Method for collision detection of an industrial robot based on vibration signals

Abstract

The application aims to provide a kind of industrial robot collision detection method based on vibration signal, utilize vibration sensor to collect multiple groups of sensor signals under collision condition;Envelope spectrum analysis is carried out on the signal, and the frequency band where the collision frequency is located is identified;VMD processing is carried out on the signal, and the interference of noise is filtered out;Time-frequency analysis is carried out on the signal after VMD, and the frequency domain, time domain and entropy value features of the signal under collision condition are extracted;The mean value of multiple collision features is taken, and the standard feature vector of collision condition is constructed;The latest data is collected, VMD filtering and feature extraction are carried out, the correlation between the data and the collision standard feature vector is calculated, and if the correlation exceeds the threshold value, it indicates that the robot collides, so as to realize the collision detection of industrial robot.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of robot technology, and particularly relates to an industrial robot collision detection method based on vibration signals. BACKGROUND

[0002] In recent decades, industrial robots have developed rapidly, and as an important part of high-end manufacturing equipment, the industrial robots have high technical added value and wide application range. People hope that robots can complete more complex work, and thus multi-robot cooperation and human-robot cooperation robots have been developed. Since industrial robots inevitably perform heavy work, and the industrial robots have many degrees of freedom and complex running conditions, human-robot cooperation may cause the industrial robots to collide with people or the surrounding environment in the case that the trajectory of the industrial robots is not completely clear. In addition, improper operation of the operator may also cause the industrial robots to run off the predetermined trajectory and thus collide. At this time, how to perform corresponding collision detection to minimize the loss becomes a crucial problem.

[0003] At present, the collision detection methods for industrial robots at home and abroad are basically divided into two kinds, one is based on external sensor detection, and the other is based on collision detection without external sensor. The collision detection method based on external sensor is to place the required sensor outside the industrial robot, and to judge whether the industrial robot collides by reading the value of the sensor. This method can generally accurately identify. However, the sensor accuracy greatly affects the result when the collision is judged by installing external sensors, and also increases the complexity of the robot structure to some extent. At present, the external sensors used for collision detection of industrial robots include the following: electronic skin, vision sensor, vibration sensor and joint torque sensor. The collision detection without external sensor generally detects the joint position by using the position detection unit of the joint motor, calculates the size of the external torque by establishing the dynamics model of the industrial robot, and sets the torque collision threshold to detect the occurrence of collision. SUMMARY

[0004] In view of the problems existing in the prior art, the purpose of the present application is to provide an industrial robot collision detection method, which aims to improve the collision recognition accuracy under the condition that only the attached vibration sensor is used without affecting the normal operation of the industrial robot. The method is applicable to an industrial robot system composed of at least one robot joint, a servo driver, a controller, a vibration sensor and a signal acquisition module. The robot joint includes a connecting rod, a reducer and a servo motor. The controller is connected with the servo driver, the servo driver is connected with the servo motor, and the servo motor is connected in series with the reducer and the connecting rod. The signal of the vibration sensor is subjected to collision frequency judgment, filtering and noise elimination, feature extraction, standard feature vector construction, and calculation of the correlation between the actual and standard feature vectors to judge whether the industrial robot collides.

[0005] The main algorithm is that a plurality of groups of sensor signals under collision conditions are collected by using a vibration sensor, envelope spectrum analysis is performed on the signals, the frequency band where the collision frequency is located is identified, VMD processing is performed on the signals, noise interference is filtered out, time-frequency analysis is performed on the signals after VMD, the frequency domain, time domain and entropy value features of the signals under the collision condition are extracted, the mean value of the plurality of collision features is taken, and the standard feature vector of the collision condition is constructed; the latest data is collected, VMD filtering and feature extraction are performed on the data, the correlation between the data and the standard feature vector of the collision is calculated, and if the correlation exceeds a threshold value, it is indicated that the robot collides, so that the collision detection of the industrial robot is realized.

[0006] The method can improve the accuracy and cost performance of collision detection, and reduce the complexity of the control system.

[0007] The technical scheme specifically adopted by the present application to solve the technical problems is:

[0008] A collision detection method for an industrial robot based on a vibration signal, based on a robot system composed of at least one robot joint, a servo driver, a controller, a vibration sensor and a signal acquisition module; the robot joint comprises a connecting rod, a speed reducer and a servo motor; the controller is connected with the servo driver, the servo driver is connected with the servo motor, and the servo motor is connected in series with the speed reducer and the connecting rod; the vibration sensor is attached to the robot base for collecting vibration signals; characterized in that:

[0009] First, a plurality of groups of sensor signals under collision conditions are collected by using the vibration sensor;

[0010] Then, envelope spectrum analysis is performed on the signals, and the frequency band where the collision frequency is located is identified;

[0011] Then, VMD processing is performed on the signals to filter out noise interference, time-frequency analysis is performed on the signals after VMD, and the frequency domain, time domain and entropy value features of the signals under the collision condition are extracted; the mean value of the plurality of collision features is taken, and the standard feature vector of the collision condition is constructed;

[0012] In use, the latest data is collected, VMD filtering and feature extraction are performed, the correlation with the standard feature vector of the collision is calculated, and if the correlation exceeds a threshold value, it is indicated that the robot collides, so that the collision detection of the industrial robot is realized.

[0013] Further, the process of constructing the standard feature vector of the collision condition specifically comprises the following steps:

[0014] Step S1: obtaining the vibration signal y1 on the industrial robot base when a collision occurs;

[0015] Step S2: envelope spectrum analysis is performed on the signal y1 to obtain a frequency value ω* corresponding to the maximum amplitude of the envelope spectrum;

[0016] Step S3: the signal y1 is subjected to VMD (Variational Modal Decomposition) processing, and a VMD modal component is selected through ω* to reconstruct the signal y1, thereby obtaining a filtered signal

[0017] Step S4: the signal is subjected to time-frequency feature extraction; time-domain features including at least: peak value F1, mean value F2, variance F3, kurtosis F4, skewness F5, waveform factor F6, peak factor F7, pulse factor F8, margin factor F9, and clearance factor F10 are obtained; frequency-domain features including at least: center frequency F11, mean square frequency F12, frequency variance F13, and frequency standard deviation F14 are obtained; and entropy value features including at least: information entropy F15, singularity entropy F16, power entropy F17, and wavelet entropy F18 are obtained. 10 11 12 13 14 15 16 17 18

[0018] Step S5: steps S1 to S4 are repeated to obtain multiple sets of feature values of the signal , the mean value of each feature value is calculated, and a standard feature vector of the signal is constructed

[0019] Further, the process of collision detection specifically includes the following steps:

[0020] Step S6: the vibration signal y2 at the latest time on the base of the industrial robot is collected, and steps S3 to S5 are performed on the signal y2 to construct a feature vector of the signal

[0021] Step S7: the Bray-Curtis correlation value R of F1 and F2 is calculated;

[0022] Step S8: a threshold value b is given, when R > b, it is judged that the robot collides at the current time, and when R ≤ b, it is judged that the robot does not collide at the current time.

[0023] Further, the vibration sensor signal attached to the base of the robot is collected through Labview by using the NI acquisition card and the signal conditioner.

[0024] Further, step S2 specifically includes the following steps:

[0025] Step S2-1: hilbert transform is performed on the signal y1, and the hilbert transform formula is:​​​​​​​​​

[0026]

[0027] where y h (t) is the function after the hilbert transform of y1;

[0028] Step S2-2: Construct the analytic signal, the expression is:

[0029] z(t) = y1(t) + jy h (t) (2)

[0030] Step S2-3: Calculate the envelope signal, the calculation formula is:

[0031]

[0032] Step S2-4: Perform FFT transform on A(t), the calculation formula is:

[0033]

[0034] Step S2-5: Calculate the frequency value ω* corresponding to the maximum of A(ω).

[0035] Further, step S3 specifically includes the following steps:

[0036] Step S3-1: Construct the VMD variational model, the expression is:

[0037]

[0038] where u k (t) is the kth modal component, K is the number of modal components, δ(t) is the impulse function, u k (t) and ω k is the kth modal component and its corresponding center frequency estimated by VMD;

[0039] Step S3-2: Establish an unconstrained augmented Lagrange equation for solving the VMD constrained variational problem:

[0040]

[0041] where α is the penalty factor, λ(t) is the Lagrange multiplier operator;

[0042] Step S3-3: Use the multiplicative operator alternating direction method to iteratively solve the unconstrained equation (6), the iterative calculation formula is:

[0043]

[0044]

[0045]

[0046] wherein w is the angular frequency, n is the iteration number, is the modal component calculated in the n+1 iteration, is the modal component center frequency calculated in the n+1 iteration, λ n+1 is the Lagrange multiplier operator calculated in the n+1 iteration;

[0047] Step S3-4: Given the initial modal component initial modal component center frequency and initial Lagrange multiplier operator λ 1 (ω), repeat step S3-3 until the following iteration stopping judgment condition is satisfied:

[0048]

[0049] wherein ε is the iteration accuracy;

[0050] Step S3-5: Calculate the frequency corresponding to the peak value of the K modal components, and reconstruct the signal y1 with the modal component whose frequency is equal to ω* to obtain the signal

[0051] Further, in step S7:

[0052] The specific calculation formula of the Bray-Curtis correlation value R of M1 and M2 is:

[0053]

[0054] Compared with the prior art, the present application and the preferred scheme thereof can well balance the accuracy and cost performance of collision detection, and reduce the complexity of the control system. BRIEF DESCRIPTION OF DRAWINGS

[0055] The present application will be further described in detail below in combination with the drawings and specific embodiments:

[0056] Figure 1 is the flow chart of the method of the embodiment of the present application;

[0057] Figure 2 is the envelope spectrum analysis schematic diagram of the method of the embodiment of the present application;

[0058] Figure 3 is the VMD decomposition schematic diagram of the method of the embodiment of the present application. DETAILED DESCRIPTION

[0059] In order to make the features and advantages of the present patent more obvious and easy to understand, the following embodiments are specifically described as follows:

[0060] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0061] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0062] The method described in this embodiment is based on a typical robot system consisting of at least one robot joint, a servo driver, a controller, a vibration sensor, and a signal acquisition module. The robot joint includes a link, a reducer, and a servo motor. The controller is connected to the servo driver, the servo driver is connected to the servo motor, and the servo motor is connected in series with the reducer and the link.

[0063] The vibration sensor is attached to the robot's base to collect vibration signals.

[0064] Preferably, in this embodiment, an NI data acquisition card and a signal conditioner are used to acquire vibration sensor signals attached to the robot base via LabVIEW.

[0065] The basic design idea of ​​this embodiment is as follows: use vibration sensors to collect sensor signals under multiple collision conditions; perform envelope spectrum analysis on the signals to identify the frequency band where the collision frequency is located; perform VMD processing on the signals to filter out noise interference; perform time-frequency analysis on the VMD signals to extract the frequency domain, time domain and entropy value features of the signals under collision conditions; take the average of multiple collision features to construct a standard feature vector of the collision conditions.

[0066] Then, by collecting the latest data, performing VMD filtering and feature extraction, and calculating its correlation with the collision standard feature vector, if the correlation exceeds the threshold, it indicates that the robot has collided, thus realizing collision detection of industrial robots.

[0067] like Figure 1 As shown, the specific implementation steps of the method in this embodiment include:

[0068] Step S1: Obtain the vibration signal y1 on the industrial robot base when the collision occurs. The specific steps are as follows: Use the NI acquisition card and signal conditioner to acquire the vibration sensor signal attached to the robot base through LabVIEW.

[0069] Step S2: As shown in the figure, envelope spectrum analysis is performed on the signal y1 to obtain a frequency value ω* corresponding to the maximum amplitude of the envelope spectrum, and the specific steps are as follows: Figure 2

[0070] Step S2-1: Perform hilbert transform on the signal y1, and the hilbert transform formula is as follows:

[0071]

[0072] Where y h (t) is the function after hilbert transform of y1;

[0073] Step S2-2: Construct the analytic signal, and the expression is as follows:

[0074] z(t)=y1(t)+jy h (t)(2)

[0075] Step S2-3: Calculate the envelope signal, and the calculation formula is as follows:

[0076]

[0077] Step S2-4: Perform FFT transform on A(t), and the calculation formula is as follows:

[0078]

[0079] Step S2-5: Calculate the frequency value ω* corresponding to the maximum A(ω).

[0080] Step S3: As shown in the figure, perform VMD processing on the signal y1, select the VMD modal component through ω* to reconstruct the signal y1, and obtain the filtered signal Figure 3 The specific steps are as follows:

[0081] Step S3-1: Construct the VMD variational model, and the expression is as follows:

[0082]

[0083] Where u k (t) is the kth modal component, K is the number of modal components, δ(t) is the impulse function, u k (t) and ω k are the kth modal component and its corresponding center frequency estimated by VMD;

[0084] Step S3-2: Establish the unconstrained augmented Lagrange equation for solving the VMD constrained variational problem:

[0085] ​​

[0086] wherein, a is a penalty factor, λ(t) is a Lagrange multiplier operator;

[0087] Step S3-3: the unconstrained equation (6) is solved by using a multiplication operator alternating direction method iteration, and the iteration calculation formula is:

[0088]

[0089]

[0090]

[0091] wherein, w is an angular frequency, n is an iteration number, is a modal component calculated in the n+1 iteration, is a modal component center frequency calculated in the n+1 iteration, λ n+1 (ω) is a Lagrange multiplier operator calculated in the n+1 iteration.

[0092] Step S3-4: given an initial modal component an initial modal component center frequency and an initial Lagrange multiplier operator λ 1 (ω), the step S3-3 is repeated until the following iteration stop judgment condition is satisfied:

[0093]

[0094] wherein, ε is an iteration precision.

[0095] Step S3-5: the frequency corresponding to the peak value of the K modal components is calculated, the modal component whose frequency is equal to ω* is reconstructed to obtain a signal .

[0096] Step S4: time-frequency feature extraction is performed on the signal , and the specific steps are as follows: time domain features such as a peak value F1, a mean value F2, a variance F3, a kurtosis F4, a skewness F5, a waveform factor F6, a peak factor F7, a pulse factor F8, a margin factor F9, and a clearance factor F 10 , frequency domain features such as a center frequency F 11 , a mean square frequency F 12 , a frequency variance F 13 , a frequency standard deviation F 14 , and entropy value features such as an information entropy F 15 , a singular entropy F 16 , a power entropy F 17 , and a wavelet entropy F 18 are obtained.

[0097] Step S5: repeat step S1-S4 to obtain multiple sets of characteristic values of the signal, calculate the mean value of each characteristic value, and construct the standard eigenvector of the signal

[0098] Step S6: collect the vibration signal y2 at the latest time on the base of the industrial robot, and perform step S3-S4 on the signal y2 to construct the eigenvector of the signal

[0099] Step S7: calculate the Bray-Curtis correlation value R of M1 and M2, and the specific steps are as follows:

[0100]

[0101] Step S8: set a threshold value b, when R>b, it is judged that the robot collides at the current time, and when R≤b, it is judged that the robot does not collide at the current time.

[0102] Those skilled in the art will understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0103] The present application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. Figure 1 The functions specified in one or more flows and / or blocks.

[0104] These computer program instructions can also be stored in a computer-readable memory that can cause the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. Figure 1 The functions specified in one or more flows and / or blocks. ​

[0105] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate a computer implemented process, so that the instructions executed on the computer or other programmable data processing devices provide a process for implementing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0106] The above description is only the preferred embodiment of the present application, not other forms of the present application, any skilled in the art can use the above disclosed technical content to change or modify the equivalent embodiments of equivalent changes. But any simple modification, equivalent change and modification of the above embodiments without departing from the technical solution of the present application, according to the technical essence of the present application, still belongs to the protection scope of the technical solution of the present application.

[0107] The patent is not limited to the above best mode, anyone can draw other various forms of vibration signal based industrial robot collision detection method under the inspiration of the patent, any equivalent change and modification made according to the patent application scope shall belong to the scope of the patent.

Claims

1. A vibration signal-based collision detection method for an industrial robot, based on a robot system composed of at least one robot joint, a servo driver, a controller, a vibration sensor, and a signal acquisition module; the robot joint comprises a connecting rod, a reducer, and a servo motor; the controller is connected to the servo driver, the servo driver is connected to the servo motor, and the servo motor is connected in series with the reducer and the connecting rod; the vibration sensor is attached to the robot base for collecting vibration signals; characterized in that: first, a plurality of sensor signals in collision situations are collected by the vibration sensor; then, envelope spectrum analysis is performed on the signals to identify the frequency band where the collision frequency is located; the signals are then processed by VMD to filter out noise interference; time-frequency analysis is performed on the VMD-processed signals to extract the frequency domain, time domain, and entropy value features of the signals in the collision situation; the mean value of the collision features of the plurality of groups is taken to construct a standard feature vector of the collision situation; in use, the latest data is collected, VMD filtering and feature extraction are performed, and the correlation with the collision standard feature vector is calculated; if the correlation exceeds a threshold value, it indicates that the robot has collided, thereby realizing collision detection of the industrial robot. The process of constructing the standard feature vector of the collision situation specifically comprises the following steps: Step S1: obtaining the vibration signal y1 on the industrial robot base when a collision occurs; Step S2: performing envelope spectrum analysis on the signal y1 to obtain the frequency value ω* corresponding to the maximum amplitude of the envelope spectrum; The process of collision detection specifically comprises the following steps:

2. The vibration signal based industrial robot collision detection method according to claim 1, characterized in that, Step S7: calculating the Bray-Curtis correlation value R of M1 and M2; Step S8: given a threshold value b, when R > b, it is judged that the robot has collided at the current time, and when R ≤ b, it is judged that the robot has not collided at the current time. The vibration sensor signal attached to the robot base is collected by a NI acquisition card and a signal conditioner through Labview. Step S3: Perform a variational mode decomposition (VMD) process on the signal y1, select a VMD modal component through ω* to reconstruct the signal y1, and obtain a filtered signal Step S4: time-frequency feature extraction is performed on the signal to obtain time-domain features including at least: peak value F1, mean value F2, variance F3, kurtosis F4, skewness F5, waveform factor F6, peak factor F7, pulse factor F8, margin factor F9, clearance factor F 10 ; frequency-domain features including at least: center of gravity frequency F 11 , mean square frequency F 12 , frequency variance F 13 , frequency standard deviation F 14 ; and entropy value features including at least: information entropy F 15 , singular entropy F 16 , power entropy F 17 , wavelet entropy F 18 ​ Step S5: repeating steps S1-S4 to obtain multiple sets of eigenvalues of the signal, calculating the mean of each eigenvalue, and constructing the standard eigenvector of the signal Step S5: repeating steps S1-S4 to obtain multiple sets of eigenvalues of the signal, calculating the mean of each eigenvalue, and constructing the standard eigenvector of the signal 3. The vibration signal based industrial robot collision detection method according to claim 2, characterized in that, Step S2 specifically comprises the following steps: Step S6: collect the vibration signal y2 at the latest time on the base of the industrial robot, and perform steps S3-S5 on the signal y2 to construct the feature vector of the signal Step S2-1: performing hilbert transformation on the signal y1, and the hilbert transformation formula is: Step S2-2: constructing the analytic signal, and the expression is:

4. The vibration signal based industrial robot collision detection method according to claim 1, characterized in that: Step S2-3: calculating the envelope signal, and the calculation formula is:

5. The vibration signal based industrial robot collision detection method according to claim 2, characterized in that, Step S2-4: performing FFT transformation on A(t), and the calculation formula is: where ω is the angular frequency; where y h (t) is the function after the hilbert transform of y1; Step S2-5: calculating the frequency value ω* corresponding to the maximum A(ω). z(t) = yl(t) + jy h (t)(2) Step S3 specifically comprises the following steps: Step S3-1: constructing a VMD variational model, and the expression is: Step S3-2: establishing an unconstrained augmented Lagrange equation for solving the VMD constrained variational problem: where α is the penalty factor, and λ(t) is the Lagrange multiplier operator; 6. The vibration signal based industrial robot collision detection method according to claim 2, characterized in that, Step S3-3: solving the unconstrained equation (6) by using the multiplicative operator alternating direction method, and the iterative calculation formula is: where ε is the iteration accuracy; where u k (t) is the kth modal component, K is the number of modal components, δ(t) is the impulse function, u k (t) and ω k k are the kth modal component and its corresponding center frequency estimated by VMD; 7.The vibration signal-based collision detection method for an industrial robot according to claim 3, characterized in that: In step S7: The specific calculation formula for calculating the Bray-Curtis correlation value R of M1 and M2 is: where ω is the angular frequency, n is the iteration number, is the modal component calculated at the n+1 iteration, is the modal component center frequency calculated at the n+1 iteration, λ n+1 is the Lagrange multiplier operator calculated at the n+1 iteration; Step S3-4: Given the initial modal component Initial modal component center frequency And initial Lagrange multiplier operator λ 1 (ω), repeat step S3-3 until the following iteration stop judgment condition is met: ​ Step S3-5: Calculate the frequency corresponding to the peak value of the K modal components, and reconstruct the signal y1 with the modal component whose frequency is equal to ω*, to obtain the signal ​ ​ ​

Images