Fault data preprocessing method for joint speed reducer for industrial robot

Abstract

The invention relates to the technical field of industrial robot fault diagnosis, in particular to a fault data preprocessing method for a joint reducer for an industrial robot, which comprises the following steps of: searching fault characteristic information of a reducer modulated by rotating speed in current signal frequency domain distribution; according to the method, speed reducer fault characteristic information is extracted and a speed reducer fault current signal time-domain waveform without rotation speed modulation interference is obtained according to the amplitude relation of the current signal and the corresponding amplitude relation before and after modulation, the current time-domain waveform eliminates the interference of the rotation speed signal, and data support and technical support are provided for subsequent accurate extraction of speed reducer fault characteristics based on the current signal.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial robot fault diagnosis, and in particular to a method for preprocessing fault data of a joint reducer for an industrial robot. Background Art

[0002] The fault diagnosis of joint reducers for industrial robots based on current signals only requires the Hall sensor to directly collect the input current of the drive motor. Compared with other signal types (such as torque, vibration, strain, etc.), no additional sensor installation space is required. In addition, in addition to low cost and no need for complex equipment, current acquisition is usually used as data backup for companies using industrial robots and will be uploaded to the cloud for industrial robot equipment status monitoring and fault diagnosis.

[0003] At present, the diagnosis of reducer faults based on current signals mainly includes three steps. First, the current signal is preprocessed, including filtering, data alignment, etc., and then the reducer fault features are extracted based on the current signal. Generally, the extracted reducer features are mainly extracted from the time domain waveform characteristics, frequency distribution characteristics, etc., and finally the machine learning method is used to complete the reducer fault identification model training and testing.

[0004] Although the current signal contains the characteristic information of the reducer fault, the characteristic information of the reducer in the current signal will be modulated by the speed. Usually, the reducer fault frequency is much higher than the speed frequency, and due to the difficulty of fine modeling of the reducer, its fault frequency is often unknown. This makes it difficult to accurately demodulate the characteristic frequency of the reducer fault by combining traditional demodulation with low-pass, band-pass and other filtering methods. Under the premise of being disturbed by the speed modulation, the current signal waveform of the same reducer fault type is not the same under different speed conditions, which seriously affects the accurate extraction of the reducer fault characteristics. This is also the key reason why it is difficult to accurately identify reducer faults based on current data-driven technology. Summary of the invention

[0005] In view of this, the purpose of the present invention is to propose a joint reducer fault data preprocessing method for an industrial robot to solve the problem of insufficient accuracy in extracting reducer fault features.

[0006] Based on the above purpose, the present invention provides a method for preprocessing fault data of a joint reducer for an industrial robot, comprising:

[0007] S1. Perform spectrum analysis on the current signal, extract the position of the frequency corresponding to the maximum amplitude in the spectrum diagram, and obtain the current frequency modulation frequency

[0008] S2. Set a certain range of allowable error And assign relevant parameters u=2, v=1, w=1, i=1. According to the allowable error and current frequency modulation frequency, calculate the maximum value i of the cycle parameter that meets the allowable error. max , w is used to accumulate the number of original fault features, i is the search parameter, according to the modulation and demodulation relationship, it is assumed that u is the multiple relationship between the modulated signal feature and the feature amplitude of the original signal, and v is the difference between a pair of modulation feature frequencies and the original signal feature frequencies and the current frequency modulation frequency The multiple relationship between them makes:

[0009]

[0010]

[0011] f n is the characteristic frequency of the reducer fault, f * is the frequency after the characteristic frequency of the reducer fault is modulated;

[0012] S3, determine the modulation frequency f i * The corresponding amplitude Is it satisfied? If it is satisfied, it is considered that the frequency is caused by noise interference, and the process goes to step S10. If it is not satisfied, the process goes to step S4. i * With another modulation frequency are the original characteristic frequencies of the reducer f n1 The corresponding modulation frequency pairs;

[0013] S4, according to the modulation frequency f i * and the current modulation frequency f c * calculate Judge f i * and The corresponding amplitude and Whether the following conditions are met, if so, execute step S5, otherwise jump to step S6;

[0014]

[0015] S5. Calculate characteristic frequency f n_w And the corresponding amplitude A n_w , let w=w+1, jump to step S10;

[0016] S6. Calculate the modulation frequency Judgement i * , and Whether the corresponding amplitude satisfies the conditions of the following formula, if so, execute step S7, otherwise jump to step S8, where: and is the original characteristic frequency f of the reducer n2 The corresponding modulation frequency pair is:

[0017]

[0018] S7. Calculate characteristic frequency f n_w and f n_w+1 And the corresponding amplitude A n_w and A n_w+1 , let w=w+2, go to step 10;

[0019] S8. Calculate f l * , and determine the modulation frequency f i * , f l * And whether the corresponding amplitude satisfies the conditions of the following formula, if so, go to step S9, otherwise go to step S10, where and f l * is the original characteristic frequency f of the reducer n3 The corresponding modulation frequency pairs;

[0020] S9. Calculate characteristic frequency f n_w 、f n_w+1 and f n_w+2 And the corresponding amplitude A n_w , A n_w+1 and A n_w+2 , and execute w=w+3 and then go to step S10;

[0021] S10, determine whether the search parameter i reaches the maximum value i max If yes, set i=i+1 and return to step S3; otherwise, go to step S11;

[0022] S11, judging whether the current signal is demodulated according to whether v is equal to 1, if yes, go to step S12, otherwise demodulate the current signal to obtain a demodulated signal y*, and perform spectrum analysis on it, according to the demodulation frequency in the demodulated signal and the current frequency modulation frequency f c * , respectively adjust the parameters u and v to twice their original values, and return to step S3 to extract the original characteristic frequency of the reducer under the modulation signal;

[0023] S12, judge f n_p and f n_q and the amplitude A n_p and An_q Does it satisfy the following formula? If yes, it is considered that there is a pair of common characteristic frequencies f in the modulated signal and the demodulated signal. n_t , go to step S13 to calculate, otherwise go to step S14:

[0024] f n_p =f n_q

[0025] |A n_p -A n_q |≤A th ;

[0026] Among them, the original characteristic frequencies of the reducer extracted from the current signal and the demodulated signal are stored in arrays P and Q respectively, f n_p and f n_q are the pth and qth characteristic frequencies in P and Q respectively, and the corresponding amplitude is A n_p and A n_q ;

[0027] S13, calculate the common characteristic frequency f n_t The corresponding amplitude is A n_t , where t is used to accumulate f w_t The number of characteristic frequencies with the same internal frequency and very close amplitude difference, let t = t + 1, and go to step S14;

[0028] S14, determine whether q is equal to w, if yes, go to step S15, otherwise, perform q=q+1 and return to step S12;

[0029] S15, execute p=p+1, judge f n_p and f n_p-1 Does it satisfy the following relationship?

[0030] f n_p <f n_p-1

[0031]

[0032] If yes, proceed to the next step, otherwise return to step S13;

[0033] S16, the filtered f n_t and A n_t Perform inverse Fourier transform to obtain the current signal that is not disturbed by speed modulation.

[0034] Preferably, in step S4, calculate The formula is:

[0035]

[0036] Preferably, in step S5, the characteristic frequency f is calculatedn_w and amplitude A n_w The formula is:

[0037]

[0038]

[0039] Preferably, in step S6, f is calculated k * The formula is:

[0040]

[0041] Preferably, in step S7, the characteristic frequency f is calculated n_w and f n_w+1 and the amplitude A n_w and A n_w+1 The formula is:

[0042]

[0043]

[0044]

[0045]

[0046] Preferably, in step S8, f is calculated l * The formula is:

[0047]

[0048] Preferably, in step S9, the characteristic frequency f is calculated n_w 、f n_w+1 and f n_w+2 and the amplitude A n_w , A n_w+1 and A n_w+2 The formula is:

[0049] f n1 =f i * +v·f c *

[0050]

[0051]

[0052]

[0053]

[0054]

[0055] Preferably, in step S11, the formula for calculating the demodulated signal y* is:

[0056]

[0057] Preferably, in step S13, the common characteristic frequency f is calculated n_t and the corresponding amplitude A n_t The formula is:

[0058] f n_t =f n_p

[0059]

[0060] The beneficial effects of the present invention: In order to solve the problem of accurately extracting the reducer fault characteristic information and eliminate the influence of speed modulation interference, the present invention proposes a joint reducer fault data preprocessing method for an industrial robot, by searching for the reducer fault characteristic information modulated by speed in the current signal frequency domain distribution, and based on the difference relationship between the reducer fault characteristic frequency and the speed frequency, as well as the corresponding amplitude relationship before and after modulation, the reducer fault characteristic information is extracted, and the reducer fault current signal time domain waveform without speed modulation interference is obtained. The current time domain waveform eliminates the interference of the speed signal, and provides data support and technical support for the subsequent accurate extraction of the reducer fault characteristics based on the current signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 Schematic diagram of the reducer fault data preprocessing process according to an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0064] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0065] The data-driven method is the mainstream method currently used for fault diagnosis of joint reducers for industrial robots. In the current signal, the characteristic frequency of reducer faults will be modulated by the speed, which seriously affects the reduction gear fault feature extraction link based on the data-driven method. The present invention proposes a method for preprocessing fault data of joint reducers for industrial robots. Based on the relationship between reducer fault information and speed modulation, the reducer fault characteristic signal is restored, the speed information interference is eliminated, and data support and technical support are provided for the accurate extraction of reducer fault features based on current signals.

[0066] In the process of preprocessing the reducer fault data, at the data sampling frequency f s When the data length is determined, the data frequency resolution is determined. In addition, considering that the reducer is a circular rotation motion, the occurrence period of different fault types is directly related to the rotation angle of the reducer. According to the reducer mechanism, assuming that the speed of the reducer input is n and the reduction ratio is r, when the reducer output rotates through the angle θ, the number of sampling points N can be calculated as follows, which is also the main basis for selecting the data preprocessing window width.

[0067]

[0068] Although the current signal contains the fault characteristic information of the reducer, it will be modulated by the speed. Assume that the characteristic frequency of the reducer fault is f n , in the current signal, the frequency used to drive the motor speed modulation is f c * , the frequency after the reducer fault characteristic frequency is modulated is f * , then the three have the following relationship:

[0069] f * =f n ±v·f c * (2)

[0070]

[0071] Among them, A * is the frequency f * The corresponding amplitude, A n is the frequency f n The corresponding amplitudes, u and v, are 2 and 1 for all formulas before demodulation.

[0072] Usually, the main energy in the current signal is used for speed frequency modulation. Therefore, in the frequency domain of the current signal, the frequency component corresponding to the maximum amplitude is the current frequency modulation frequency f. c * Therefore, based on the energy distribution of the current signal in the frequency domain, we can first obtain the speed frequency in the current signal. and amplitude

[0073] Assume f i * , is the original characteristic frequency f of the reducer n1 A pair of characteristic modulation frequencies, the frequency and amplitude should satisfy the following conditions at the same time,

[0074]

[0075]

[0076] in, is the allowable error considering the influence of noise.

[0077] In order to further reduce noise interference, the present invention uses and The weighted average of the original characteristic frequency f of the reducer n1 The amplitude A n1 ,

[0078]

[0079]

[0080] Considering that there are multiple original characteristic frequencies of the reducer, the frequency amplitude of the original characteristic frequency of the reducer after being modulated by the speed may be superimposed. Assume f i * and is the original characteristic frequency f of the reducer n1 A pair of characteristic modulation frequencies, and and is the original characteristic frequency f of the reducer n2 A pair of characteristic modulation frequencies, then is the common frequency of the two, and its amplitude If there is a superposition relationship, then while satisfying formula (4), the following conditions should also be met:

[0081]

[0082]

[0083] in, for The corresponding amplitude.

[0084] In this case, the original characteristic frequency and amplitude of the reducer (f n1 According to formula (6), the calculation is as follows:

[0085]

[0086] Consider the relationship between the amplitudes of the three original characteristic frequencies of the reducer after being modulated. Assuming f i * and and and and f l * They are the original characteristic frequencies f of the reducer respectively. n1 , f n2 and f n3 The corresponding modulation frequency pair, f n1 and f n2 The modulation signal characteristic has a common frequency, f n2 and f n3 The modulation signal characteristic has a common frequency, and The corresponding amplitudes at all have a superposition relationship and should satisfy formula (4), formula (8) and the following conditions at the same time:

[0087]

[0088]

[0089] in, f l * The corresponding amplitude.

[0090] At this time, the original characteristic frequency and amplitude of the reducer (f n1 and f n2 Calculated according to formula (6) and formula (10) respectively) as follows:

[0091]

[0092]

[0093]

[0094]

[0095] The superposition of amplitudes of more original characteristic frequencies of reducers after modulation rarely occurs in actual projects and is not considered in the present invention.

[0096] In order to avoid the problem of wrong extraction or omission of the original characteristic frequency of the reducer in the above extraction method, the present invention demodulates the original collected signal according to formula (19), and extracts the original characteristic frequency and amplitude of the reducer for the second time according to the above method. At this time, u and v involved in the above method should be set to 4 and 2 respectively, and the original characteristic frequency and amplitude of the reducer extracted before and after demodulation are compared to determine whether there is an equivalent relationship.

[0097]

[0098] Assume that the original characteristic frequencies of the reducer extracted from the current signal and the demodulated signal are stored in arrays P and Q respectively, f n_p and f n_q are the pth and qth characteristic frequencies in P and Q respectively, and the corresponding amplitude is A n_p and A n_q , if the following conditions are met, it is considered that the characteristic frequencies extracted from the current signal and the demodulated signal have a common frequency,

[0099] f n_p =f n_q (20)

[0100] |A n_p -A n_q |≤A th (twenty one)

[0101] Then the final reducer original characteristic frequency f n and amplitude A n Determine according to the following formula:

[0102] f n_t =f n_p (twenty two)

[0103]

[0104] After the original characteristic frequency and amplitude of the reducer are determined, the original characteristic frequency and amplitude are subjected to inverse Fourier transform to obtain the time domain waveform of the reducer fault current signal with the speed modulation interference eliminated.

[0105] The specific process is as follows Figure 1 shown.

[0106] Based on this, the present invention proposes a method for preprocessing fault data of a joint reducer for an industrial robot, and the preprocessing process thereof is performed according to the following steps.

[0107] Step 1,

[0108] Perform spectrum analysis on the current signal y and extract the position of the frequency corresponding to the maximum amplitude in the spectrum diagram. This frequency is the current frequency modulation frequency f c * .

[0109] Step 2,

[0110] Set a certain range of tolerance And assign relevant parameters u=2, v=1, w=1, i=1, and calculate the maximum value i of the loop parameter max =N, w is used to accumulate the number of original fault features, i is the search parameter, according to the modulation and demodulation relationship, it is assumed that u is the multiple relationship between the modulated signal feature and the feature amplitude of the original signal, v is a pair of modulation feature frequency and the difference between the original signal feature frequency and the current frequency modulation frequency f c * The multiple relationship between them.

[0111] Step 3,

[0112] Determine the modulation frequency f i * The corresponding amplitude Is it satisfied? If yes, it is considered that the frequency is formed by noise interference, and go directly to step 10; if no, go to step 4.

[0113] Step 4,

[0114] According to formula (4), And judge f i * , Does the corresponding amplitude satisfy the condition of formula (5)? ​​If yes, go to step 5; if not, go to step 6.

[0115] Step 5,

[0116] Calculate the characteristic frequency f according to formula (6) and formula (7): n_w and the amplitude A n_w , let, w=w+1, and go to step 10.

[0117] Step 6,

[0118] According to formula (8) And judge f i * , and Whether the corresponding amplitude satisfies the condition of formula (9), if yes, go to step 7, otherwise, go to step 8.

[0119] Step 7,

[0120] Calculate the characteristic frequency f according to formula (6) and formula (10) to (12): n_w and f n_w+1 and the amplitude A n_w and A n_w+1 , let w=w+2, and go to step 10.

[0121] Step 8,

[0122] According to formula (13), f is calculated l * , and determine the modulation frequency f i * , f l * And whether the corresponding amplitude satisfies the condition of formula (14), if yes, go to step 9, otherwise, go to step 10.

[0123] Step 9,

[0124] Calculate the characteristic frequency f according to formula (6), formula (8) and formula (15) to (18): n_w 、f n_w+1 and f n_w+2 and the amplitude A n_w , A n_w+1 and A n_w+2 , and execute w=w+3 and then go to step 10.

[0125] Step 10,

[0126] Determine whether the search parameter i reaches the maximum value i max If yes, set i=i+1 and return to step 3; otherwise, go to step 11.

[0127] Step 11,

[0128] Determine whether the current signal is demodulated based on whether v is equal to 1. If yes, go to step 12. If not, demodulate the current signal according to formula (19) to obtain the demodulated signal y*, and perform spectrum analysis on it. According to the demodulation frequency in the demodulated signal and the current frequency modulation frequency , adjust the parameters u and v to twice their original values ​​respectively, and return to step 3 to extract the original characteristic frequency of the reducer under the modulated signal.

[0129] Step 12,

[0130] Judgement n_p and f n_q and the amplitude A n_p and A n_q Do the conditions of formula (20) and formula (21) meet at the same time? If yes, it is considered that there is a pair of common characteristic frequencies f in the modulated signal and the demodulated signal. n_t , go to step 13 to calculate, otherwise, go to step 14.

[0131] Step 13,

[0132] Calculate the common characteristic frequency f according to formula (22) and formula (23): n_t The corresponding amplitude is A n_t , where t is used to accumulate f w_t The number of characteristic frequencies with the same internal frequency and very close amplitude difference, let t=t+1, and go to step 14.

[0133] Some fault features are found in both the modulated signal and the demodulated signal, but there are common features and non-common features in the modulated signal and the demodulated signal, which are stored in f n_w In f, as long as the common features are n_w There are two identical values ​​in f. Assume that 5 features are extracted from the modulated signal and 4 features are extracted from the demodulated signal. These extracted features are stored in f n_w So f n_w There are 9 characteristic frequencies in the example, but these 9 characteristic frequencies are repeated and need further comparison, so the f n_p and f n_q , f n_p and f n_q Represents f n_w The two different frequencies in the equation are compared and screened one by one, and the repetition frequency found is counted into f n_t , is the final required frequency. So f n_t and f n_w There is a necessary relationship that f n_t <=f n_w / 2, so the maximum number of t in the above example will not exceed 4.

[0134] Step 14,

[0135] Determine whether q is equal to w. If yes, go to step 15. If not, perform q=q+1 and then return to step 12.

[0136] Step 15,

[0137] Execute p=p+1 to prevent the common characteristic frequency f n_t The same frequency is screened twice to determine fn_p and f n_p-1 Does it satisfy the following relationship?

[0138] f n_p <f n_p-1 (twenty four)

[0139]

[0140] If yes, go to the next step; if no, return to step 13.

[0141] Step 16,

[0142] After filtering, f n_t and A n_t Perform inverse Fourier transform to obtain the current signal that is not disturbed by speed modulation.

[0143] Those skilled in the art should understand that the discussion of any of the above embodiments is only exemplary and is not intended to imply that the scope of the present invention is limited to these examples; under the concept of the present invention, the technical features in the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other changes in different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity. Any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for preprocessing fault data of a joint reducer for an industrial robot, characterized in that: The method comprises: S1. Perform spectrum analysis on the current signal, extract the position of the frequency corresponding to the maximum amplitude in the spectrum diagram, and obtain the current frequency modulation frequency S2. Set a certain range of allowable error And assign relevant parameters u=2, v=1, w=1, i=1. According to the allowable error and current frequency modulation frequency, calculate the maximum value i of the cycle parameter that meets the allowable error. max , where w is used to accumulate the number of original fault features, i is the search parameter, and according to the modulation and demodulation relationship, it is assumed that u is the multiple relationship between the modulated signal feature and the feature amplitude of the original signal, and v is the difference between a pair of modulation feature frequencies and the original signal feature frequencies and the current frequency modulation frequency. The multiple relationship between them makes: f n is the characteristic frequency of the reducer fault, f * is the frequency after the characteristic frequency of the reducer fault is modulated; S3, determine the modulation frequency f i * The corresponding amplitude Is it satisfied? If it is satisfied, it is considered that the frequency is caused by noise interference, and the process goes to step S10. If it is not satisfied, the process goes to step S4. i * With another modulation frequency are the original characteristic frequencies of the reducer f n1 The corresponding modulation frequency pairs; S4, according to the modulation frequency f i * and current modulation frequency calculate Judgement i * and The corresponding amplitude and Whether the following conditions are met, if so, execute step S5, otherwise jump to step S6; S5. Calculate characteristic frequency f n_w And the corresponding amplitude A n_w , let w=w+1, jump to step S10; S6. Calculate the modulation frequency Judgement i * , and Whether the corresponding amplitude satisfies the conditions of the following formula, if so, execute step S7, otherwise jump to step S8, where: and is the original characteristic frequency f of the reducer n2 The corresponding modulation frequency pair is: S7. Calculate characteristic frequency f n_w and f n_w+1 And the corresponding amplitude A n_w and A n_w+1 , let w=w+2, go to step 10; S8. Calculate f l * , and determine the modulation frequency f i * , f l * And whether the corresponding amplitude satisfies the conditions of the following formula, if so, go to step S9, otherwise go to step S10, where and f l * is the original characteristic frequency f of the reducer n3 The corresponding modulation frequency pairs; S9. Calculate characteristic frequency f n_w 、f n_w+1 and f n_w+2 And the corresponding amplitude A n_w , A n_w+1 and A n_w+2 , and execute w=w+3 and then go to step S10; S10, determine whether the search parameter i reaches the maximum value i max If yes, set i=i+1 and return to step S3; otherwise, go to step S11; S11, judging whether the current signal is demodulated according to whether v is equal to 1, if yes, go to step S12, otherwise demodulate the current signal to obtain a demodulated signal y*, and perform spectrum analysis on it, according to the demodulation frequency in the demodulated signal and the current frequency modulation frequency , respectively adjust the parameters u and v to twice their original values, and return to step S3 to extract the original characteristic frequency of the reducer under the modulation signal; S12, judge f n_p and f n_q and the amplitude A n_p and A n_q Does it satisfy the following formula? If yes, it is considered that there is a pair of common characteristic frequencies f in the modulated signal and the demodulated signal. n_t , go to step S13 to calculate, otherwise go to step S14: f n_p =f n_q |A n_p -A n_q |≤A th ; Among them, the original characteristic frequencies of the reducer extracted from the current signal and the demodulated signal are stored in arrays P and Q respectively, f n_p and f n_q are the pth and qth characteristic frequencies in P and Q respectively, and the corresponding amplitude is A n_p and A n_q ; S13, calculate the common characteristic frequency f n_t The corresponding amplitude is A n_t , t is used to accumulate f w_t The number of characteristic frequencies with the same internal frequency and very close amplitude difference, let t = t + 1, and go to step S14; S14, determine whether q is equal to w, if yes, go to step S15, otherwise, perform q=q+1 and return to step S12; S15, execute p=p+1, judge f n_p and f n_p-1 Does it satisfy the following relationship? f n_p <f n_p-1 If yes, proceed to the next step, otherwise return to step S13; S16, the filtered f n_t and A n_t Perform inverse Fourier transform to obtain the current signal that is not disturbed by speed modulation.

2. The fault data preprocessing method for a joint reducer for an industrial robot according to claim 1, characterized in that: In step S4, calculate The formula is:

3. The fault data preprocessing method for a joint reducer for an industrial robot according to claim 1, characterized in that: In step S5, the characteristic frequency f is calculated n_w and amplitude A n_w The formula is:

4. The fault data preprocessing method for a joint reducer for an industrial robot according to claim 1, characterized in that: In step S6, calculate The formula is:

5. The fault data preprocessing method for a joint reducer for an industrial robot according to claim 1, characterized in that: In step S7, the characteristic frequency f is calculated n_w and f n_w+1 and the amplitude A n_w and A n_w+1 The formula is:

6. The fault data preprocessing method for a joint reducer for an industrial robot according to claim 1, characterized in that: In step S8, f is calculated l * The formula is:

7. The fault data preprocessing method for a joint reducer for an industrial robot according to claim 1, characterized in that: In step S9, the characteristic frequency f is calculated n_w 、f n_w+1 and f n_w+2 and the amplitude A n_w , A n_w+1 and A n_w+2 The formula is:

8. The fault data preprocessing method for a joint reducer for an industrial robot according to claim 1, characterized in that: In step S11, the formula for calculating the demodulated signal y* is: Where n is the number of sampling points, y is the modulation signal, and f s is the sampling rate.

9. The fault data preprocessing method for a joint reducer for an industrial robot according to claim 1, characterized in that: In step S13, the common characteristic frequency f is calculated n_t and the corresponding amplitude A n_t The formula is: f n_t =f n_p

Images