth handling operation is used as the first factor. The first factor is multiplied by the exponential decay factor to obtain the product result. The normalization function is applied to the product result to obtain the degree of influence of component o in the xth handling operation. The exponential decay factor is obtained in the following way: For each other component a in the engineering robot, calculate the duration of the continuous identical change moments between component o and component a in the xth handling operation, and record it as the first time length; obtain the number of identical change moments between component o and component a in the xth handling operation, and record it as the first change quantity; and record the product of the first time length and the first change quantity as the cooperative influence value corresponding to component a. Summing the synergistic impact values ​​corresponding to all components a, we obtain the total synergistic impact. Using the natural constant as the base and the negative value of the sum of the synergistic effects as the exponential function, the value of the exponential function is calculated, which is the exponential decay factor.

8. The method of claim 1, wherein, The specific methods for combining all the data collected by the engineering robot to achieve fault prediction include: The normalized value of the product of the number of handling tasks completed by the engineering robot at the current acquisition time and the mean of all pressure data collected for the handling tasks at the current acquisition time is denoted as the adjustment weight at the current acquisition time. An autoregressive model is trained based on the pressure data from the first preset number of consecutive handling processes preceding the current handling operation and the pressure data collection time, to obtain the collection time-pressure curve. The sum of the impact of the robot's components on all handling operations is recorded as the first weight of the robot's components. The product of the adjustment weight at the current collection time and the first weight of the robot's components is used as the autocorrelation coefficient of the pressure data collected at the component positions of the robot in the autoregressive model. The independent variable of the time-pressure curve is the collection time, and the dependent variable is the pressure data of each component. Based on the difference between the pressure data of each component in the time-pressure curve and the pressure data at the current time of acquisition, fault prediction of engineering robots can be achieved.

9. The method of claim 8, wherein, The method for predicting engineering robot faults based on the difference between the pressure data of each component in the time-pressure curve and the pressure data at the current time of acquisition includes the following specific methods: The pressure data of each component in the time-pressure curve is taken as the standard value corresponding to the pressure data. The engineering robot is deemed to have malfunctioned when the absolute value of the difference between the standard value corresponding to the pressure data at the current acquisition time and the pressure data at the current acquisition time is greater than or equal to the fault threshold. If the absolute value of the difference between the standard value corresponding to the pressure data at the current acquisition time and the pressure data at the current acquisition time is less than the fault threshold, the engineering robot is determined to be without fault.

10. An engineered robot failure prediction system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein, When the processor executes the computer program, it implements the steps of the method as claimed in any one of claims 1-9.

Images