Reversing box off-circle fault identification method based on fusion index

Abstract

The invention, which relates to the technical field of fault diagnosis, discloses a fusion index-based commutation box off-circle fault identification method comprising the following identification steps: S1, arranging a vibration acceleration sensor on a commutation box bearing pedestal to collect vibration acceleration data; s2, analyzing vibration acceleration data by adopting EEMD, averaging the obtained I MF, and reducing white noise components in the data; and S3, carrying out an SE-SS method on the average intrinsic mode function, and enhancing the characteristics of the off-circle fault in the signal. According to the invention, through a data cleaning method, white noise components in collected signals are reduced to the maximum extent by EEMD; an SE-SS method is adopted to enhance the playing circle fault features in the average intrinsic mode function, the playing circle fault can be monitored in real time, an alarm can be given in advance in the early development stage of the playing circle fault, the features of the playing circle fault in the signal can be enhanced by using the method provided by the invention, the fault features can be conveniently identified, and the detection accuracy is improved. The application requirements of the reversing box of the rail engineering vehicle for early diagnosis and alarm of the off-road fault are met.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a method for identifying a reversing box ring fault based on a fusion index. Background Art

[0002] The reversing box is one of the core components of a mechanically driven rail engineering vehicle. It connects the engine and the wheelset assembly, and can not only transmit motion but also carry a large amount of load. During the use of the reversing box, the ring-playing fault is a very difficult fault to diagnose. The specific manifestation is that the size matching between the inner ring of the bearing and the shaft is inappropriate, which causes the inner ring of the bearing to slip during use, and then causes dry friction between the inner ring of the bearing and the shaft. This fault manifests as intermittent slippage in the early stage, which is difficult to detect using vibration and temperature signals. In the middle and late stages of the fault, it will cause the bearing and shaft to heat up rapidly in a very short time. If it continues to run, there is a risk of shaft melting; if it is stopped for maintenance, the inner ring of the bearing and the shaft will be welded during cooling, resulting in the unfavorable consequences of the bearing being unable to be disassembled and the shaft being severely deformed. Therefore, the fault must be identified in the early stage of the fault. To this end, we propose a reversing box ring-playing fault identification method based on fusion indicators. Summary of the Invention

[0003] In order to solve the above technical problems, a method for identifying the ring fault of the reversing box bearing based on fusion indicators is provided. This technical solution can realize the efficient early identification of the ring fault of the reversing box bearing in the rail engineering vehicle.

[0004] To achieve the above objectives, the present invention adopts a technical solution: a method for identifying a reversing gear ring fault based on a fusion index, wherein the identification steps include:

[0005] S1, the vibration acceleration sensor is set on the reversing box bearing seat to collect vibration acceleration data;

[0006] S2. Analyze the vibration acceleration data using EEMD and average the obtained IMFs to reduce the white noise component in the data.

[0007] S3. Apply SE-SS method to the average intrinsic mode function to enhance the characteristics of looping fault in the signal;

[0008] S4. Construct a fusion index that can distinguish looping faults and extract fusion features;

[0009] S5. Alarm the loop fault based on the extracted fusion feature index.

[0010] Preferably, in step S1, a vibration acceleration sensor is installed on the reversing box bearing seat. Based on the principle of piezoelectric effect, the internal piezoelectric crystal generates a charge signal under the action of vibration acceleration, and collects vibration acceleration data in real time. The data includes vibration frequency, amplitude and change trend information.

[0011] Preferably, in step S2, EEMD is used to analyze the vibration acceleration data, and the obtained IMF is averaged to reduce the white noise component in the data. The specific steps are:

[0012] A1. Add a random white noise signal to the original signal sequence x(t). The expression is:

[0013] Ex(t)=x(t)+N(t)

[0014] Where N(t) represents the random white noise signal sequence, and x(t) represents the original signal sequence;

[0015] A2. Decompose Ex(t) into IMF components Ec according to the calculation method j (t) and a residual component Er j (t), the expression is:

[0016]

[0017] A3. Repeat steps A1 and A2, adding a new random white noise signal sequence with the same amplitude in each calculation to obtain different groups of IMF components and residual components;

[0018] A4. Eliminate the effect of random white noise in the IMF component and average the IMF components obtained from each decomposition to obtain the final EEMD decomposition result, which is expressed as:

[0019]

[0020] Where n is the number of ensemble averages in the EEMD decomposition process.

[0021] Preferably, in step S3, the SE-SS method is performed on the average intrinsic mode function to enhance the characteristics of the looping fault in the signal. Specifically, the steps include:

[0022] The double envelope transformation of the average IMF target air volume EEc(t) is expressed as follows:

[0023] SES(t)=|EA(EEc(t)) 2 |

[0024] Perform linear interpolation on SES(t) after double envelope transformation to obtain J data sets of the same length. The data set is processed for spectrum synchronization, and the processing formula is:

[0025]

[0026] in is the spectrum synchronization value of the sequence.

[0027] Preferably, Expressed as the Fourier transform of the dataset, it is expressed as:

[0028]

[0029] Among them F j is the spectrum of the jth data set after Fourier transform, N is the total amount of data in the jth data set, and fs is the sampling frequency of the data set in Hz;

[0030] In each selected local frequency band, the theoretically calculated fault frequency component fc is located at the center of the frequency band. The appropriate amplitude of the fault frequency component fc is selected, and the frequency bands corresponding to the frequencies at different positions in the spectrum are averaged. The spectrum is then transformed from the frequency domain to the frequency domain. Convert to discrete points For subsequent analysis, each frequency f is represented by its nearest discrete point d. Then, the fault characteristic frequency fc(k) is converted to a discrete point dc(k), and the corresponding frequency value is the value closest to fc, where k = 1, 2, ..., K, and K is the total number of fault characteristic frequencies counted. During the analysis process, the width of the local frequency band is set to the same to facilitate the spectrum synchronization calculation operation. Given the bandwidth in the local frequency band is bw, the width dw of the local frequency band of the discrete point after conversion is expressed as:

[0031]

[0032] Where R(·) represents the rounding operation.

[0033] Preferably, the kth local frequency band ψ in the discrete points k Expressed as:

[0034]

[0035] where ψ i,k represents the kth local segment ψ k The i-th discrete point in , i = 1, 2, ..., d w+1 ;

[0036] In the kth local segment ψ k According to the discrete points {ψ i ,K} are sorted in descending order according to their values, and a new sequence π is obtained. i,k ;

[0037] The fault center point value on the frequency band is expressed as:

[0038]

[0039] Then the discrete point values on the frequency band except the fault center point can be expressed as:

[0040]

[0041] where ξ{·} represents the median calculation.

[0042] Preferably, the first 50% of the high-amplitude fault characteristic frequency component components or the center frequency component components in the local frequency band are averaged to enhance the fault characteristics thereof, further suppress other frequency component components in the local frequency band by eliminating occasional protruding peaks, and reduce the amplitude of abnormal values;

[0043] The relative amplitude of the center frequency component can be expressed as:

[0044] υ s =g s -ξ{g}

[0045] Among them, g s ={g i}i=d w / 2+1 is the central discrete point there.

[0046] Preferably, in step S4, a fusion index that can distinguish the looping fault is constructed, and the fusion feature extraction step is performed by finding the characteristic frequency of the looping fault. s Find the fault characteristic frequency f d The corresponding amplitude A d , taking into account the difference between the theoretical calculated value and the measured value of the fault characteristic frequency, f d The error of ±1% around is the search area, and the maximum value found is A d , and calculate the average amplitude value in the area, which is A m , the expression is:

[0047] A d =max{A|0.99f s ≤f s ≤1.01f s}

[0048] A m =mean{A|0.99f s ≤f s ≤1.01f s}

[0049] The fusion index is A d and A m The ratio is expressed as:

[0050] HI=A d / A m

[0051] Use the HI indicator to identify the reversing box ring fault.

[0052] Preferably, the fusion characteristic indicators in step S5 are obtained by analyzing and comprehensively processing multi-source data during the operation of the equipment. When the equipment is running, the numerical changes of these fusion characteristic indicators are monitored in real time. When one or some characteristic indicators deviate from the range of normal operation and reach the set alarm threshold, an audible and visual alarm is issued.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] The present invention uses a data cleaning method to reduce the white noise component in the collected signal by EEMD to the greatest extent; adopts the SE-SS method to enhance the loop fault characteristics in the average intrinsic mode function; and performs fault identification by constructing a fusion index that can distinguish loop faults. The loop fault can be monitored in real time and an early alarm can be issued in its early development. The proposed method can enhance the loop fault characteristics in the signal, facilitate the identification of fault characteristics, and solve the application demand of the rail engineering vehicle reversing box for early diagnosis and alarm of loop faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of a method for identifying a reversing box ring fault based on fusion indicators provided by the present invention;

[0056] Figure 2 This is an original envelope spectrum of the vibration signal of the reversing box ring provided by the present invention;

[0057] Figure 3 This is an IMF envelope spectrum of the vibration signal of the reversing box ring provided by the present invention;

[0058] Figure 4 This is a SE-SS envelope spectrum of the IMF of the vibration signal of the reversing box ring provided by the present invention;

[0059] Figure 5 This is a SE-SS envelope spectrum diagram of a normal reversing box vibration signal IMF provided by the present invention. DETAILED DESCRIPTION

[0060] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0061] Reference Figure 1 As shown in FIG, a method for identifying a reversing box ring fault based on a fusion index, the identification steps include:

[0062] S1, the vibration acceleration sensor is set on the reversing box bearing seat to collect vibration acceleration data;

[0063] S2. Analyze the vibration acceleration data using EEMD and average the obtained IMFs to reduce the white noise component in the data.

[0064] S3. Apply SE-SS method to the average intrinsic mode function to enhance the characteristics of looping fault in the signal;

[0065] S4. Construct a fusion index that can distinguish looping faults and extract fusion features;

[0066] S5. Alarm the loop fault based on the extracted fusion feature index.

[0067] This application installs the vibration acceleration sensor on the reversing box bearing seat, which can directly obtain the vibration condition of the bearing seat. Because the ring failure will directly cause the vibration characteristics of the bearing seat to change, this installation position can accurately capture the vibration information related to the ring failure, providing a reliable data basis for subsequent fault diagnosis; by collecting vibration acceleration data in real time, the operating status of the reversing box can be continuously monitored. Once abnormal vibration occurs in the equipment, it can be discovered and analyzed in time to avoid further deterioration of the fault and ensure the stable operation of the equipment;

[0068] EEMD can decompose complex vibration signals into multiple intrinsic mode functions (IMFs) and average these IMFs, effectively reducing the white noise component in the data. The data after noise removal is purer, making subsequent fault feature extraction more accurate and improving the reliability of fault diagnosis. White noise can mask the true vibration signal characteristics. By reducing the white noise component, the vibration characteristics related to the hoop fault can be highlighted, making these characteristics easier to detect and analyze.

[0069] The SE-SS method can further enhance the features related to the looping fault in the signal. Although the data processed by EEMD has removed the noise, the fault features may still not be obvious enough. The SE-SS method can make the looping fault features more prominent, improve the recognition of the fault features, and facilitate subsequent fault diagnosis.

[0070] The enhanced fault signature enables the system to more sensitively detect early signs of a power failure, even if the fault is relatively minor. This improves the sensitivity of fault diagnosis, enables early warning, and buys more time for equipment maintenance.

[0071] The construction of fusion index integrates characteristic information from multiple aspects, not just relying on a single signal feature. By fusing different features, it can more comprehensively describe the characteristics of the looping fault and improve the accuracy of fault diagnosis. Because looping faults may exhibit different characteristics in multiple aspects, combining these characteristics can more accurately determine the existence and extent of the fault.

[0072] Accurate fault alarms can ensure the safe operation of the equipment, avoid equipment accidents caused by ring failures, ensure the safety of the production process, and protect the safety of personnel and equipment.

[0073] In step S1, the vibration acceleration sensor is installed on the bearing seat of the reversing box. Based on the principle of piezoelectric effect, the internal piezoelectric crystal generates a charge signal under the action of vibration acceleration, and the vibration acceleration data is collected in real time. The data includes vibration frequency, amplitude and change trend information.

[0074] The reversing box bearing seat of this application is a key part that bears load and transmits power during equipment operation. Abnormal conditions of the ring failure will directly cause changes in the vibration characteristics of the bearing seat. Installing the sensor here can directly obtain key information closely related to the operating status of the reversing box, providing the most direct and effective data source for subsequent fault diagnosis, avoiding information deviation or loss caused by indirect measurement;

[0075] The collected data includes multi-dimensional information such as vibration frequency, amplitude, and trend. Frequency reflects the periodic nature of vibration. Different fault types often correspond to specific frequency components, and frequency analysis can provide a preliminary assessment of the nature of the fault. Amplitude reflects the intensity of the vibration, and changes in amplitude can intuitively reflect the severity of the fault. Trend information can demonstrate the evolution of vibration status over time, helping to predict the direction of fault development.

[0076] In step S2, EEMD is used to analyze the vibration acceleration data, and the obtained IMF is averaged to reduce the white noise component in the data. The specific steps are:

[0077] A1. Add a random white noise signal to the original signal sequence x(t). The expression is:

[0078] Ex(t)=x(t)+N(t)

[0079] Where N(t) represents the random white noise signal sequence, and x(t) represents the original signal sequence;

[0080] A2. Decompose Ex(t) into IMF components Ec according to the calculation method j (t) and a residual component Er j (t), the expression is:

[0081]

[0082] A3. Repeat steps A1 and A2, adding a new random white noise signal sequence with the same amplitude in each calculation to obtain different groups of IMF components and residual components;

[0083] A4. Eliminate the effect of random white noise in the IMF component and average the IMF components obtained from each decomposition to obtain the final EEMD decomposition result, which is expressed as:

[0084]

[0085] Where n is the number of ensemble averages in the EEMD decomposition process.

[0086] In the S2 step of this application, EEMD is used to analyze the vibration acceleration data and average the IMF to reduce noise. The addition of white noise can suppress modal aliasing, separate different vibration modes, and facilitate accurate analysis of fault characteristics. Multiple decompositions and averaging improve the decomposition accuracy and reduce the uncertainty error of a single decomposition. At the same time, the randomness of white noise is utilized to offset each other during averaging, reducing its impact on the results and making the IMF component purer. This series of operations enhances the reliability of the analysis of vibration acceleration data, can accurately extract fault characteristics, provide strong support for fault diagnosis and early warning, and reduce equipment failure losses.

[0087] In step S3, the SE-SS method is applied to the average intrinsic mode function to enhance the characteristics of the looping fault in the signal. The specific steps include:

[0088] The double envelope transformation of the average IMF target air volume EEc(t) is expressed as follows:

[0089] SES(t)=|EA(EEc(t)) 2 |

[0090] Perform linear interpolation on SES(t) after double envelope transformation to obtain J data sets of the same length. The data set is processed for spectrum synchronization, and the processing formula is:

[0091]

[0092] in is the spectrum synchronization value of the sequence.

[0093] in Expressed as the Fourier transform of the dataset, it is expressed as:

[0094]

[0095] Among them F j is the spectrum of the jth data set after Fourier transform, N is the total amount of data in the jth data set, and fs is the sampling frequency of the data set in Hz;

[0096] In each selected local frequency band, the theoretically calculated fault frequency component fc is located at the center of the frequency band. The appropriate amplitude of the fault frequency component fc is selected, and the frequency bands corresponding to the frequencies at different positions in the spectrum are averaged. The spectrum is then transformed from the frequency domain to the frequency domain. Convert to discrete points For subsequent analysis, each frequency f is represented by its nearest discrete point d. Then, the fault characteristic frequency fc(k) is converted to a discrete point dc(k), and the corresponding frequency value is the value closest to fc, where k = 1, 2, ..., K, and K is the total number of fault characteristic frequencies counted. During the analysis process, the width of the local frequency band is set to the same to facilitate the spectrum synchronization calculation operation. Given the bandwidth in the local frequency band is bw, the width dw of the local frequency band of the discrete point after conversion is expressed as:

[0097]

[0098] Where R(·) represents the rounding operation.

[0099] The kth local frequency band ψ in the discrete point k Expressed as:

[0100]

[0101] where ψ i,k represents the kth local segment ψ k The i-th discrete point in , i = 1, 2, ..., d w+1 ;

[0102] In the kth local segment ψ k According to the discrete points {ψ i ,K} are sorted in descending order according to their values, and a new sequence π is obtained. i,k ;

[0103] The fault center point value on the frequency band is expressed as:

[0104]

[0105] Then the discrete point values on the frequency band except the fault center point can be expressed as:

[0106]

[0107] where ξ{·} represents the median calculation.

[0108] The first 50% of high-amplitude fault characteristic frequency components or center frequency components in the local frequency band are averaged to enhance their fault characteristics. By eliminating occasional protruding peaks, other frequency components in the local frequency band are further suppressed and the amplitude of abnormal values is reduced.

[0109] The relative amplitude of the center frequency component can be expressed as:

[0110] υ s =g s -ξ{g}

[0111] Among them, g s ={g i}i=d w / 2+1 is the central discrete point there.

[0112] The double envelope transform of this application can further extract key information from the signal, emphasize the characteristic components related to the looping fault, and make the fault characteristics more prominent in the signal. Linear interpolation and spectrum synchronization processing regularize the data, facilitate subsequent analysis, and ensure that different data sets are compared at the same scale, so as to more accurately capture the fault frequency-related information. By calculating the fault frequency component components in the selected local frequency band and converting the spectrum into discrete points, the fault characteristic frequency can be accurately located. Setting the same local frequency band width facilitates spectrum synchronization calculation, making the analysis process more systematic and comparable, improving the accuracy of fault frequency identification, and providing a reliable basis for fault diagnosis. The discrete points are sorted and the fault center point value and other discrete point values are calculated. Then, the high-amplitude fault characteristic frequency component components in the local frequency band are averaged, effectively eliminating accidental protruding peaks, suppressing other frequency component components, and reducing the amplitude of outliers. This greatly weakens the interference components in the signal, highlights the true fault characteristics, and reduces the possibility of misjudgment. The relative amplitude of the center frequency component components is calculated, further enhancing the recognition of the fault characteristics. After this series of processing, the characteristics of the ring-flipping fault are more obvious, which facilitates subsequent fault judgment and early warning based on these characteristics, improves the detection capability and reliability of the entire fault diagnosis system for the ring-flipping fault, and helps to detect equipment failures in a timely manner and take measures.

[0113] In step S4, a fusion index that can distinguish the looping fault is constructed, and the fusion feature extraction step is performed by finding the characteristic frequency of the looping fault. s Find the fault characteristic frequency f d The corresponding amplitude A d , taking into account the difference between the theoretical calculated value and the measured value of the fault characteristic frequency, f d The error of ±1% around is the search area, and the maximum value found is A d , and calculate the average amplitude value in the area, which is Am , the expression is:

[0114] A d =max{A|0.99f s ≤f s ≤1.01f s}

[0115] A m =mean{A|0.99f s ≤f s ≤1.01f s}

[0116] The fusion index is A d and A m The ratio is expressed as:

[0117] HI=A d / A m

[0118] Use the HI indicator to identify the reversing box ring fault.

[0119] In step S5, the fusion characteristic indicators are obtained by analyzing and comprehensively processing multi-source data during the operation of the equipment. When the equipment is running, the numerical changes of these fusion characteristic indicators are monitored in real time. When one or some characteristic indicators deviate from the range of normal operation and reach the set alarm threshold, an audible and visual alarm is issued.

[0120] In step S5 of the present application, the fusion characteristic indicators are monitored in real time. Once one or some characteristic indicators deviate from the normal range and reach the alarm threshold, an audible and visual alarm is issued. This real-time monitoring function can timely detect changes in the operating status of the equipment and issue an early warning in the early stage of the ring fault, so that the operator can take quick measures to avoid further deterioration of the fault and reduce equipment damage and production losses. The fusion characteristic indicator is obtained by analyzing and comprehensively processing multi-source data during the operation of the equipment. It integrates information from many aspects and can more comprehensively reflect the operating status of the equipment. The comprehensive utilization of multi-source data makes up for the limitations of a single data source, provides clues about the ring fault from different angles, further improves the accuracy and reliability of fault diagnosis, and provides a more scientific basis for equipment maintenance and management. The audible and visual alarm method is simple and intuitive, and the operator can quickly detect abnormal conditions of the equipment, which facilitates timely inspection and processing. At the same time, the fault diagnosis system based on fusion indicators has clear judgment criteria and operating procedures, which reduces the difficulty of fault judgment for operators, improves the efficiency of equipment maintenance, and makes the operation and management of the equipment more standardized and intelligent.

[0121] Example 1

[0122] A certain type of reversing box experienced repeated ring-playing failures. To identify the failures and provide early warning, two normal reversing boxes and three reversing boxes at risk of ring-playing failures were tested. The reversing boxes were numbered 1-5, with 1-2 being normal reversing boxes and 3-5 being ring-playing reversing boxes.

[0123] Taking the No. 3 reversing box as an example, EEMD is used to decompose its signal and eliminate the white noise component to the greatest extent. Figure 2 is the envelope spectrum of the vibration signal before EEMD, Figure 3 This is the average IMF envelope spectrum after EEMD, compared with Figure 2 and Figure 3 It can be seen that the white noise component in the signal is significantly reduced. Since the ringing fault occurs on axis 3 and the speed sensor is installed on axis 1, the speed of axis 3 is 1.15 times that of axis 1. The amplitude corresponding to the fault coefficient 1.15 needs to be found. The frequency coefficient close to 1.137 is found, and the corresponding amplitude is 0.2928, which is not very prominent. Then the SE-SS method is used to obtain Figure 4 As a result, the amplitude reaches 0.3806, which is quite prominent, but the value itself is still not large enough. Figure 4 In contrast, the same method was used to process the No. 2 reversing box. Figure 5 As shown in Figure 1, its amplitude at the position of 1.137 is 0.4504. Although it is not prominent, the value is large. In order to more accurately identify the ring-playing fault, the HI index is constructed. As shown in Table 1, the HI index can accurately identify the ring-playing fault and is not affected by the individual differences of the commutation box.

[0124] Table 1 below is a comparison table of amplitude and HI of the present invention

[0125]

[0126] Table 1

[0127] Example 2

[0128] Another model of reversing box frequently experienced looping failures. To effectively identify this failure and issue early warnings, two reversing boxes with normal performance and three reversing boxes with potential looping failures were selected for testing. These reversing boxes were numbered 6-10, with No. 6-7 being normal reversing boxes and No. 8-10 being those with looping problems.

[0129] Taking the No. 8 reversing box as an example, the EEMD technology is used to decompose the collected signal to remove the white noise component to the greatest extent possible. In order to more accurately identify the ring-playing fault, the HI index is constructed. The specific situation is shown in Table 2. With the help of the HI index, the ring-playing fault can be accurately identified and will not be affected by the individual differences of the reversing box.

[0130] Table 2 below is a comparison table of amplitude and HI of the present invention

[0131]

[0132] Table 2

[0133] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. Various changes and improvements are possible without departing from the spirit and scope of the present invention, and such changes and improvements fall within the scope of the invention as claimed.

Claims

1. A method for identifying reversing gear ring faults based on fusion indicators, characterized in that: The identification steps include: S1, the vibration acceleration sensor is set on the reversing box bearing seat to collect vibration acceleration data; S2. Analyze the vibration acceleration data using EEMD and average the obtained IMFs to reduce the white noise component in the data. S3. Apply SE-SS method to the average intrinsic mode function to enhance the characteristics of looping fault in the signal; S4. Construct a fusion index that can distinguish looping faults and extract fusion features; S5. Alarm the loop fault based on the extracted fusion feature index.

2. The method for identifying a reversing gear ring fault based on fusion indicators according to claim 1 is characterized in that: In step S1, the vibration acceleration sensor is installed on the bearing seat of the reversing box. Based on the principle of piezoelectric effect, the internal piezoelectric crystal generates a charge signal under the action of vibration acceleration, and the vibration acceleration data is collected in real time. The data includes vibration frequency, amplitude and change trend information.

3. The method for identifying a reversing gear ring fault based on fusion indicators according to claim 1 is characterized in that: In step S2, EEMD is used to analyze the vibration acceleration data, and the obtained IMF is averaged to reduce the white noise component in the data. The specific steps are as follows: A1. Add a random white noise signal to the original signal sequence x(t). The expression is: Ex(t)=x(t)+N(t) Where N(t) represents the random white noise signal sequence, and x(t) represents the original signal sequence; A2. Decompose Ex(t) into IMF components Ec according to the calculation method j (t) and a residual component Er j (t), the expression is: A3. Repeat steps A1 and A2, adding a new random white noise signal sequence with the same amplitude in each calculation to obtain different groups of IMF components and residual components; A4. Eliminate the effect of random white noise in the IMF component and average the IMF components obtained from each decomposition to obtain the final EEMD decomposition result, which is expressed as: Where n is the number of ensemble averages in the EEMD decomposition process.

4. The method for identifying a reversing gear ring fault based on fusion indicators according to claim 1 is characterized in that: In step S3, the SE-SS method is applied to the average intrinsic mode function to enhance the characteristics of the looping fault in the signal. The specific steps include: The double envelope transformation of the average IMF target air volume EEc(t) is expressed as follows: SES(t)=|EA(EEc(t)) 2 | Perform linear interpolation on SES(t) after double envelope transformation to obtain J data sets of the same length. The data set is processed for spectrum synchronization, and the processing formula is: in is the spectrum synchronization value of the sequence.

5. The method for identifying a reversing gear ring fault based on fusion indicators according to claim 4 is characterized in that: in Expressed as the Fourier transform of the dataset, it is expressed as: Among them F j is the spectrum of the jth data set after Fourier transform, N is the total amount of data in the jth data set, and fs is the sampling frequency of the data set in Hz; In each selected local frequency band, the theoretically calculated fault frequency component fc is located at the center of the frequency band. The appropriate amplitude of the fault frequency component fc is selected, and the frequency bands corresponding to the frequencies at different positions in the spectrum are averaged. The spectrum is then transformed from the frequency domain to the frequency domain. Convert to discrete points For subsequent analysis, each frequency f is represented by its nearest discrete point d. Then, the fault characteristic frequency fc(k) is converted to a discrete point dc(k), and the corresponding frequency value is the value closest to fc, where k = 1, 2, ..., K, and K is the total number of fault characteristic frequencies counted. During the analysis process, the width of the local frequency band is set to the same to facilitate the spectrum synchronization calculation operation. Given the bandwidth in the local frequency band is bw, the width dw of the local frequency band of the discrete point after conversion is expressed as: Where R(·) represents the rounding operation.

6. The method for identifying a reversing gear ring fault based on fusion indicators according to claim 5 is characterized in that: The kth local frequency band ψ in the discrete point k Expressed as: where ψ i,k represents the kth local segment ψ k The i-th discrete point in , i = 1, 2, ..., d w+1 ; In the kth local segment ψ k According to the discrete points {ψ i ,K} are sorted in descending order according to their values, and a new sequence π is obtained. i,k ; The fault center point value on the frequency band is expressed as: Then the discrete point values on the frequency band except the fault center point can be expressed as: where ξ{·} represents the median calculation.

7. The method for identifying a reversing gear ring fault based on fusion indicators according to claim 6 is characterized in that: The first 50% of the high-amplitude fault characteristic frequency components or the center frequency components in the local frequency band are averaged to enhance the fault characteristics. By eliminating occasional protruding peaks, other frequency components in the local frequency band can be further suppressed and the amplitude of abnormal values can be reduced; The relative amplitude of the center frequency component can be expressed as: u s =g s -ξ{g} Among them, g s ={g i }i=d w / 2+1 is the central discrete point there.

8. The method for identifying a reversing gear ring fault based on fusion indicators according to claim 1 is characterized in that: In step S4, a fusion index that can distinguish the looping fault is constructed, and the fusion feature extraction step is performed by finding the characteristic frequency of the looping fault. s Find the fault characteristic frequency f d The corresponding amplitude A d , taking into account the difference between the theoretical calculated value and the measured value of the fault characteristic frequency, f d The error of ±1% around is the search area, and the maximum value found is A d , and calculate the average amplitude value in the area, which is A m , the expression is: A d =max{A|0.99f s ≤f s ≤1.01f s } Ad=max{A|0.99fs≤fs≤1.01fs} A m =mean{A|0.99f s ≤f s ≤1.01f s } The fusion index is A d and A m The ratio is expressed as: HI=A d / THE m Use the HI indicator to identify the reversing box ring fault.

9. The method for identifying a reversing gear ring fault based on fusion indicators according to claim 1 is characterized in that: In step S5, the fusion characteristic indicators are obtained by analyzing and comprehensively processing multi-source data during the operation of the equipment. When the equipment is running, the numerical changes of these fusion characteristic indicators are monitored in real time. When one or some characteristic indicators deviate from the range of normal operation and reach the set alarm threshold, an audible and visual alarm is issued.

Images