A method, device, equipment and storage medium for determining mechanical arm faults

Abstract

The present invention relates to the field of robotic arm technology, and provides a robotic arm fault identification method, apparatus, device, and storage medium. The robotic arm fault identification method includes: obtaining multiple first real-time data information of a first target identification area of ​​the robotic arm; extracting first real-time feature values ​​of the multiple first real-time data information, and classifying the first real-time feature values ​​to obtain multiple first real-time feature values ​​of different categories; calculating the corresponding robotic arm first fault probability based on the first real-time feature values ​​of the same category; calculating the total robotic arm first fault probability based on the robotic arm first fault probabilities corresponding to the multiple first real-time feature values ​​of different categories; and realizing fault identification of the first target identification area based on the total robotic arm first fault probability. The present invention comprehensively considers the influence of different types of feature values ​​on robotic arm fault identification, and can accurately realize the identification of robotic arm faults.

Description

Technical Field

[0001] The present invention relates to the technical field of robotic arms, and in particular to a method, device, equipment and storage medium for determining robotic arm faults. Background Art

[0002] A robotic arm, also known as a manipulator, is the most widely used automated mechanical device in robotics technology. It can be found in industrial manufacturing, medical treatment, entertainment services, military, semiconductor manufacturing, and space exploration. To ensure the proper operation of a robotic arm, its real-time working status is typically monitored to predict potential failures and prevent potential safety incidents.

[0003] For example, Chinese invention patent application number CN202210996051.9 discloses a PLC-based intelligent fault detection method for a machining robot arm, which can determine the robot arm's fault information. Another example is Chinese invention patent application number CN201510031469.6, which discloses an iterative learning fault diagnosis method for a single-joint robot arm system. This method can detect and estimate system faults and has a certain adaptability to different types of faults. Another example is Chinese invention patent application number CN202110589534.2, which discloses a real-time industrial robot arm fault identification method. Its estimation results can meet the accuracy and real-time requirements of practical applications.

[0004] The above-mentioned existing technologies only identify robot arm failures based on single-dimensional mechanical features, and do not consider integrating multiple different types of robot arm features to obtain the overall failure probability of the robot arm. Their accuracy needs to be improved. Summary of the Invention

[0005] One purpose of the present invention is to solve the problem that the existing technology only identifies robot arm faults based on a single dimension of mechanical features, and does not consider the integration of multiple different types of robot arm features to obtain the overall failure probability of the robot arm, and its accuracy needs to be improved. A robot arm fault identification method is provided, which can calculate the first failure probability of the robot arm according to multiple different types of first real-time feature values ​​in the target judgment area, and calculate the total probability of the first failure of the robot arm according to multiple robot arm first failure probabilities, and finally realize fault identification based on the total probability of the first failure of the robot arm.

[0006] The robot arm fault identification method comprises the following steps:

[0007] Acquire a plurality of first real-time data information of a first target discrimination area of ​​the robotic arm;

[0008] extracting a plurality of first real-time feature values ​​of the first real-time data information, and classifying the first real-time feature values ​​to obtain a plurality of first real-time feature values ​​of different categories;

[0009] Calculate the first failure probability of the corresponding robotic arm according to the first real-time characteristic value of the same category;

[0010] Calculating a total probability of a first failure of the robotic arm based on first failure probabilities of the robotic arm corresponding to a plurality of first real-time characteristic values ​​of different categories;

[0011] Fault discrimination in the first target discrimination area is achieved according to the total probability of the first fault of the robotic arm.

[0012] The robot arm fault judgment method obtains multiple first real-time data information of the first target judgment area, classifies the first real-time feature values, obtains multiple first real-time feature values ​​of different categories, and then calculates the corresponding robot arm first fault probability based on the first real-time feature values ​​of the same category. Finally, based on the first failure probabilities of the robot arm corresponding to the multiple first real-time feature values ​​of different categories, the total probability of the first failure of the robot arm is calculated. The fault judgment of the first target judgment area is realized based on the obtained total probability of the first failure of the robot arm. It combines the first failure probabilities of the robot arm corresponding to the first real-time feature values ​​of different categories, comprehensively considers the influence of different types of feature values ​​on the robot arm fault judgment, and can accurately realize the judgment of the robot arm fault.

[0013] Preferably, the robot arm fault determination method further includes the following steps:

[0014] Acquire a plurality of second real-time data information of a second target discrimination area of ​​the robotic arm;

[0015] extracting a plurality of second real-time feature values ​​of the second real-time data information, and classifying the second real-time feature values ​​to obtain a plurality of second real-time feature values ​​of different categories;

[0016] Calculating the corresponding second failure probability of the robotic arm according to the second real-time characteristic value of the same category;

[0017] Obtaining a first interference factor of the first target discrimination area on the second target discrimination area and a second interference factor of the external environment on the second target discrimination area;

[0018] Calculating a total probability of a second failure of the robotic arm based on the second failure probabilities, the first interference factor, and the second interference factor corresponding to the second real-time characteristic values ​​of the plurality of different categories;

[0019] Fault discrimination in the second target discrimination area is achieved according to the second total fault probability of the robotic arm.

[0020] Preferably, the specific method for obtaining the first interference factor includes the following steps:

[0021] For the selected first real-time feature value and the second real-time feature value of the same category, obtaining a true value and a measured value of the second real-time feature value while excluding the influence of an external environment and the first real-time feature value of other categories on the second real-time feature value;

[0022] Calculating a first interference function of the first real-time eigenvalue to the second real-time eigenvalue by obtaining true values ​​and measured values ​​of the second real-time eigenvalue corresponding to a plurality of different first real-time eigenvalues;

[0023] A first interference factor is obtained according to the first interference function and the collected first real-time characteristic value.

[0024] Preferably, the specific method for obtaining the second interference factor includes the following steps:

[0025] For the selected external environmental factors of the same category and the second real-time characteristic value, obtaining a true value and a measured value of the second real-time characteristic value while excluding the influence of the first real-time characteristic value and the external environmental factors of other categories on the second real-time characteristic value;

[0026] By obtaining the true values ​​and measured values ​​of the second real-time characteristic values ​​corresponding to a plurality of different external environmental factors, a second interference function of the selected external environmental factors of the same category on the second real-time characteristic values ​​is calculated overall;

[0027] A second interference factor is obtained according to the second interference function and the collected real-time external environmental factor value.

[0028] Preferably, the first real-time data information includes noise information, vibration information, wear information, torque information and current information of the first target discrimination area.

[0029] Preferably, the specific method for calculating the first failure probability of the corresponding robotic arm according to the first real-time characteristic value of the same category includes the following steps:

[0030] Acquire first time information corresponding to the first real-time feature value of the same category;

[0031] Constructing a first real-time characteristic value curve function according to the first real-time characteristic value of the same category and the first time information;

[0032] Get the preset time period The actual first real-time characteristic value of the same category within the preset time period is obtained based on the first real-time characteristic curve function a first overall deviation between the predicted first real-time feature values ​​of the same category within the same category;

[0033] Calculating a first actual fault probability according to the first real-time characteristic value of the same category;

[0034] Acquire a preset time period based on the first real-time characteristic curve function a first predicted failure probability within the range of 0 to 100, and correcting the first predicted failure probability by using the first overall deviation;

[0035] Calculating a first failure probability of the robotic arm according to the first actual failure probability and the corrected first predicted failure probability;

[0036] in, Indicates the current moment, Indicates the preset time width.

[0037] Preferably, the specific method for calculating the corresponding second failure probability of the robotic arm according to the second real-time characteristic value of the same category includes the following steps:

[0038] Acquire second time information corresponding to the second real-time feature value of the same category;

[0039] Constructing a second real-time characteristic value curve function according to the second real-time characteristic value of the same category and the second time information;

[0040] Get the preset time period The actual second real-time characteristic value of the same category within the preset time period is obtained based on the second real-time characteristic curve function a second overall deviation between predicted second real-time feature values ​​of the same category within the same category;

[0041] Calculating a second actual fault probability according to the second real-time characteristic value of the same category;

[0042] Acquire a preset time period based on the second real-time characteristic curve function a second predicted failure probability within the range of 0 to 1, and correcting the second predicted failure probability by using the second overall deviation;

[0043] Calculating a second failure probability of the robotic arm according to the second actual failure probability and the corrected second predicted failure probability;

[0044] in, Indicates the current moment, Indicates the preset time width.

[0045] The present invention also provides a robot arm fault identification device, which includes:

[0046] An information acquisition module, configured to acquire a plurality of first real-time data information of a first target discrimination area of ​​the robotic arm;

[0047] a feature extraction module, configured to extract a plurality of first real-time feature values ​​of the first real-time data information, and classify the first real-time feature values ​​to obtain a plurality of first real-time feature values ​​of different categories;

[0048] a calculation module, configured to calculate the corresponding first failure probability of the robotic arm based on the first real-time eigenvalues ​​of the same category, and calculate the total first failure probability of the robotic arm based on the first failure probabilities of the robotic arm corresponding to the first real-time eigenvalues ​​of multiple different categories;

[0049] A discrimination module is used to realize fault discrimination of the first target discrimination area according to the total probability of the first fault of the robotic arm.

[0050] The present invention also provides a robot arm fault determination device, which includes:

[0051] Controller;

[0052] a memory storing executable instructions;

[0053] The executable instructions can be run on the controller and implement the robotic arm fault determination method.

[0054] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program implements the robotic arm fault determination method when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The present invention can be further understood from the following description in conjunction with the accompanying drawings. The components in the figures are not necessarily drawn to scale, but rather the emphasis is placed on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0056] Figure 1 This is a schematic diagram of the overall process of a method for determining a fault of a robotic arm according to an embodiment of the present invention;

[0057] Figure 2 This is a partial flow chart of a method for determining a robot arm fault in one embodiment of the present invention;

[0058] Figure 3 is a flowchart of a specific method for calculating the first failure probability of a corresponding robotic arm in one embodiment of the present invention;

[0059] Figure 4 is a flowchart of a specific method for obtaining a first interference factor in one embodiment of the present invention;

[0060] Figure 5 is a flowchart of a specific method for calculating the corresponding second failure probability of a robotic arm in one embodiment of the present invention;

[0061] Figure 6 It is a schematic diagram of the overall structure of a robot arm fault judgment device in one embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with its embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0063] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly attached to the other element or there may be an intermediate element. When an element is referred to as being "connected to" another element, it may be directly connected to the other element or there may be an intermediate element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only implementation methods.

[0064] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0065] The "first" and "second" in the present invention do not represent specific quantities and orders, but are only used to distinguish names.

[0066] like Figure 1 As shown, a method for determining a robot arm fault in one embodiment of the present invention includes the following steps:

[0067] S1, obtaining a plurality of first real-time data information of a first target discrimination area of ​​the robotic arm. The first target discrimination area may be any joint shaft portion. Preferably, the first real-time data information includes noise information, vibration information, wear information, torque information, and current information of the first target discrimination area.

[0068] S2: extracting a plurality of first real-time feature values ​​of the first real-time data information, and classifying the first real-time feature values ​​to obtain a plurality of first real-time feature values ​​of different categories.

[0069] The plurality of first real-time data information includes different categories. After extracting feature values ​​from the first real-time data information, the first real-time feature values ​​are classified according to properties and / or sources of the first real-time feature values.

[0070] Here, the plurality of first real-time data information may be classified first, and then feature value extraction may be performed on the classified first real-time data information to obtain a plurality of first real-time feature values ​​of different categories.

[0071] S3, calculating the first failure probability of the corresponding robotic arm according to the first real-time characteristic value of the same category.

[0072] For the first real-time feature value of the same category, the trained fault neural network model is used for identification to output the corresponding first fault probability of the robotic arm.

[0073] The fault neural network model is trained using a large number of first real-time eigenvalue data sets of different categories collected in advance. The first real-time eigenvalue data sets include first real-time eigenvalues ​​and corresponding fault probabilities. That is, the first real-time eigenvalues ​​of the same category serve as the input of the fault neural network model, and the fault probability calibrated with the first real-time eigenvalues ​​of the same category serves as the output of the fault neural network model.

[0074] For the first failure probability of the robotic arm corresponding to the first real-time eigenvalue of the same category, the first failure probability eigenvalues ​​of the robotic arm of the same category can also be collected in advance, and the corresponding first failure probability of the robotic arm can be calculated by comparing the similarity between the first real-time eigenvalue of the same category and the first failure probability eigenvalue of the robotic arm.

[0075] Preferably, in step S3, as Figure 3 As shown, the specific method for calculating the first failure probability of the corresponding robotic arm according to the first real-time characteristic value of the same category includes the following steps:

[0076] S30: Acquire first time information corresponding to the first real-time feature value of the same category.

[0077] S31 : Constructing a first real-time feature value curve function according to the first real-time feature values ​​of the same category and the first time information.

[0078] Specifically, the first real-time characteristic value curve function includes but is not limited to a linear function, a piecewise function, and a quadratic function. The first real-time characteristic value curve function can be constructed with the first time information as the independent variable and the first real-time characteristic value as the dependent variable, and then the first real-time characteristic value curve function is fitted according to the plurality of first real-time characteristic values ​​of the same category and the first time information to determine the relevant parameters of the first real-time characteristic value curve function, thereby constructing the first real-time characteristic value curve function. For example, it is assumed that the first real-time characteristic value curve function is a linear function and is expressed as ,in, Respectively represent the first real-time feature value and the first time information, Respectively represent the proportional coefficient and the preset constant term. Based on the least square method, the first real-time feature values ​​of the same category and the first time information are used to obtain the Fitting is performed to determine the proportional coefficient and the preset constant term, thereby constructing a first real-time eigenvalue curve function.

[0079] S32, obtaining a preset time period The actual first real-time characteristic value of the same category within the preset time period is obtained based on the first real-time characteristic curve function The first overall deviation between the predicted first real feature values ​​within the same category.

[0080] S33: Calculate a first actual fault probability according to the first real-time characteristic value of the same category.

[0081] Specifically, the first actual failure probability may be calculated by the similarity between the first real-time feature value of the same category and the corresponding first failure probability feature value of the robotic arm.

[0082] S34, obtaining a preset time period based on the first real-time characteristic curve function A first predicted failure probability within the range is calculated, and the first predicted failure probability is corrected by the first overall deviation.

[0083] Specifically, the preset time period can be obtained based on the first real-time characteristic curve function. The first real-time feature prediction value is obtained, and a first predicted failure probability is obtained based on the first real-time feature prediction value. For example, the corresponding first predicted failure probability can be calculated by comparing the similarity between the first real-time feature prediction value of the same category and the first failure probability feature value of the robotic arm.

[0084] S35: Calculate a first failure probability of the robotic arm according to the first actual failure probability and the corrected first predicted failure probability.

[0085] in, Indicates the current moment, Indicates the preset time width.

[0086] The first overall deviation can be obtained by calculating the average of the deviations between multiple actual first real-time characteristic values ​​and the predicted first real-time characteristic values. Correcting the first predicted failure probability using the first overall deviation is specifically as follows: first predicted failure probability = first predicted failure probability × (1 + first overall deviation).

[0087] The first failure probability of the robotic arm=(first actual failure probability+first predicted failure probability) / 2.

[0088] By calculating the first overall deviation and correcting the first predicted failure probability based on the first overall deviation, the first failure probability of the robotic arm is calculated according to the first actual failure probability and the corrected first predicted failure probability, taking into account the first actual failure probability at the current moment and the next moment. The first predicted failure probability can, to a certain extent, improve the accuracy of the first failure probability of the robotic arm.

[0089] S4, calculating the total probability of the first failure of the robotic arm according to the first failure probabilities of the robotic arm corresponding to the first real-time characteristic values ​​of multiple different categories.

[0090] S5: Implement fault discrimination in the first target discrimination area according to the total probability of the first fault of the robotic arm.

[0091] After obtaining the first failure rate of the robotic arm, the first failure probabilities of the robotic arm corresponding to multiple first real-time feature values ​​of different categories are combined to obtain the total probability of the first failure of the robotic arm, and the total probability of the first failure of the robotic arm is compared with the preset first failure probability threshold to determine whether a fault occurs in the first target judgment area.

[0092] Exemplarily, the total probability of the first failure of the robotic arm is equal to the average value or weighted average value of the first failure probabilities of the robotic arm corresponding to the first real-time characteristic values ​​of multiple different categories.

[0093] The robot arm fault judgment method obtains multiple first real-time data information of the first target judgment area, classifies the first real-time feature values, obtains multiple first real-time feature values ​​of different categories, and then calculates the corresponding robot arm first fault probability based on the first real-time feature values ​​of the same category. Finally, based on the first failure probabilities of the robot arm corresponding to the multiple first real-time feature values ​​of different categories, the total probability of the first failure of the robot arm is calculated. The fault judgment of the first target judgment area is realized based on the obtained total probability of the first failure of the robot arm. It combines the first failure probabilities of the robot arm corresponding to the first real-time feature values ​​of different categories, comprehensively considers the influence of different types of feature values ​​on the robot arm fault judgment, and can accurately realize the judgment of the robot arm fault.

[0094] In one embodiment, Figure 2 As shown, the robot arm fault determination method further includes the following steps:

[0095] S6: Acquire multiple second real-time data information of a second target discrimination area of ​​the robotic arm. The second target discrimination area may be any joint axis. Preferably, the second real-time data information includes noise information, vibration information, wear information, torque information, and current information of the second target discrimination area. The second target discrimination area is different from the first target discrimination area.

[0096] S7: Extracting multiple second real-time feature values ​​from the second real-time data information and classifying the second real-time feature values ​​to obtain multiple second real-time feature values ​​of different categories. The multiple second real-time data information may include different categories. After extracting feature values ​​from the second real-time data information, classify the second real-time feature values ​​based on their properties and / or sources.

[0097] S8, calculating the corresponding second failure probability of the robotic arm according to the second real-time characteristic value of the same category.

[0098] By comparing the similarity between the second real-time eigenvalue of the same category and the second fault probability eigenvalue of the robotic arm, the corresponding second fault probability of the robotic arm is calculated, or the second fault probability of the robotic arm corresponding to the second real-time eigenvalue of the same category is identified through a pre-trained fault neural network model.

[0099] S9: Obtain a first interference factor of the first target discrimination area on the second target discrimination area and a second interference factor of the external environment on the second target discrimination area.

[0100] like Figure 4 As shown, the specific method for obtaining the first interference factor of the first target discrimination area to the second target discrimination area includes the following steps:

[0101] S90: Obtain a first interference function of the first real-time feature value with respect to the second real-time feature value of each category.

[0102] Through the method of multiple experiments, for the first real-time eigenvalue and the second real-time eigenvalue of the same selected category, the true value and the measured value of the second real-time eigenvalue are obtained while excluding the influence of the external environment and the first real-time eigenvalue of other categories on the second real-time eigenvalue. By obtaining the true value and the measured value of the second real-time eigenvalue corresponding to multiple different first real-time eigenvalues, the first interference function of the first real-time eigenvalue on the second real-time eigenvalue is calculated as a whole.

[0103] The true value of the second real-time characteristic value here can be understood as the value of the second real-time characteristic value under ideal conditions and without interference, that is, the theoretical reference value obtained after excluding the influence of the target interference source (that is, the first real-time characteristic value) and other external factors. The measured value can be understood as the value of the second real-time characteristic value actually observed, at which time the first real-time characteristic value will interfere with it, that is, the second real-time characteristic value data actually observed after the introduction of the target interference source (here refers to the first real-time characteristic value of the specified category). The specific method of excluding the influence of the external environment and other categories of the first real-time characteristic value on the second real-time characteristic value includes:

[0104] 1. Implement environmental isolation. For example, build a closed experimental environment, using shielding covers, electromagnetic isolation chambers, or vacuum chambers to eliminate external interference such as temperature, humidity, and electromagnetic fields. Physical blocking devices (such as optical partitions and signal filters) can also be used to cut off the transmission path from the first real-time eigenvalue to the second real-time eigenvalue.

[0105] 2. Control relevant variables. For example, fix irrelevant variables and set the first real-time eigenvalues ​​of other categories to constant values ​​(such as zero or maintain the lowest energy state).

[0106] Assume that the true value of the second real-time feature value is , under the influence of the first real-time characteristic value, the measurement value of the second real-time characteristic value is , then the first real-time feature value of the same category The first interference coefficient for the second real-time characteristic value , for multiple different first real-time feature values ​​of the same category The first interference coefficient for the second real-time characteristic value And perform fitting to obtain the first interference function.

[0107] in, Indicates the first real-time eigenvalue under the influence of the first The measurement value of the second real-time characteristic value, Indicates the first real-time eigenvalue under the influence of the first The true value of the second real-time eigenvalue, Indicates the first real-time eigenvalue under the influence of the first An interference coefficient.

[0108] S91: Obtain a second interference function of the external environment on the second eigenvalue.

[0109] Through the method of multiple experiments, for the selected external environmental factors of the same category and the second real-time characteristic value, the true value and measured value of the second real-time characteristic value are obtained while excluding the influence of the first real-time characteristic value and the external environmental factors of other categories on the second real-time characteristic value. By obtaining the true value and measured value of the second real-time characteristic value corresponding to multiple different external environmental factors, the second interference function of the selected external environmental factors of the same category on the second real-time characteristic value is calculated as a whole.

[0110] For example, assuming that there is no influence of the first real-time feature value and other types of external environmental factors on the second real-time feature value, the true value of the second real-time feature value is , under the influence of the selected external environmental factors of the same category, the measured value of the second real-time characteristic value is , then the selected external environmental factors of the same category The second interference coefficient for the second real-time characteristic value , for multiple different first real-time feature values ​​of the same category Interference coefficient for the second real-time eigenvalue And perform fitting to obtain the second interference function.

[0111] in, Indicates the influence of external environmental factors The measurement value of the second real-time characteristic value, Indicates that under the influence of external environmental factors The true value of the second real-time eigenvalue, Indicates that under the influence of external environmental factors A second interference coefficient.

[0112] External environmental factors include but are not limited to ambient temperature, humidity, and ambient noise.

[0113] S92. After obtaining the first interference function and the second interference function, based on the collected first real-time feature value and real-time external environmental factor value, obtain the first interference factor of the first target discrimination area to the second target discrimination area according to multiple first interference coefficients, and obtain the second interference factor of the external environment to the second target discrimination area according to multiple second interference coefficients.

[0114] Here, the first interference factor may be an average value or a weighted average value of a plurality of first interference coefficients, and the second interference factor may be an average value or a weighted average value of a plurality of second interference coefficients.

[0115] S10 , calculating a total second failure probability of the robotic arm according to the second failure probabilities, the first interference factor, and the second interference factor corresponding to the second real-time characteristic values ​​of different categories.

[0116] As a preferred technical solution, in step S10, based on the second fault probabilities, the first interference factors, and the second interference factors corresponding to the second real-time characteristic values ​​of multiple different categories, the specific formula for calculating the total probability of the second fault of the robotic arm is: .

[0117] in, represents the total probability of the second failure of the robot arm, Indicates the The second failure probability of the robot arm corresponding to the second real-time eigenvalue of each category, represents the number of the second real-time feature value categories, Indicates the The second conditional function of the second failure probability of the robot arm corresponding to the second real-time eigenvalue of the category, Indicates the The adjustment coefficient of the second failure probability of the robot arm corresponding to the second real-time eigenvalue of each category, represents the first interference factor, represents the second interference factor, Indicates the The first interference factor compensation function corresponding to the second real-time eigenvalue of each category, Indicates the A second interference factor compensation function corresponding to the second real-time characteristic value of each category.

[0118] Here, the The adjustment coefficient of the second failure probability of the robot arm corresponding to the second real-time eigenvalue of each category is mainly used to correct The dimension, distribution deviation, or highlighting the influence of the feature, the second condition function is mainly used to Map it to a reasonable range, such as limiting the probability value to [0,1], or introduce nonlinear correction to make basic corrections to the failure probability of a single feature and eliminate the deviation of the feature itself.

[0119] Due to external factors such as environmental noise and load mutation, the failure probability may deviate from the theoretical value, so the first interference factor and the second interference factor are introduced. Since the first interference factor and the second interference factor may interfere from different dimensions (such as temperature and load), the average value of the compensation function of the two is taken. , comprehensively correct the influence of interference.

[0120] Total probability of the second failure of the robotic arm It mainly integrates the contributions of multi-category features through summation and averaging to avoid the limitations of a single feature.

[0121] The purpose of the first interference factor compensation function and the second interference factor compensation function is to more accurately calculate the influence of the first interference factor and the second interference factor on the second failure probability of the robotic arm, and use the first interference factor and the second interference factor to correct and compensate the second failure probability of the robotic arm to improve the accuracy of the second failure probability of the robotic arm. The first interference factor compensation function and the second interference factor compensation function. Preferably, when the first interference factor is the average or weighted average of multiple first interference coefficients, and the second interference factor is the average or weighted average of multiple second interference coefficients, , Of course, for different calculation methods of the first interference factor and the second interference factor, the corresponding first interference factor compensation function and the second interference factor compensation function are also different.

[0122] No. The second conditional function of the second failure probability of the robot arm corresponding to the second real-time eigenvalue of the category can be expressed as .in, For the The second preset probability threshold of the second failure probability of the manipulator arm corresponding to the second real-time eigenvalue of each category. By setting the second conditional function and the second preset probability threshold, the second failure probability of the manipulator arm that is less than the second preset probability threshold can be discarded. When the second failure probability of the manipulator arm is less than the second preset probability threshold, it means to some extent that the second failure probability of the manipulator arm corresponding to the second real-time eigenvalue is low and can be ignored. Therefore, by setting the second conditional function and the second preset probability threshold, the fault discrimination of the second target discrimination area through the total probability of the second failure of the manipulator arm can be made more practical, thereby improving the accuracy of fault discrimination.

[0123] No. The adjustment coefficient of the second failure probability of the robotic arm corresponding to the second real-time characteristic value of each category can be adjusted according to actual conditions, which will not be elaborated here.

[0124] S11 , performing fault discrimination in the second target discrimination area according to the second total fault probability of the robotic arm.

[0125] In some cases, such as when the second target discrimination area is adjacent to the first target discrimination area, the first real-time data information of the first target discrimination area may interfere with the second real-time data information. One example is when the first target discrimination area fails, such as due to wear or insufficient lubrication, resulting in abnormal noise, torque, or vibration, which may affect the collected second real-time data information.

[0126] Changes in the external environment may sometimes interfere with the collection of the second real-time data information. For example, excessive environmental noise may cause the collected noise data information to be too large, or the generation of a current magnetic field may cause the collected current data information to be distorted.

[0127] By obtaining a first interference factor of the first target discrimination area on the second target discrimination area and a second interference factor of the external environment on the second target discrimination area, and taking into account the interference factors of the first target discrimination area and the external environment on the second real-time data information of the second target discrimination area, the accuracy of fault discrimination in the second target discrimination area can be improved.

[0128] In one embodiment, the specific formula for calculating the total probability of the first failure of the robotic arm based on the first failure probabilities of the robotic arm corresponding to the first real-time feature values ​​of multiple different categories is: .

[0129] in, represents the total probability of the first failure of the robot arm, Indicates the The first failure probability of the robotic arm corresponding to the first real-time eigenvalue of each category, represents the number of first real-time feature value categories, Indicates the The first conditional function of the first failure probability of the robotic arm corresponding to the first real-time eigenvalue of each category, Indicates the The adjustment coefficient of the first failure probability of the robotic arm corresponding to the first real-time characteristic value of each category can be set and adjusted according to actual conditions and will not be repeated here.

[0130] For the The first conditional function of the first failure probability of the robot arm corresponding to the first real-time eigenvalue of each category can be expressed as .in, No. By setting the first conditional function and the first preset probability threshold, the first failure probability of the robot arm that is less than the first preset probability threshold can be discarded.

[0131] Specifically, the first real-time characteristic value of each category is first adjusted by the corresponding adjustment coefficient, and then filtered by the conditional function. If it is less than the corresponding first preset probability threshold, it is retained; otherwise, it is set to zero to filter out the lower first failure probability of the robotic arm.

[0132] Adjustment coefficient Reflect the priority of different features. For example, vibration abnormality is more likely to reflect the failure of the robot arm than temperature abnormality. The corresponding adjustment coefficient The value is larger to amplify its impact. The first preset probability threshold is used to achieve noise filtering. If the failure probability of a feature is extremely low, even after adjustment, it is still small, indicating that its contribution to the current failure can be ignored and directly excluded. The total probability of the first failure of the robot arm By fusing multi-feature results (sum + average), the impact of differences in the number of features can be eliminated, making the results more universal.

[0133] When the robotic arm's first failure probability is less than the first preset probability threshold, this means, to some extent, that the robotic arm's first failure probability corresponding to the first real-time characteristic value is low and can be ignored. Therefore, by setting the first conditional function and the first preset probability threshold, fault determination for the first target identification area based on the robotic arm's first failure total probability can be made more realistic, thereby improving fault determination accuracy.

[0134] In one embodiment, in step S8, as Figure 5 As shown, the specific method for calculating the corresponding second failure probability of the robotic arm according to the second real-time characteristic value of the same category includes the following steps:

[0135] S80: Acquire second time information corresponding to the second real-time feature value of the same category.

[0136] S81: Construct a second real-time feature value curve function according to the second real-time feature value of the same category and the second time information.

[0137] Specifically, the second real-time characteristic value curve function includes but is not limited to a linear function, a piecewise function, and a quadratic function. The second real-time characteristic value curve function can be constructed with the second time information as the independent variable and the second real-time characteristic value as the dependent variable, and then the second real-time characteristic value curve function is fitted according to the plurality of second real-time characteristic values ​​of the same category and the second time information to determine the relevant parameters of the second real-time characteristic value curve function and construct the second real-time characteristic value curve function. For example, it is assumed that the second real-time characteristic value curve function is a linear function and is expressed as ,in, Respectively represent the second real-time characteristic value and the second time information, Respectively represent the proportional coefficient and the preset constant term. Based on the least square method, the second real-time feature values ​​of the same category and the second time information can be used to obtain the Perform fitting to determine the proportional coefficient and preset constant term, thereby constructing a second real-time eigenvalue curve function.

[0138] S82, obtaining a preset time period The actual second real-time characteristic value of the same category within the preset time period is obtained based on the second real-time characteristic curve function The second overall deviation between the predicted second real feature values ​​within the same category.

[0139] S83: Calculate a second actual fault probability according to the second real-time characteristic value of the same category.

[0140] Specifically, the second actual failure probability may be calculated by the similarity between the second real-time feature value of the same category and the corresponding second failure probability feature value of the robotic arm.

[0141] S84, obtaining a preset time period based on the second real-time characteristic curve function A second predicted failure probability within the range is calculated, and the second predicted failure probability is corrected by the second overall deviation.

[0142] S85: Calculate the second failure probability of the robotic arm based on the second actual failure probability and the corrected second predicted failure probability. Indicates the current moment, Indicates the preset time width.

[0143] Specifically, the preset time period can be obtained based on the second real-time characteristic curve function. The second real-time feature prediction value is used to obtain a second predicted failure probability based on the second real-time feature prediction value. For example, the corresponding second predicted failure probability can be calculated by comparing the similarity between the second real-time feature prediction value of the same category and the second failure probability feature value of the robotic arm.

[0144] The second overall deviation can be obtained by calculating the average of the deviations between multiple actual second real-time characteristic values ​​and the predicted second real-time characteristic values. Correcting the second predicted failure probability using the second overall deviation is specifically as follows: second predicted failure probability = second predicted failure probability * (1 + second overall deviation).

[0145] The second failure probability of the robotic arm=(second actual failure probability+second predicted failure probability) / 2.

[0146] By calculating the second overall deviation and correcting the second predicted failure probability based on the second overall deviation, the second failure probability of the robot arm is calculated according to the second actual failure probability and the corrected second predicted failure probability, taking into account the second actual failure probability at the current moment and the next moment. The second predicted failure probability can, to a certain extent, improve the accuracy of the second failure probability of the robotic arm.

[0147] The present invention also provides a robot arm fault identification device, such as Figure 6 As shown, it includes an information acquisition module, a feature extraction module, a calculation module and a discrimination module.

[0148] The information acquisition module is used to obtain multiple first real-time data information of the first target discrimination area of ​​the robotic arm; the feature extraction module is used to extract the first real-time feature values ​​of the multiple first real-time data information, and classify the first real-time feature values ​​to obtain multiple first real-time feature values ​​of different categories.

[0149] The calculation module is used to calculate the corresponding first failure probability of the robotic arm based on the first real-time eigenvalue of the same category, and calculate the total probability of the first failure of the robotic arm based on the first failure probabilities of the robotic arm corresponding to the first real-time eigenvalues ​​of multiple different categories; the judgment module is used to realize the fault judgment of the first target judgment area based on the total probability of the first failure of the robotic arm.

[0150] The robot arm fault judgment device obtains multiple first real-time data information of the first target judgment area, classifies the first real-time feature values, obtains multiple first real-time feature values ​​of different categories, and then calculates the corresponding robot arm first fault probability based on the first real-time feature values ​​of the same category. Finally, based on the first failure probabilities of the robot arm corresponding to the multiple first real-time feature values ​​of different categories, the total probability of the first failure of the robot arm is calculated. The fault judgment of the first target judgment area is realized based on the obtained total probability of the first failure of the robot arm. It combines the first failure probabilities of the robot arm corresponding to the first real-time feature values ​​of different categories, comprehensively considers the influence of different types of feature values ​​on the robot arm fault judgment, and can accurately realize the judgment of the robot arm fault.

[0151] The present invention also provides a robot arm fault determination device, which includes: a controller; and a memory storing executable instructions.

[0152] The executable instructions can be run on the controller and implement the robotic arm fault determination method.

[0153] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program implements the robotic arm fault determination method when executed by a processor.

[0154] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0155] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be determined by the appended claims.

Claims

1. A method for determining a robot arm fault, characterized in that: The robot arm fault determination method comprises the following steps: Acquire a plurality of first real-time data information of a first target discrimination area of ​​the robotic arm; extracting a plurality of first real-time feature values ​​of the first real-time data information, and classifying the first real-time feature values ​​to obtain a plurality of first real-time feature values ​​of different categories; Calculating the first failure probability of the corresponding robotic arm according to the first real-time characteristic value of the same category; Calculating a total probability of a first failure of the robotic arm based on first failure probabilities of the robotic arm corresponding to a plurality of first real-time characteristic values ​​of different categories; Implementing fault discrimination in the first target discrimination area according to the total probability of the first fault of the robotic arm; The specific method for calculating the corresponding first failure probability of the robotic arm includes the following steps: Acquire first time information corresponding to the first real-time feature value of the same category; Constructing a first real-time characteristic value curve function according to the first real-time characteristic value of the same category and the first time information; Get the preset time period The actual first real-time characteristic value of the same category within the preset time period is obtained based on the first real-time characteristic curve function a first overall deviation between the predicted first real-time feature values ​​of the same category within the same category; Calculating a first actual fault probability according to the first real-time characteristic value of the same category; Acquire a preset time period based on the first real-time characteristic curve function a first predicted failure probability within the range of 0 to 100, and correcting the first predicted failure probability by using the first overall deviation; Calculating a first failure probability of the robotic arm according to the first actual failure probability and the corrected first predicted failure probability; in, Indicates the current moment, Represents a preset time width, the first overall deviation is obtained by calculating the average value of the deviation between multiple actual first real-time feature values ​​and predicted first real-time feature values; the first actual failure probability is calculated by the similarity between the first real-time feature value of the same category and the corresponding first failure probability feature value of the robotic arm; the preset time period is obtained based on the first real-time feature curve function A first real-time feature prediction value is obtained, and a first predicted failure probability is obtained according to the first real-time feature prediction value.

2. The method for determining a robot arm fault according to claim 1, wherein: The robot arm fault determination method further comprises the following steps: Acquire a plurality of second real-time data information of a second target discrimination area of ​​the robotic arm; extracting a plurality of second real-time feature values ​​of the second real-time data information, and classifying the second real-time feature values ​​to obtain a plurality of second real-time feature values ​​of different categories; Calculating the corresponding second failure probability of the robotic arm according to the second real-time characteristic value of the same category; Obtaining a first interference factor of the first target discrimination area on the second target discrimination area and a second interference factor of the external environment on the second target discrimination area; Calculating a total probability of a second failure of the robotic arm based on the first interference factor, the second interference factor, and the second failure probabilities corresponding to the second real-time characteristic values ​​of different categories; Fault discrimination in the second target discrimination area is achieved according to the second total fault probability of the robotic arm.

3. The method for determining a mechanical arm fault according to claim 2, wherein: The specific method for obtaining the first interference factor includes the following steps: For the selected first real-time feature value and the second real-time feature value of the same category, obtaining a true value and a measured value of the second real-time feature value while excluding the influence of an external environment and the first real-time feature value of other categories on the second real-time feature value; Calculating a first interference function of the first real-time eigenvalue to the second real-time eigenvalue by obtaining true values ​​and measured values ​​of the second real-time eigenvalue corresponding to a plurality of different first real-time eigenvalues; A first interference factor is obtained according to the first interference function and the collected first real-time characteristic value.

4. The method for determining a mechanical arm fault according to claim 3, wherein: The specific method for obtaining the second interference factor includes the following steps: For the selected external environmental factors of the same category and the second real-time characteristic value, obtaining a true value and a measured value of the second real-time characteristic value while excluding the influence of the first real-time characteristic value and the external environmental factors of other categories on the second real-time characteristic value; By obtaining the true values ​​and measured values ​​of the second real-time characteristic values ​​corresponding to a plurality of different external environmental factors, a second interference function of the selected external environmental factors of the same category on the second real-time characteristic values ​​is calculated overall; A second interference factor is obtained according to the second interference function and the collected real-time external environmental factor value.

5. The method for determining a robot arm fault according to claim 1, wherein: The first real-time data information includes noise information, vibration information, wear information, torque information and current information of the first target identification area.

6. The method for determining a robot arm fault according to claim 2, wherein: The specific method for calculating the corresponding second failure probability of the robotic arm according to the second real-time characteristic value of the same category includes the following steps: Acquire second time information corresponding to the second real-time feature value of the same category; Constructing a second real-time characteristic value curve function according to the second real-time characteristic value of the same category and the second time information; Get the preset time period The actual second real-time characteristic value of the same category within the preset time period is obtained based on the second real-time characteristic curve function a second overall deviation between predicted second real-time feature values ​​of the same category within the same category; Calculating a second actual fault probability according to the second real-time characteristic value of the same category; Acquire a preset time period based on the second real-time characteristic curve function a second predicted failure probability within the range of 0 to 1, and correcting the second predicted failure probability by using the second overall deviation; Calculating a second failure probability of the robotic arm according to the second actual failure probability and the corrected second predicted failure probability; in, Indicates the current moment, Indicates the preset time width.

7. A robot arm fault identification device, used to implement the robot arm fault identification method according to any one of claims 1 to 6, characterized in that: The mechanical arm fault determination device comprises: An information acquisition module, configured to acquire a plurality of first real-time data information of a first target discrimination area of ​​the robotic arm; a feature extraction module, configured to extract a plurality of first real-time feature values ​​of the first real-time data information, and classify the first real-time feature values ​​to obtain a plurality of first real-time feature values ​​of different categories; a calculation module, configured to calculate the corresponding first failure probability of the robotic arm based on the first real-time eigenvalues ​​of the same category, and calculate the total first failure probability of the robotic arm based on the first failure probabilities of the robotic arm corresponding to the first real-time eigenvalues ​​of multiple different categories; a discrimination module, configured to discriminate the fault of the first target discrimination area according to the total probability of the first fault of the robotic arm; The specific method for calculating the corresponding first failure probability of the robotic arm includes the following steps: Acquire first time information corresponding to the first real-time feature value of the same category; Constructing a first real-time characteristic value curve function according to the first real-time characteristic value of the same category and the first time information; Get the preset time period The actual first real-time characteristic value of the same category within the preset time period is obtained based on the first real-time characteristic curve function a first overall deviation between the predicted first real-time feature values ​​of the same category within the same category; Calculating a first actual fault probability according to the first real-time characteristic value of the same category; Acquire a preset time period based on the first real-time characteristic curve function a first predicted failure probability within the range of 0 to 100, and correcting the first predicted failure probability by using the first overall deviation; Calculating a first failure probability of the robotic arm according to the first actual failure probability and the corrected first predicted failure probability; in, Indicates the current moment, Represents a preset time width, the first overall deviation is obtained by calculating the average value of the deviation between multiple actual first real-time feature values ​​and predicted first real-time feature values; the first actual failure probability is calculated by the similarity between the first real-time feature value of the same category and the corresponding first failure probability feature value of the robotic arm; the preset time period is obtained based on the first real-time feature curve function A first real-time feature prediction value is obtained, and a first predicted failure probability is obtained according to the first real-time feature prediction value.

8. A robot arm fault identification device, characterized in that: The mechanical arm fault determination device comprises: Controller; a memory storing executable instructions; The executable instructions can be run on the controller and implement the robotic arm fault determination method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for determining a robot arm fault according to any one of claims 1 to 6 is implemented.

Images