Robotic arm fault detection method, device, equipment, storage medium and program product

Abstract

The present invention provides a method, device, equipment, storage medium and program product for detecting faults in a robotic arm, which relates to the field of intelligent manufacturing technology. The method comprises: obtaining input data to be detected; inputting the input data to be detected into a normal model and a fault model respectively, obtaining a first discrimination result output by the normal model and a second discrimination result output by the fault model; making a decision based on the first discrimination result and the second discrimination result by using a Bayesian decision method to obtain a fault detection result of the robotic arm. In the above manner, firstly, discrimination is performed respectively by using a normal model and a fault model to obtain two discrimination results, and then further decision is made on the two discrimination results by using a Bayesian decision method to obtain a final fault detection result. In the process of robotic arm fault detection, the Bayesian decision method is introduced to optimize the discrimination results of the normal model and the fault model, which can improve the accuracy and reliability of robotic arm fault detection.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing technology, and in particular to a robot arm fault detection method, device, equipment, storage medium and program product. Background Art

[0002] Robotic arm fault detection is a crucial aspect of industrial automation, particularly in applications requiring high reliability and safety. With technological advancements, robotic arms are becoming increasingly widespread, playing a vital role in diverse fields, including manufacturing, healthcare, aerospace, and the service industry. However, robotic arm failures can not only disrupt production but also potentially lead to safety incidents. Therefore, developing effective fault detection methods is crucial to ensuring the stable operation of robotic arms.

[0003] Currently, most robotic arm fault detection methods operate within a "passive" framework. These methods rely on the input and output data of the robotic arm system to generate a residual signal, which is then used to determine whether the robotic arm has failed. In this passive fault detection process, the detection unit does not modify the input data of the robotic arm system, and the diagnostic process does not actively interfere with or affect the natural evolution of the robotic arm system.

[0004] However, in real industrial environments, the noise and uncertainty of robotic arm systems can mask fault signals, making early fault detection difficult. In closed-loop systems, the controller may compensate for anomalies caused by the fault, further reducing the detectability of the fault signal. Consequently, existing robotic arm fault detection methods have low accuracy and reliability when the fault signature is unclear or similar to normal conditions. Summary of the Invention

[0005] The present invention provides a robot arm fault detection method, device, equipment, storage medium and program product, which are used to solve the defects of low accuracy and reliability of the robot arm fault detection method in the prior art.

[0006] The present invention provides a method for detecting faults of a robotic arm, comprising: obtaining input data to be detected; the input data to be detected is determined based on the motion trajectory of the robotic arm; inputting the input data to be detected into a normal model and a fault model respectively, obtaining a first discrimination result output by the normal model and a second discrimination result output by the fault model; making a decision based on the first discrimination result and the second discrimination result by using a Bayesian decision method, and obtaining a fault detection result of the robotic arm.

[0007] According to a robot arm fault detection method provided by the present invention, a decision is made through a Bayesian decision method based on a first discrimination result and a second discrimination result to obtain a robot arm fault detection result, including: correcting the first discrimination result based on a first Kalman filter to obtain a third discrimination result; correcting the second discrimination result based on a second Kalman filter to obtain a fourth discrimination result; and making a decision on the third discrimination result and the fourth discrimination result through a Bayesian decision method to obtain a robot arm fault detection result.

[0008] According to a robotic arm fault detection method provided by the present invention, obtaining input data to be detected includes: obtaining a preset motion trajectory of the robotic arm; performing trajectory optimization on the preset motion trajectory based on a differential evolution algorithm to obtain a motion trajectory; wherein the trajectory optimization process takes minimizing an optimization function as an optimization goal, and the optimization function is determined based on the Bayesian risk upper bound of a Bayesian decision method; and tracking and detecting the motion trajectory to obtain the input data to be detected.

[0009] According to a robot arm fault detection method provided by the present invention, the expression of the optimization function is:

[0010] ;

[0011] Among them, H0 is the normal model; H1 is the fault model; P(H0) is the prior probability of the normal model; P(H1) is the prior probability of the fault model; μ0 is the mean of the Gaussian distribution with uncertainty in the normal model; μ1 is the mean of the Gaussian distribution with uncertainty in the fault model; σ0 is the variance of the Gaussian distribution with uncertainty in the normal model; σ1 is the variance of the Gaussian distribution with uncertainty in the fault model.

[0012] According to a robotic arm fault detection method provided by the present invention, before obtaining the input data to be detected, it also includes: constructing a robotic arm model; the robotic arm model is a mathematical model used to describe the dynamic characteristics of the robotic arm; based on the robotic arm model, a normal model is constructed; the normal model is a mathematical model used to describe the dynamic characteristics of the robotic arm when no fault occurs; determining the fault coefficient of the robotic arm; based on the fault coefficient and the robotic arm model, a fault model is constructed; the fault model is a mathematical model used to describe the dynamic characteristics of the robotic arm when a fault occurs.

[0013] According to a robot arm fault detection method provided by the present invention, the expression of the normal model is:

[0014] ;

[0015] ;

[0016] Among them, x(k) represents the system state of the manipulator at time k; x(k-1) represents the system state of the manipulator at time k-1; u(k) represents the motor input of the manipulator at time k; y(k) represents the system output of the manipulator at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; the expression of the fault model is:

[0017] ;

[0018] ;

[0019] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; and Z is the fault coefficient.

[0020] The present invention also provides a robotic arm fault detection device, including: an acquisition module for acquiring input data to be detected; the input data to be detected is determined based on the motion trajectory of the robotic arm; a discrimination module for inputting the input data to be detected into a normal model and a fault model respectively, to obtain a first discrimination result output by the normal model and a second discrimination result output by the fault model; a detection module for making a decision based on the first discrimination result and the second discrimination result through a Bayesian decision method to obtain a fault detection result of the robotic arm.

[0021] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any one of the above-mentioned robot arm fault detection methods is implemented.

[0022] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned robot arm fault detection methods.

[0023] The present invention also provides a computer program product, comprising a computer program, which implements any of the above-mentioned robot arm fault detection methods when executed by a processor.

[0024] The robot arm fault detection method, device, equipment, storage medium and program product provided by the present invention obtain input data to be detected; the input data to be detected is determined based on the motion trajectory of the robot arm; the input data to be detected is input into the normal model and the fault model respectively to obtain a first discrimination result output by the normal model and a second discrimination result output by the fault model; based on the first discrimination result and the second discrimination result, a decision is made by the Bayesian decision method to obtain the fault detection result of the robot arm. In the above manner, first, the normal model and the fault model are used to perform discrimination respectively to obtain two discrimination results, and then the two discrimination results are further decided by the Bayesian decision method to obtain the final fault detection result. In the process of robot arm fault detection, the Bayesian decision method is introduced to optimize the discrimination results of the normal model and the fault model, which can improve the accuracy and reliability of robot arm fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0026] Figure 1 This is one of the flow charts of the robot arm fault detection method provided by the present invention.

[0027] Figure 2 This is the second flow chart of the robot arm fault detection method provided by the present invention.

[0028] Figure 3 It is a schematic diagram of the principle of Bayesian decision-making provided by the present invention.

[0029] Figure 4 It is a structural diagram of the robotic arm fault detection device provided in an embodiment of the present application.

[0030] Figure 5 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0032] See also Figure 1, Figure 1 This is one of the flow charts of the robot arm fault detection method provided by the present invention. In this embodiment, the robot arm fault detection method includes steps S110 to S130, each of which is as follows:

[0033] S110: Obtain input data to be detected.

[0034] The input data to be detected is determined based on the motion trajectory of the robotic arm.

[0035] Generally, the motion trajectory of the robotic arm can reflect the fault condition of the robotic arm: if the robotic arm has not failed, the robotic arm can move according to the preset motion trajectory (i.e., the preset reference trajectory); if the robotic arm has failed, the motion trajectory of the robotic arm will deviate from the preset motion trajectory.

[0036] Specifically, the motion trajectory of the robot arm is tracked and detected to obtain input data to be detected.

[0037] S120: Input the input data to be detected into the normal model and the fault model respectively, and obtain a first discrimination result output by the normal model and a second discrimination result output by the fault model.

[0038] In this embodiment, the normal model is a mathematical model used to describe the dynamic characteristics of the robot arm when no fault occurs; the fault model is a mathematical model used to describe the dynamic characteristics of the robot arm when a fault occurs.

[0039] Among them, the normal model and the fault model can be constructed based on the robotic arm model, and the robotic arm model is a mathematical model used to describe the dynamic characteristics of the robotic arm.

[0040] Specifically, based on the dynamic characteristics of the robotic arm, a robotic arm model is constructed. The expression of the robotic arm model is as follows:

[0041] ;

[0042] ;

[0043] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix.

[0044] Furthermore, different fault coefficients (ie, multiplicative fault coefficients) are set for the robot arm model to obtain a normal model and a fault model.

[0045] Specifically, determine the fault coefficient corresponding to the normal model and the fault coefficient corresponding to the fault model; determine the normal model based on the fault coefficient corresponding to the normal model and the robotic arm model; determine the fault model based on the fault coefficient corresponding to the fault model and the robotic arm model.

[0046] Generally, the fault coefficient value corresponding to the normal model is 1. At this time, the expressions of the robot arm model and the normal model are the same, that is, the expression of the normal model is as follows:

[0047] ;

[0048] ;

[0049] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix.

[0050] The value of the fault coefficient corresponding to the fault model can be set according to actual needs. The value range of the fault coefficient corresponding to the fault model is (0, 1). The expression of the fault model is as follows:

[0051] ;

[0052] ;

[0053] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; Z is the fault coefficient corresponding to the fault model.

[0054] Specifically, the input data to be detected is substituted into the expression of the normal model and the expression of the fault model respectively, so as to obtain the first discrimination result output by the normal model and the second discrimination result output by the fault model.

[0055] S130: Based on the first discrimination result and the second discrimination result, a decision is made by using a Bayesian decision method to obtain a fault detection result of the robotic arm.

[0056] The robot arm fault detection method provided in this embodiment obtains input data to be detected; the input data to be detected is determined based on the motion trajectory of the robot arm; the input data to be detected is input into the normal model and the fault model respectively to obtain a first discrimination result output by the normal model and a second discrimination result output by the fault model; based on the first discrimination result and the second discrimination result, a decision is made through the Bayesian decision method to obtain the fault detection result of the robot arm. In the above manner, first, the normal model and the fault model are used to perform discrimination respectively to obtain two discrimination results, and then the two discrimination results are further decided by the Bayesian decision method to obtain the final fault detection result. In the process of robot arm fault detection, the Bayesian decision method is introduced to optimize the discrimination results of the normal model and the fault model, which can improve the accuracy and reliability of robot arm fault detection.

[0057] In some embodiments, based on the first discrimination result and the second discrimination result, a decision is made through a Bayesian decision method to obtain a fault detection result of the robotic arm, including: correcting the first discrimination result based on the first Kalman filter to obtain a third discrimination result; correcting the second discrimination result based on the second Kalman filter to obtain a fourth discrimination result; and making a decision on the third discrimination result and the fourth discrimination result through a Bayesian decision method to obtain a fault detection result of the robotic arm.

[0058] See also Figure 2 , Figure 2 This is the second flow chart of the robot arm fault detection method provided by the present invention.

[0059] like Figure 2 As shown, after obtaining the input data to be detected, the input data to be detected is input into the normal model and the fault model respectively to obtain a first discrimination result output by the normal model and a second discrimination result output by the fault model.

[0060] Considering that the data directly output by the normal model and the fault model will be affected by noise, in order to further improve the accuracy of detection, the data directly output by the model can be optimized through the Kalman filter to improve the accuracy and stability of fault detection.

[0061] The Kalman filter is an effective recursive filter that uses the state equation and measurement equation of a linear system to estimate the system state. It is widely used in signal processing, navigation systems, robot positioning, financial forecasting, and other fields. The core idea of ​​the Kalman filter is to use the state estimate and uncertainty (covariance) to update the estimated value. It consists of two main steps: prediction and update.

[0062] The Kalman filter operates in two key stages: first, the prediction stage, which uses the state transition equation and process noise to predict the system's current state based on the state estimate at the previous moment. Second, the update stage, when new observation data arrives, compares the predicted state with the actual observation data, combines the observation equation and observation noise, and calculates an optimized posterior state estimate. This posterior estimate not only considers the error between the predicted and observed values ​​but also incorporates prior knowledge of the system and noise characteristics, thereby improving the accuracy and stability of the state estimate.

[0063] Specifically, the first discrimination result is corrected based on the first Kalman filter to obtain a third discrimination result.

[0064] Among them, the expression of the first Kalman filter (that is, the Kalman filter applicable to the normal model) is as follows:

[0065] (1) State estimation:

[0066] ;

[0067] (2) Error covariance prediction:

[0068] ;

[0069] (3) Calculate the Kalman gain:

[0070] ;

[0071] (4) Status update:

[0072] ;

[0073] (5) Error covariance update:

[0074] .

[0075] Furthermore, the second discrimination result is corrected based on the second Kalman filter to obtain a fourth discrimination result.

[0076] The expression of the second Kalman filter (i.e., the Kalman filter applicable to the fault model) is as follows:

[0077] (1) State estimation:

[0078] ;

[0079] (2) Error covariance prediction:

[0080] ;

[0081] (3) Calculate the Kalman gain:

[0082] ;

[0083] (4) Status update:

[0084] ;

[0085] (5) Error covariance update:

[0086] ;

[0087] in, is the state estimate of the robot arm at time k, is the optimal state estimate of the robotic arm at time k-1, is the optimal state estimate of the robot arm at time k, u(k) represents the motor input of the robot arm at time k, is the prior error covariance at time k, is the error covariance after the update at time k, is the updated error covariance at time k-1, y(k) is the system state measurement at time k, K(k) is the Kalman gain at time k, Q is the covariance of the process noise, R is the covariance of the measurement noise, A is the first coefficient matrix, B is the second coefficient matrix, C is the third coefficient matrix, and Z is the fault coefficient corresponding to the fault model.

[0088] It should be noted that in actual application scenarios, the Kalman filter may not be used. Instead, the output of the normal model and / or the fault model may be used directly to judge the fault condition of the robotic arm. However, considering that there may be a large amount of process noise and measurement noise in the working environment of the robotic arm, resulting in inaccurate model output, the Kalman filter is used for correction to provide the optimal solution for state estimation, namely, the minimum mean square error estimation.

[0089] Furthermore, a Bayesian decision method is used to make a decision on the third discrimination result and the fourth discrimination result to obtain a fault detection result of the robotic arm.

[0090] The Bayesian decision method, a statistical decision theory based on Bayes' theorem, is an effective decision-making tool in the face of uncertainty and risk. Its core lies in optimizing the decision-making process by leveraging prior knowledge and probability distributions. The Bayesian decision method first establishes a probabilistic model to describe the relationship between decision variables and their probability distributions. Subsequently, when new observations emerge, the prior probability is updated using Bayes' theorem to obtain the posterior probability distribution, allowing for a more accurate assessment of the likelihood and impact of various decisions.

[0091] During the decision-making process, the Bayesian decision method uses different decision rules based on the posterior probability distribution, such as the maximum a posteriori probability decision rule or the expected loss minimization decision rule, to select the optimal solution. This method not only effectively utilizes existing prior knowledge but also continuously optimizes the accuracy and effectiveness of decisions through repeated updates of observational data. Therefore, Bayesian decision making is widely used in machine learning, artificial intelligence, statistical inference, and various decision analysis fields, providing an in-depth and comprehensive solution to decision-making problems in complex and dynamic environments.

[0092] In this embodiment, the Bayesian decision method is used to consider the prior probability and the posterior probability, and a decision is made on the third discrimination result and the fourth discrimination result to obtain the fault detection result of the robotic arm, thereby reducing the false detection rate to a certain extent.

[0093] See also Figure 3 , Figure 3 It is a schematic diagram of the principle of Bayesian decision-making provided by the present invention.

[0094] In the field of fault detection, given a set of input data and output data as well as normal model and fault model, the decision principle of the Bayesian decision method is as follows: Figure 3 shown.

[0095] like Figure 2 and Figure 3 As shown, H0 represents the normal model, H1 represents the fault model, P(H i ) represents the prior probability of model i (i=0,1), Figure 3 The shaded area in (i.e., the overlapping area under the two curves) represents the probability of incorrect model selection (i.e., the Bayesian risk).

[0096] Specifically, the Bayesian decision method can produce two decision regions; the decision region of the normal model is recorded as , the decision region of the fault model is recorded as .

[0097] If the output y of the model falls within If the output y of the model falls within the range of If the output of the fault model falls within the region, the output of the fault model is selected. Fault detection is achieved by determining the decision region where the output of the model falls.

[0098] It should be noted that, in actual application, the fault detection result of the robotic arm can be directly judged by the first discrimination result output by the normal model and / or the second discrimination result output by the fault model, or it can be judged according to the third discrimination result output by the first Kalman filter and / or the fourth discrimination result output by the second Kalman filter, or it can be judged by the Bayesian decision method based on the third discrimination result and the fourth discrimination result.

[0099] In some embodiments, obtaining input data to be detected includes: obtaining a preset motion trajectory of the robotic arm; performing trajectory optimization on the preset motion trajectory based on a differential evolution algorithm to obtain a motion trajectory; wherein the trajectory optimization process takes minimizing an optimization function as an optimization goal, and the optimization function is determined based on the Bayesian risk upper bound of a Bayesian decision method; tracking and detecting the motion trajectory to obtain the input data to be detected.

[0100] Most existing robot arm fault detection methods are performed under a "passive" framework. In order to overcome the limitations of passive fault detection methods, this embodiment improves the existing detection strategy and adopts an active fault detection method.

[0101] Active fault detection methods enhance the detectability of fault signatures by designing specific input data (e.g., input signals) to detect. This is a more advanced fault detection strategy that proactively intervenes in the robot system's operation to enhance the detectability of fault signatures. Compared to traditional passive fault detection methods, the most important task of active fault detection methods is to design appropriate input data to detect. This requires that the designed input data not only improves fault detection performance but also achieves the robot's control objectives.

[0102] It is understandable that different input data will lead to different error rates in the Bayesian decision method. Therefore, in this embodiment, the control input is designed using the trajectory optimization method to reduce the Bayesian risk.

[0103] Bayesian risk The expression is as follows:

[0104] ;

[0105] In general, the above Bayesian risk expression is not suitable as an optimization metric because the integral in Bayesian risk cannot usually be evaluated in closed form.

[0106] Based on this, in this embodiment, Battacharyya Bound is used as the optimization indicator, and the upper bound of the Bayesian risk is given by Battacharyya Bound.

[0107] The expression for Battacharyya Bound is as follows:

[0108] ;

[0109] Assuming that the system uncertainty is Gaussian distribution, the mean of the Gaussian distribution of the normal model uncertainty is μ0, the variance of the Gaussian distribution of the normal model uncertainty is σ0, the mean of the Gaussian distribution of the fault model uncertainty is μ1, and the variance of the Gaussian distribution of the fault model uncertainty is σ1, then the above formula can be calculated:

[0110] ;

[0111] The expression on the right side of the equation can be understood as the upper bound of Bayesian risk. .

[0112] Since the logarithmic function is a monotonically increasing function, the logarithm of Battacharyya Bound can be taken as the optimization function. The expression of the optimization function is as follows:

[0113] ;

[0114] Among them, H0 is the normal model; H1 is the fault model; P(H0) is the prior probability of the normal model; P(H1) is the prior probability of the fault model; μ0 is the mean of the Gaussian distribution with uncertainty in the normal model; μ1 is the mean of the Gaussian distribution with uncertainty in the fault model; σ0 is the variance of the Gaussian distribution with uncertainty in the normal model; σ1 is the variance of the Gaussian distribution with uncertainty in the fault model.

[0115] Specifically, if Figure 2 As shown, the preset motion trajectory of the robotic arm is first obtained, and the preset motion trajectory is optimized based on the differential evolution algorithm to obtain the motion trajectory.

[0116] Among them, the trajectory optimization process takes minimizing the optimization function as the optimization goal, and the optimization function is determined by the Bayesian risk upper bound based on the Bayesian decision method.

[0117] Differential Evolution (DE) is a population-based direct search optimization algorithm that solves various optimization problems by simulating the principles of natural selection and genetic evolution. Due to its simplicity, strong robustness, easy implementation, and insensitivity to parameters, it has been widely used in engineering and scientific research.

[0118] The differential evolution algorithm consists of the following steps:

[0119] (1) Initialization: Randomly generate a set of initial solutions, each of which represents an individual in the population.

[0120] (2) Difference operation: select three different individuals and calculate their differences.

[0121] (3) Crossover operation: cross the difference with the fourth individual to generate a new individual.

[0122] (4) Selection operation: Compare the newly generated individuals with the original individuals, usually using a greedy strategy, that is, selecting individuals with good fitness.

[0123] (5) Termination condition: If the termination condition is met (such as reaching the maximum number of iterations or the quality of the solution reaches a preset threshold), the algorithm ends.

[0124] In this embodiment, the optimization goal of the differential evolution algorithm is to minimize the optimization function.

[0125] Furthermore, the motion trajectory is tracked and detected to obtain input data to be detected.

[0126] In some embodiments, the optimization function is expressed as:

[0127] ;

[0128] Among them, H0 is the normal model; H1 is the fault model; P(H0) is the prior probability of the normal model; P(H1) is the prior probability of the fault model; μ0 is the mean of the Gaussian distribution with uncertainty in the normal model; μ1 is the mean of the Gaussian distribution with uncertainty in the fault model; σ0 is the variance of the Gaussian distribution with uncertainty in the normal model; σ1 is the variance of the Gaussian distribution with uncertainty in the fault model.

[0129] In some embodiments, before obtaining the input data to be detected, it also includes: constructing a robotic arm model; the robotic arm model is a mathematical model used to describe the dynamic characteristics of the robotic arm; based on the robotic arm model, constructing a normal model; the normal model is a mathematical model used to describe the dynamic characteristics of the robotic arm when no failure occurs; determining the failure coefficient of the robotic arm; based on the failure coefficient and the robotic arm model, constructing a failure model; the failure model is a mathematical model used to describe the dynamic characteristics of the robotic arm when a failure occurs.

[0130] In some embodiments, the normal model is expressed as:

[0131] ;

[0132] ;

[0133] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix.

[0134] The expression of the fault model is:

[0135] ;

[0136] ;

[0137] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; and Z is the fault coefficient.

[0138] Specifically, based on the dynamic characteristics of the robotic arm, a robotic arm model is constructed. The expression of the robotic arm model is as follows:

[0139] ;

[0140] ;

[0141] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix.

[0142] Furthermore, different fault coefficients (ie, multiplicative fault coefficients) are set for the robot arm model to obtain a normal model and a fault model.

[0143] Specifically, determine the fault coefficient corresponding to the normal model and the fault coefficient corresponding to the fault model; determine the normal model based on the fault coefficient corresponding to the normal model and the robotic arm model; determine the fault model based on the fault coefficient corresponding to the fault model and the robotic arm model.

[0144] Generally, the fault coefficient value corresponding to the normal model is 1. At this time, the expressions of the robot arm model and the normal model are the same, that is, the expression of the normal model is as follows:

[0145] ;

[0146] ;

[0147] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix.

[0148] The value of the fault coefficient corresponding to the fault model can be set according to actual needs. The value range of the fault coefficient corresponding to the fault model is (0, 1). The expression of the fault model is as follows:

[0149] ;

[0150] ;

[0151] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; Z is the fault coefficient corresponding to the fault model.

[0152] The robotic arm fault detection method provided in this embodiment adopts a Bayesian decision method and utilizes prior knowledge and probabilistic models to optimize the fault detection process, significantly improving the fault detection performance and reliability. The method first establishes a mathematical model of the robotic arm, including a normal model and a fault model, and designs Kalman filters suitable for different models to improve state estimation accuracy. Through the Bayesian decision rule, combined with an optimization function (such as Battacharyya Bound), the optimal model is selected to achieve accurate fault diagnosis given the system input and output data. In addition, a differential evolution algorithm is used to optimize the system control input and design a control signal suitable for fault detection, thereby enhancing the system's detection capability and robustness for potential fault characteristics.

[0153] The present invention also provides a specific example of a method for detecting a mechanical arm fault, the main implementation steps of which can be referred to in Figure 2 ,Considering the normal model and the fault model, the correct model is selected through the Bayesian decision method, and the ,manipulator is judged to be in the normal model or the fault model, ,thus achieving fault detection.

[0154] The upper bound of the Bayesian risk can be given by the Battacharyya Bound. The transformed Battacharyya Bound is used as the optimization target to optimize the trajectory of the robotic arm. The differential evolution algorithm is used to find the trajectory with the minimum Battacharyya Bound. The input data is given according to the optimized trajectory, thereby reducing the detection error rate and realizing the design of friendly input data.

[0155] In this embodiment, it is assumed that the expression of a motor-driven single-rod robotic arm model is:

[0156] ;

[0157] Where J is the moment of inertia, θ is the rotation angle of the robot arm, and τ is the motor torque.

[0158] After discretization, the expression of the single-link robotic arm model is as follows:

[0159] ;

[0160] The robotic arm model can be further derived.

[0161] Furthermore, the robot arm fault detection method is verified through simulation.

[0162] Specifically, in the simulation, the value of the fault coefficient corresponding to the normal model is 1, and the expression of the normal model is as follows:

[0163] ;

[0164] ;

[0165] Specifically, in the simulation, considering the failure of the robot arm actuator, assuming that the actuator loses 30% of its efficiency, the failure coefficient Z corresponding to the failure model is 1-30%, that is, Z=0.7. The expression of the failure model is as follows:

[0166] ;

[0167] ;

[0168] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents the process noise; and v(k) represents the measurement noise.

[0169] It can be understood that in this embodiment, the system actuator output after the robot arm fails is The expression is as follows:

[0170] .

[0171] It should be noted that the fault type considered in this embodiment is a loss of 30% of the efficiency. In fact, it can also be extended to other fault types, such as a loss of M (the value of M is between 0 and 1) of efficiency. In this case, the fault coefficient Z corresponding to the fault model is 1-M.

[0172] In this embodiment, assuming that the control target is to rotate the motor angle of the robotic arm from 0 to 0.5 rad, in the differential evolution algorithm, the number of samples is 50, the mutation factor is 0.9, and the number of optimizations is 10. Subsequent simulations are performed to verify the robotic arm fault detection method.

[0173] The simulation results show that before and after optimization using the differential evolution algorithm, the robotic arm can move according to the preset motion trajectory, without affecting the realization of the robotic arm's working objectives; when the robotic arm fails, optimization using the differential evolution algorithm can improve the accuracy and reliability of fault detection, greatly reduce the Bayesian risk, and improve the accuracy of fault detection.

[0174] The present application also provides a device for detecting a mechanical arm fault. Figure 4 , Figure 4 4 is a schematic diagram of the structure of the robot arm fault detection device provided in the embodiment of the present application. In the embodiment of the present application, the robot arm fault detection device includes an acquisition module 410, a determination module 420 and a detection module 430.

[0175] The acquisition module 410 is used to acquire input data to be detected.

[0176] The input data to be detected is determined based on the motion trajectory of the robotic arm.

[0177] The discrimination module 420 is used to input the input data to be detected into the normal model and the fault model respectively, and obtain a first discrimination result output by the normal model and a second discrimination result output by the fault model.

[0178] The detection module 430 is used to make a decision through a Bayesian decision method based on the first discrimination result and the second discrimination result to obtain a fault detection result of the robotic arm.

[0179] In some embodiments, the detection module 430 is used to correct the first discrimination result based on the first Kalman filter to obtain a third discrimination result; correct the second discrimination result based on the second Kalman filter to obtain a fourth discrimination result; and make a decision on the third discrimination result and the fourth discrimination result through a Bayesian decision method to obtain a fault detection result of the robotic arm.

[0180] In some embodiments, the acquisition module 410 is used to obtain a preset motion trajectory of the robotic arm; based on the differential evolution algorithm, the preset motion trajectory is optimized to obtain the motion trajectory; wherein, the trajectory optimization process takes minimizing the optimization function as the optimization goal, and the optimization function is determined based on the Bayesian risk upper bound of the Bayesian decision method; the motion trajectory is tracked and detected to obtain the input data to be detected.

[0181] In some embodiments, the optimization function is expressed as:

[0182] ;

[0183] Among them, H0 is the normal model; H1 is the fault model; P(H0) is the prior probability of the normal model; P(H1) is the prior probability of the fault model; μ0 is the mean of the Gaussian distribution with uncertainty in the normal model; μ1 is the mean of the Gaussian distribution with uncertainty in the fault model; σ0 is the variance of the Gaussian distribution with uncertainty in the normal model; σ1 is the variance of the Gaussian distribution with uncertainty in the fault model.

[0184] In some embodiments, the robotic arm fault detection device further includes a building module.

[0185] A construction module is used to construct a robot arm model; the robot arm model is a mathematical model used to describe the dynamic characteristics of the robot arm; based on the robot arm model, a normal model is constructed; the normal model is a mathematical model used to describe the dynamic characteristics of the robot arm when no failure occurs; the failure coefficient of the robot arm is determined; based on the failure coefficient and the robot arm model, a failure model is constructed; the failure model is a mathematical model used to describe the dynamic characteristics of the robot arm when a failure occurs.

[0186] In some embodiments, the normal model is expressed as:

[0187] ;

[0188] ;

[0189] Among them, x(k) represents the system state of the manipulator at time k; x(k-1) represents the system state of the manipulator at time k-1; u(k) represents the motor input of the manipulator at time k; y(k) represents the system output of the manipulator at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; the expression of the fault model is:

[0190] ;

[0191] ;

[0192] Among them, x(k) represents the system state of the robot arm at time k; x(k-1) represents the system state of the robot arm at time k-1; u(k) represents the motor input of the robot arm at time k; y(k) represents the system output of the robot arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; and Z is the fault coefficient.

[0193] The embodiment of the present application also provides an electronic device, Figure 5 is a schematic diagram of the structure of the electronic device provided in the embodiment of the present application, such as Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other via the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the robot arm fault detection method.

[0194] Furthermore, the logic instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0195] An embodiment of the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the robot arm fault detection method provided by the above methods is implemented.

[0196] An embodiment of the present application also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the robotic arm fault detection method provided by the above methods.

[0197] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0198] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0199] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for detecting a fault in a robotic arm, characterized in that: include: Get the input data to be tested; The input data to be detected is determined based on the motion trajectory of the robotic arm; Inputting the input data to be detected into a normal model and a fault model respectively, obtaining a first discrimination result output by the normal model and a second discrimination result output by the fault model; Based on the first discrimination result and the second discrimination result, a decision is made by using a Bayesian decision method to obtain a fault detection result of the robotic arm; The method of making a decision based on the first discrimination result and the second discrimination result by using a Bayesian decision method to obtain a fault detection result of the robotic arm includes: Correcting the first discrimination result based on the first Kalman filter to obtain a third discrimination result; Correcting the second discrimination result based on the second Kalman filter to obtain a fourth discrimination result; Making a decision on the third discrimination result and the fourth discrimination result by using the Bayesian decision method to obtain a fault detection result of the robotic arm; The obtaining of input data to be detected includes: Obtaining a preset motion trajectory of the robotic arm; Based on the differential evolution algorithm, the preset motion trajectory is optimized to obtain the motion trajectory; wherein the trajectory optimization process takes minimizing an optimization function as an optimization goal, and the optimization function is determined based on the Bayesian risk upper bound of the Bayesian decision method; Tracking and detecting the motion trajectory to obtain the input data to be detected; The expression of the optimization function is: ； Wherein, H0 is the normal model; H1 is the fault model; P(H0) is the prior probability of the normal model; P(H1) is the prior probability of the fault model; μ0 is the uncertain mean of the Gaussian distribution of the normal model; μ1 is the uncertain mean of the Gaussian distribution of the fault model; σ0 is the uncertain variance of the Gaussian distribution of the normal model; σ1 is the uncertain variance of the Gaussian distribution of the fault model; Before obtaining the input data to be detected, the method further includes: Constructing a robotic arm model; the robotic arm model is a mathematical model used to describe the dynamic characteristics of the robotic arm; Based on the robot arm model, constructing the normal model; the normal model is a mathematical model used to describe the dynamic characteristics of the robot arm when no fault occurs; determining a failure coefficient of the robotic arm; Based on the fault coefficient and the robot arm model, constructing the fault model; the fault model is a mathematical model for describing the dynamic characteristics of the robot arm when a fault occurs; The expression of the normal model is: ； ； Wherein, x(k) represents the system state of the manipulator at time k; x(k-1) represents the system state of the manipulator at time k-1; u(k) represents the motor input of the manipulator at time k; y(k) represents the system output of the manipulator at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; The expression of the fault model is: ； ； Among them, x(k) represents the system state of the robotic arm at time k; x(k-1) represents the system state of the robotic arm at time k-1; u(k) represents the motor input of the robotic arm at time k; y(k) represents the system output of the robotic arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; and Z is the fault coefficient.

2. A robot arm fault detection device, characterized in that: include: An acquisition module is used to obtain input data to be detected; The input data to be detected is determined based on the motion trajectory of the robotic arm; A discrimination module, configured to input the input data to be detected into a normal model and a fault model respectively, and obtain a first discrimination result output by the normal model and a second discrimination result output by the fault model; a detection module, configured to make a decision by using a Bayesian decision method based on the first discrimination result and the second discrimination result to obtain a fault detection result of the robotic arm; The detection module is configured to correct the first discrimination result based on the first Kalman filter to obtain a third discrimination result; Correcting the second discrimination result based on the second Kalman filter to obtain a fourth discrimination result; Making a decision on the third discrimination result and the fourth discrimination result by using the Bayesian decision method to obtain a fault detection result of the robotic arm; The acquisition module is used to acquire the preset motion trajectory of the robotic arm; Based on a differential evolution algorithm, the preset motion trajectory is optimized to obtain the motion trajectory; wherein the trajectory optimization process takes minimizing an optimization function as an optimization goal, and the optimization function is determined based on the Bayesian risk upper bound of the Bayesian decision method; the motion trajectory is tracked and detected to obtain the input data to be detected; The expression of the optimization function is: ； Wherein, H0 is the normal model; H1 is the fault model; P(H0) is the prior probability of the normal model; P(H1) is the prior probability of the fault model; μ0 is the uncertain mean of the Gaussian distribution of the normal model; μ1 is the uncertain mean of the Gaussian distribution of the fault model; σ0 is the uncertain variance of the Gaussian distribution of the normal model; σ1 is the uncertain variance of the Gaussian distribution of the fault model; A construction module is configured to construct a robot arm model; the robot arm model is a mathematical model used to describe the dynamic characteristics of the robot arm; based on the robot arm model, the normal model is constructed; the normal model is a mathematical model used to describe the dynamic characteristics of the robot arm when no fault occurs; a fault coefficient of the robot arm is determined; based on the fault coefficient and the robot arm model, the fault model is constructed; the fault model is a mathematical model used to describe the dynamic characteristics of the robot arm when a fault occurs; The expression of the normal model is: ； ； Wherein, x(k) represents the system state of the manipulator at time k; x(k-1) represents the system state of the manipulator at time k-1; u(k) represents the motor input of the manipulator at time k; y(k) represents the system output of the manipulator at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; The expression of the fault model is: ； ； Among them, x(k) represents the system state of the robotic arm at time k; x(k-1) represents the system state of the robotic arm at time k-1; u(k) represents the motor input of the robotic arm at time k; y(k) represents the system output of the robotic arm at time k; w(k) represents process noise; v(k) represents measurement noise; A is the first coefficient matrix; B is the second coefficient matrix; C is the third coefficient matrix; D is the fourth coefficient matrix; E is the fifth coefficient matrix; and Z is the fault coefficient.

3. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the robot arm fault detection method according to claim 1 is implemented.

4. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the robot arm fault detection method as claimed in claim 1 is implemented.

5. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the robot arm fault detection method as claimed in claim 1 is implemented.

Images