Bearing assembly and related machinery and methods
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- AB SKF SKF PATENT DEPARTMENT
- Filing Date
- 2024-10-23
- Publication Date
- 2026-05-26
Smart Images
Figure CN122095231A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a bearing assembly and a method for monitoring bearing assemblies in a machine.
[0002] More specifically, the present invention relates to the determination of bearing failure in bearing assemblies. Background Technology
[0003] Machines (such as trucks) consist of multiple axles.
[0004] Each axle is supported by a bearing.
[0005] The bearings are monitored to detect bearing failures.
[0006] Typically, each bearing is equipped with a vibration sensor, and the vibration sensor delivers a signal indicating the vibration of the bearing.
[0007] For example, Fast Fourier Analysis can be used to perform spectral analysis on a signal to obtain a spectrum, thereby identifying harmonics that represent defects in a bearing, such as defects on the raceway of the inner ring of the bearing.
[0008] However, some defects in the bearing cannot be identified from this spectrum.
[0009] It is also known to determine the root mean square of the set of vibration measurements contained in the signal.
[0010] However, the root mean square mass of the vibration measurement set is sensitive to changes in bearing speed, which may lead to false alarms. Summary of the Invention
[0011] Therefore, the present invention aims to improve the detection of bearing failures based on vibration measurement results.
[0012] According to one aspect, a method for monitoring a bearing assembly in a machine, the bearing assembly including a bearing and a vibration sensor for measuring the vibration of the bearing, the bearing being provided with an inner ring and an outer ring capable of rotating concentrically relative to each other.
[0013] The method includes the following steps:
[0014] a. Determine the value of at least one first statistical parameter equal to the Hjorth complexity parameter based on at least one set of vibration measurements delivered by the sensor during a training period after the sensor is implemented in the machine, and when the rotational speed of one of the inner and outer rings relative to the other ring is contained within a predetermined interval.
[0015] b. Model the values of at least the first statistical parameter using a normal distribution during the training period.
[0016] c. Determine the detection threshold based at least on the normal distribution during the training period.
[0017] d. Determine at least a first value for the first statistical parameter based on a set of vibration measurements delivered by the sensor during normal operation of the machine and when the rotational speed of one revolution is contained within the predetermined range.
[0018] e. Determine the score during normal operation of the machine based at least on a first value of the Hjorth complexity parameter and the normal distribution modeled on the value of the first statistical parameter.
[0019] f. Compare the score with the detection threshold, and
[0020] g. Determine the bearing failure based on the results of the comparison.
[0021] Since the determination of the Hjorth complexity parameter is insensitive to changes in the bearing's rotational speed, the first statistical parameter in the vibration measurement set that is equal to the Hjorth complexity parameter allows for accurate prediction of bearing damage by analyzing only the evolution of the first statistical parameter.
[0022] Preferably,
[0023] Step (a) further includes determining the value of at least a second statistical parameter based on the set of vibration measurements delivered by the sensor during the training period.
[0024] - Step (b) further includes modeling the value of the second statistical parameter using a normal distribution during the training period.
[0025] - Step (c) includes determining the detection threshold based on a normal distribution modeling the values of the first statistical parameter during the training period and a normal distribution modeling the values of the second statistical parameter.
[0026] Step (d) further includes determining a first value for the second statistical parameter based on a set of vibration measurements delivered by the sensor during normal operation, and
[0027] - Step (e) includes determining a score during normal operation of the machine based on a first value of the Hjorth complexity parameter, a first value of the second statistical parameter, a normal distribution modeled on the value of the first statistical parameter, and a normal distribution modeled on the value of the second statistical parameter.
[0028] To increase diagnostic reliability, as shown above, several statistical parameters (one of which is equal to the Hjorth complexity parameter) are determined based on the recorded vibration measurement results of bearing 4. By analyzing the evolution of these parameters, early prediction of damage to bearing 4 can be made.
[0029] By using several statistical parameters to predict bearing failures, different types of defects in bearings can be detected.
[0030] Furthermore, the detection threshold is determined based on vibration measurements of the bearings in the machine, allowing the threshold to be set according to the application of the machine to achieve accurate monitoring of the bearings.
[0031] Advantageously, when the machine is a mobile machine, if the kurtosis of the vibration measurement set is less than a predetermined kurtosis threshold, the value of each statistical parameter is determined during the training period and during the normal operation of the machine.
[0032] Preferably, steps (d), (e), (f), and (g) are repeated for a predetermined duration.
[0033] Advantageously, determining the detection threshold involves summing the values of each normal distribution contained in a first interval with a lower limit equal to 1 minus a predetermined first quantile, the detection threshold being equal to the sum.
[0034] Preferably, the scoring includes defining the score for each statistical parameter:
[0035] - Perform Z-score normalization on the first value of each statistical parameter based on the mean and variance of a normal distribution modeling the values of the statistical parameters, to obtain standardized values for the statistical parameters.
[0036] - Summing each standardized value contained in a second interval with a lower limit equal to 1 minus a predetermined second quantile, the sum being equal to the score.
[0037] Advantageously, comparing the score with the detection threshold includes comparing the score with the detection threshold minus the number of standardized values included in the second interval.
[0038] Preferably, the bearing is considered defective when the score is greater than the detection threshold.
[0039] Advantageously, the second statistical parameter includes:
[0040] - The root mean square of the set of vibration measurement results, or
[0041] - The sum of the areas of identical squares, each square being defined by the diagonal of two vibration measurements connecting the set of vibration measurements, or
[0042] - The dispersion entropy of the measurement result set, or
[0043] - The permutation entropy of the measurement result set, or
[0044] - The average value of a predetermined number of the strongest harmonics in the spectrum of the measurement result set, wherein the first statistical parameter and the second statistical parameter are different.
[0045] Preferably, the method further includes defining a third statistical parameter, which is equal to the square of the sum of the first statistical parameter and the second statistical parameter.
[0046] According to one aspect, a bearing device is proposed, comprising a bearing and a vibration sensor for measuring the vibration of the bearing, the bearing being provided with an inner ring and an outer ring capable of rotating concentrically relative to each other.
[0047] The bearing assembly further includes:
[0048] - A first determining component, configured to determine, based on vibration measurements delivered by the sensor when the rotational speed of one of the inner and outer rings relative to the other ring falls within a predetermined range, the value of at least a first statistical parameter equal to the Hjorth complexity parameter.
[0049] - A modeling component for modeling the values of at least the first statistical parameter using a normal distribution during a training period after the sensor is implemented in the machine.
[0050] - A second determining component, configured to determine a detection threshold based at least on the normal distribution during the training time period.
[0051] - A third determining component, configured to determine a score during normal operation of the machine based at least on a first value of the Hjorth complexity parameter and a first normal distribution modeled on the value of the first statistical parameter, and
[0052] - A comparison component for comparing the score with the detection threshold and for determining the bearing failure based on the result of the comparison.
[0053] According to another aspect, a machine is proposed that includes a bearing device as defined above. Attached Figure Description
[0054] Other advantages and features of the invention will become apparent upon reviewing the detailed description of the non-limiting embodiments and the accompanying drawings, in which:
[0055] Figure 1 A rotating machine according to the present invention is illustrated schematically;
[0056] Figure 2 An example of a monitoring device for monitoring bearing devices according to the present invention is shown schematically;
[0057] Figure 3 An example of statistical parameters according to the present invention is illustrated schematically;
[0058] Figure 4 An example of another statistical parameter according to the invention is illustrated schematically.
[0059] Figure 5 An algorithm for calculating another statistical parameter is illustrated schematically;
[0060] Figure 6 and Figure 7 An example of a method for monitoring bearing devices according to the present invention is shown. Detailed Implementation
[0061] Reference Figure 1 , Figure 1 A partial longitudinal section of machine 1 is schematically shown.
[0062] Machine 1 includes a housing 2 and a shaft 3 supported in the housing 2 by roller bearings 4.
[0063] Machine 1 can be a mobile machine, such as a truck, and axle 3 is the truck's axle supported by roller bearing 4.
[0064] In the variant, machine 1 can be a stationary machine, such as a machine tool.
[0065] The roller bearing 4 is provided with an inner ring 5 mounted on the shaft 3 and an outer ring 6 mounted in a hole in the housing 2. The outer ring 6 surrounds the inner ring 5 radially. The inner ring 5 and the outer ring 6 rotate concentrically relative to each other.
[0066] The roller bearing 4 is further provided with a row of rolling elements 7 radially positioned between the inner raceway of the inner ring 5 and the outer raceway of the outer ring 6. In the example shown, the rolling elements 7 are balls. Alternatively, the roller bearing may include other types of rolling elements 7, such as rollers. In the example shown, the roller bearing includes a single row of rolling elements 7. Alternatively, the roller bearing 4 may include several rows of rolling elements.
[0067] Sensor 8 is installed in seat 2 to measure the rotational speed of bearing 4.
[0068] Sensor 8 can be mounted on the hole in seat 2.
[0069] In a variant, sensor 8 can be mounted on outer ring 6 or on outer ring 6 and in the hole of seat 2.
[0070] Sensor 8 delivers a signal indicating the vibration of bearing 4 to monitoring device 9.
[0071] In the variant, sensor 8 is located in monitoring device 9.
[0072] Bearing 4 and monitoring device 9 form a bearing assembly.
[0073] Figure 2 An example of monitoring device 9 is shown.
[0074] The monitoring device 9 includes a first determining component 10 for determining the value of at least a statistical parameter P1 equal to the Hjorth complexity parameter based on vibration measurement results delivered by the sensor 8.
[0075] The vibration measurement results set includes a predetermined number of signal samples x delivered by sensor 8. i , where i is an integer ranging from 1 to M, for example, equal to eight thousand samples.
[0076] The monitoring device 9 also includes a modeling component 11, a second determination component 12, a third determination component 13, a comparison component 14, a memory 15, and an implementation component 16.
[0077] In the variant, the memory is located outside the monitoring device 9.
[0078] The first determining component 10 implements the first algorithm ALGO1 to determine the first statistical parameter P1 in the vibration measurement result set that is equal to the Hjorth complexity parameter, such that:
[0079]
[0080] Where Var(X) is the variance of variable X, d(X) is the first derivative of variable X, d²(X) is the second derivative of variable X, and Mobility equals:
[0081]
[0082] The first determining component 10 can implement an additional algorithm to determine additional statistical parameters.
[0083] Assume that the first determining component 10 also implements the second algorithm ALGO2, the third algorithm ALGO3, the fourth algorithm ALGO4, the fifth algorithm ALGO5, the sixth algorithm ALGO6, the seventh algorithm ALGO7 and the eighth algorithm ALGO8.
[0084] Based on the set of vibration measurement results delivered by the sensors, the second algorithm ALGO2 determines the second statistical parameter P2, the third algorithm ALGO3 determines the third statistical parameter P3, the fourth algorithm ALGO4 determines the fourth statistical parameter P4, the fifth algorithm ALGO5 determines the fifth statistical parameter P5, the sixth algorithm ALGO6 determines the sixth statistical parameter P6, the seventh algorithm ALGO7 determines the seventh statistical parameter P7, and the eighth algorithm ALGO8 determines the eighth statistical parameter P8.
[0085] The seven statistical parameters P1, P2, P3, P4, P5, P6, and P7 are different from each other.
[0086] For example, the second algorithm ALGO2 determines a second statistical parameter P2 that is equal to the root mean square of the set of vibration measurement results, such that:
[0087]
[0088] The third algorithm ALGO3 and the fourth algorithm ALGO4, for example, determine the third statistical parameter P3 and the fourth statistical parameter P4 based on the sum of the areas of identical squares, respectively.
[0089] Each square is defined by the diagonal of two vibration measurement results that connect the sets of vibration measurement results.
[0090] For example, the third statistical parameter P3 is equal to the sum of identical areas, each area A1 being defined by the diagonal D1 connecting sample S1 in the measurement result set with the third sample S3 in the measurement result set following said sample S1, such as... Figure 3As shown, ti is the sampling time, and i varies between 1 and M, such that:
[0091]
[0092] For example, the fourth statistical parameter P4 is equal to the sum of identical areas, each area A11, A12, A13, A14 being defined by diagonals D11, D12, D13, D14 connecting samples S1, S2, S3, S4 in the measurement result set with the next samples S2, S3, S4, S5 in the measurement result set after those samples S1, S2, S3, S4, such that:
[0093]
[0094] The fifth algorithm ALGO5 determines the fifth statistical parameter P5, which is, for example, equal to the dispersion entropy of the measurement result set.
[0095] M. Rostaghi and H. Azami, in their paper "Dispersion Entropy: A measure for time series analysis" published in IEEE Signal Processing Letters, Vol. 23, No. 5, pp. 610-614, disclose an algorithm for determining the dispersion entropy value DE of a univariate signal X of length N, where X = {X1, X2, ..., XN}, and N is an integer.
[0096] A univariate signal X, for example, equals a set of vibration measurement results including a predetermined number M samples x. i .
[0097] The algorithm includes Figure 5 The four main steps are shown.
[0098] During step 17, samples Xj of the univariate signal X (j varies from 1 to N) are mapped to class c labeled from 1 to c.
[0099] The univariate signal X is mapped to the signal Y = {Y1, Y2, ..., YN} using the normal cumulative distribution function (NCDF).
[0100] For each sample Yj of the signal Y (j is an integer ranging from 1 to N), the following constraints are defined: Values, where:
[0101]
[0102] The round() operator increments or decrements a number to the next integer.
[0103] During step 18, embedding vectors are created using the embedding dimension m. This makes each embedding vector Equals time series.
[0104]
[0105] Where d is the time delay, and k varies between 1 and N-(m-1)d.
[0106] Then, each time series Mapped to walking pattern ,in, .
[0107] It can be assigned to each time series The number of possible dispersion patterns, etc. m Because the signal has m members, and each member can be an integer from 1 to c.
[0108] During step 19, for c m For each of the possible walking patterns, the relative frequency is obtained as follows:
[0109]
[0110] The number() operator converts a value into a number.
[0111] It shows the assignment to Walking pattern The number is divided by the total number of embedded signals with an embedding dimension m.
[0112] During step 20, the walk entropy value DE is determined:
[0113]
[0114] The sixth algorithm ALGO6 determines the sixth statistical parameter P6, which is, for example, equal to the permutation entropy of the measurement result set.
[0115] Reference: “Efficiently measuring complexity on the basis of real-world data”, VA Unakafova, K. Keller, Entropy, 15(10), 4392–4415.
[0116] Permutation entropy provides a quantification measure of the complexity of vibration measurement results by capturing the sequential relationships between the values of vibration measurement results and extracting the probability distribution of sequential patterns.
[0117] The seventh algorithm ALGO7 determines the seventh statistical parameter P7, which is, for example, equal to the average of a predetermined number of the strongest harmonics in the spectrum of the measurement result set.
[0118] For example, the Fast Fourier Transform algorithm is implemented to determine the spectrum of the measurement result set, and the average value is determined from the one hundred strongest harmonics of the spectrum.
[0119] The eighth parameter P8 is determined according to the eighth algorithm ALGO8 implemented by the first determining component 10.
[0120] The eighth parameter P8 is, for example, equal to the square root of the sum of the statistical parameters:
[0121]
[0122] The following section shows an example of a method for implementing monitoring device 9.
[0123] The method includes a first part and a second part. The first part is used to determine a detection threshold Sd during a training period after the sensor 8 is implemented in the machine 1. The detection threshold Sd is used to determine a fault in the bearing 4. The second part is used to determine a fault in the bearing 4 during normal operation of the machine 1 after the training period.
[0124] Figure 6 An example of the first part of the method is shown.
[0125] Suppose that one of the inner rings 5 and outer rings 6 rotates relative to the other of the inner rings 6 and outer rings 5.
[0126] During step 21, the first part of the method begins when the predetermined duration between two consecutive method executions is reached.
[0127] The predetermined duration is, for example, 24 hours, such that a training period is performed once a day for, for example, one month after the sensor 8 is implemented in the machine 1.
[0128] When the first part of the method (step 21) begins, during step 22, if the rotational speed of the circle is contained within a predetermined range, then in step 23, a set of vibration measurement results, including xi signal samples delivered by sensor 8, is stored in, for example, memory 15, where i varies between 1 and an integer M.
[0129] M is, for example, equal to 8000, so that the set of vibration measurement results includes 8000 samples xi.
[0130] The predetermined range is defined by a lower limit and an upper limit.
[0131] The lower and upper limits are determined based on the type (mobile or stationary) and purpose of machine 1.
[0132] For mobile machines, the lower limit is, for example, 50 miles per hour, while the upper limit is infinity.
[0133] If the rotational speed of the circle is less than the lower limit, the first part of the method returns to step 21, waiting for the next start of the first part of the method during the training period.
[0134] When xi samples are stored in memory 15 (step 24), if machine 1 is a mobile machine, during step 25, implementation component 16 determines the kurtosis value of the vibration measurement result set.
[0135] If machine 1 is a fixed machine, the first part of the method proceeds from step 24 to step 26.
[0136] If the determined kurtosis value is greater than the predetermined kurtosis threshold, the first part of the method returns to step 21.
[0137] If the determined kurtosis value is less than a predetermined kurtosis threshold, during step 26, the first determining component 10 implements eight algorithms ALGO1, ALGO2, ALGO3, ALGO4, ALGO5, ALGO6, ALGO7, and ALGO8 to determine the value of each of eight statistical parameters P1, P2, P3, P4, P5, P6, P7, and P8 based on the sample xi.
[0138] The eight values of the eight statistical parameters P1, P2, P3, P4, P5, P6, P7, and P8 are stored, for example, in memory 15.
[0139] The kurtosis threshold is, for example, equal to 6.
[0140] Repeat steps 22, 23, 24, 25, and 26 until a predetermined number of values for each statistical parameter P1, P2, P3, P4, P5, P6, P7, and P8 are stored in memory 15, for example, until forty values of each statistical parameter P1, P2, P3, P4, P5, P6, P7, and P8 are stored in memory 15.
[0141] As long as the predetermined number of values for each statistical parameter P1, P2, P3, P4, P5, P6, P7, P8 have not yet been stored in memory 15 (step 27), the first part of the method returns to step 21.
[0142] When a predetermined number of values for each statistical parameter P1, P2, P3, P4, P5, P6, P7, P8 are stored in memory 15 (step 27), during step 28, modeling component 11 models the stored values of each statistical parameter P1, P2, P3, P4, P5, P6, P7, P8 using a normal distribution defined by the average value and the variance of the values of the statistical parameters.
[0143] The value of the first statistical parameter P1 is determined by the mean μ1 and variance σ1. 2 Modeling the first normal distribution.
[0144] The value of the second statistical parameter P2 is determined by the mean μ2 and variance σ2. 2 Modeling the second normal distribution.
[0145] The value of the third statistical parameter P3 is determined by the mean μ3 and variance σ3. 2 Modeling the third normal distribution.
[0146] The value of the fourth statistical parameter P4 is determined by the mean μ4 and variance σ4. 2 Modeling the fourth normal distribution.
[0147] The value of the fifth statistical parameter P5 is determined by the mean μ5 and variance σ5. 2 The fifth normal distribution model.
[0148] The value of the sixth statistical parameter P6 is determined by the mean μ6 and variance σ6. 2 Modeling the sixth normal distribution.
[0149] The value of the seventh statistical parameter P7 is determined by the mean μ7 and variance σ7. 2 Modeling the seventh normal distribution.
[0150] The value of the eighth statistical parameter P8 is determined by the mean μ8 and variance σ8. 2 Modeling the eighth normal distribution.
[0151] During step 29, the second determining component 12 determines the detection threshold Sd by adding the values in the first interval of the first distribution to the seventh distribution that are included in the first interval with a lower limit equal to 1 minus a predetermined first quantile.
[0152] The first quantile is selected based on the sensitivity of detecting bearing failures.
[0153] The lower the predetermined first quantile, the greater the fault detection sensitivity of this method.
[0154] The first quantile is pre-defined as, for example, 80%.
[0155] Figure 7 An example of the second part of the method during normal operation of machine 1 after the training period is shown.
[0156] Assuming that one of the inner circle 5 and the outer circle 6 is rotated relative to the other of the inner circle 6 and the outer circle 5, determine the first normal distribution to the eighth normal distribution, and determine the detection threshold Sd.
[0157] If machine 1 is a mobile machine, then the second part of the method includes steps 21, 22, 23, 24, 25 and 26.
[0158] If machine 1 is a fixed machine, then the second part of the method does not include step 25, and the second part of the method proceeds from step 24 to step 26.
[0159] When the value of each of the eight statistical parameters P1, P2, P3, P4, P5, P6, P7, and P8 is determined based on the sample xi, in step 30, the third determining component 13 determines the score Sc based on the value of each of the eight statistical parameters P1, P2, P3, P4, P5, P6, P7, and P8 determined during the training period and the first normal distribution to the seventh normal distribution.
[0160] The third determining component 13 performs Z-score normalization for each value of statistical parameters P1, P2, P3, P4, P5, P6, P7, P8 based on the mean and variance of the normal distribution associated with statistical parameters P1, P2, P3, P4, P5, P6, P7, P8, to obtain a standardized value for each value of statistical parameters P1, P2, P3, P4, P5, P6, P7, P8.
[0161] The value of each of the statistical parameters P1, P2, P3, P4, P5, P6, P7, and P8 is denoted as Val1, Val2, Val3, Val4, Val5, Val6, Val7, and Val8.
[0162] The standardized values of each Val1, Val2, Val3, Val4, Val5, Val6, Val7, Val8 are denoted as ValS1, ValS2, ValS3, ValS4, ValS5, ValS6, ValS7, ValS8.
[0163] The following equation compares each standardized value ValS1, ValS2, ValS3, ValS4, ValS5, ValS6, ValS7, ValS8 with the values Val1, Val2, Val3, Val4, Val5, Val6, Val7, Val8, the mean μ1, μ2, μ3, μ4, μ5, μ6, μ7, μ8, and the variance σ1. 2 σ2 2σ3 2 σ4 2 σ5 2 σ6 2 σ7 2 σ8 2 Connecting them:
[0164]
[0165] Here, j varies between 1 and 7.
[0166] The score Sc is equal to the sum of the standardized values ValSj contained in the second interval, which is equal to 1 minus the predetermined second quantile.
[0167] The second quantile is pre-determined to be, for example, 90%.
[0168] During step 31, the comparison unit 14 compares the score Sc with the detection threshold Sd minus the number of standardized values contained in the second interval determined in step 30.
[0169] Subtracting the number of standardized values contained in the second interval as determined in step 30 allows the detection threshold Sd to be weighted based on the number of standardized values contained in the second interval.
[0170] If the score Sc is less than the detection threshold Sd minus the number of standardized values contained in the second interval (step 32), then bearing 4 is considered to be functioning. The second part of the method returns to step 21.
[0171] If the score Sc is greater than the detection threshold Sd minus the number of standardized values contained in the second interval (step 32), then bearing 4 is considered defective, and during step 33, comparison component 14 delivers an alarm to prevent bearing 4 from being defective.
[0172] Since the determination of the Hjorth complexity parameter is insensitive to changes in the rotational speed of bearing 4, the first statistical parameter P1, which is equal to the Hjorth complexity parameter in the set of vibration measurement results, allows for accurate prediction of bearing 4 damage by analyzing only the evolution of the first statistical parameter.
[0173] To increase diagnostic reliability, as shown above, several statistical parameters (one of which is equal to the Hjorth complexity parameter) are determined based on the recorded vibration measurement results of bearing 4. By analyzing the evolution of these parameters, early prediction of damage to bearing 4 can be made.
[0174] By using several statistical parameters to predict the failure of bearing 4, different types of defects in bearing 4 can be detected.
[0175] Furthermore, the detection threshold Sd is determined based on the vibration measurement results of the bearing 4 in machine 1, allowing the threshold to be defined according to the application of machine 1 to obtain accurate monitoring of bearing 4.
Claims
1. A method for monitoring a bearing assembly in a machine (1), the bearing assembly comprising a bearing (4) and a vibration sensor (8) for measuring the vibration of the bearing, the bearing (4) being provided with an inner ring (5) and an outer ring (6) capable of rotating concentrically relative to each other, the method comprising the steps of: a. Determine the value of at least a first statistical parameter equal to the Hjorth complexity parameter (P1) based on at least one set of vibration measurements delivered by the sensor (8) during a training period after the sensor (8) is implemented in the machine (1) and when the rotational speed of one of the inner rings (4) and the outer rings (5) relative to the other ring is contained within a predetermined interval. b. Model the value of at least the first statistical parameter (P1) using a normal distribution during the training period. c. Determine the detection threshold (Sd) based at least on the normal distribution during the training period. d. Determine at least a first value of the first statistical parameter (P1) based on a set of vibration measurements delivered by the sensor during normal operation of the machine and when the rotational speed of the one revolution is contained within the predetermined range. e. Determine the score (Sc) during normal operation of the machine (1) based at least on the first value of the Hjorth complexity parameter (P1) and the normal distribution modeled on the value of the first statistical parameter. f. Compare the score (Sc) with the detection threshold (Sd), and g. Determine the fault of the bearing (4) based on the results of the comparison.
2. The method according to claim 1, characterized in that: Step (a) further includes determining the value of at least a second statistical parameter based on the set of vibration measurements delivered by the sensor (8) during the training period. Step (b) further includes modeling the values of the second statistical parameters (P2, P3, P4, P5, P6, P7) using a normal distribution during the training period. - Step (c) includes determining the detection threshold (Sd) based on a normal distribution modeling the values of the first statistical parameter during the training period and a normal distribution modeling the values of the second statistical parameter. - Step (d) further includes determining a first value for the second statistical parameter (P2, P3, P4, P5, P6, P7) based on the set of vibration measurements delivered by the sensor (8) during normal operation, and - Step (e) includes determining a score (Sc) during normal operation of the machine (1) based on a first value of the Hjorth complexity parameter (P1), a first value of the second statistical parameter (P2, P3, P4, P5, P6, P7), a normal distribution modeled on the value of the first statistical parameter, and a normal distribution modeled on the value of the second statistical parameter.
3. The method according to claim 1 or 2, characterized in that, When the machine (1) is a mobile machine, if the kurtosis of the vibration measurement result set is less than a predetermined kurtosis threshold, the value of each statistical parameter (P1, P2, P3, P4, P5, P6, P7) is determined during the training period and during the normal operation of the machine.
4. The method according to any one of claims 1 to 3, characterized in that, Determining the detection threshold (Sd) involves summing the values of each normal distribution contained in a first interval with a lower limit equal to 1 minus a predetermined first quantile, the detection threshold being equal to the sum.
5. The method according to any one of claims 1 to 4, characterized in that, Determining the score (Sc) involves considering each statistical parameter: - Perform Z-score normalization of the first value of each statistical parameter based on the mean and variance of a normal distribution modeling the values of the second statistical parameter, to obtain standardized values for the statistical parameters, and - Summing each standardized value contained in a second interval with a lower limit equal to 1 minus a predetermined second quantile, the sum being equal to the score.
6. The method according to claim 5, characterized in that, Comparing the score (Sc) with the detection threshold (Sd) involves comparing the score (Sc) with the detection threshold (Sd) minus the number of standardized values contained in the second interval.
7. The method according to any one of claims 1 to 6, characterized in that, When the score (Sc) is greater than the detection threshold (Sd), the bearing (4) is considered defective.
8. The method according to any one of claims 2 to 7, when referring to claim 2, characterized in that, The second statistical parameters (P2, P3, P4, P5, P6, P7) include: - The root mean square of the set of vibration measurement results, or - The sum of the areas of identical squares, each square being defined by the diagonal of two vibration measurements connecting the set of vibration measurements, or - The spread entropy of the set of measurement results, or - The permutation entropy of the set of measurement results, or - The average value of a predetermined number of the strongest harmonics in the spectrum of the measurement result set, wherein the first statistical parameter and the second statistical parameter are different.
9. The method according to any one of claims 2 to 7, when referring to claim 2, characterized in that, The method further includes defining a third statistical parameter (P8), which is equal to the square of the sum of the first statistical parameter and the second statistical parameter.
10. A bearing assembly comprising a bearing (4) and a vibration sensor for measuring the vibration of the bearing, the bearing (4) being provided with an inner ring (5) and an outer ring (6) capable of rotating concentrically relative to each other, the bearing assembly further comprising: - A first determining component (10) is used to determine the value of at least a first statistical parameter equal to the Hjorth complexity parameter (P1) based on vibration measurements delivered by the sensor when the rotational speed of one of the inner rings (4) and the outer rings (5) relative to the other ring is contained within a predetermined range. - A modeling component (11) for modeling the value of at least the first statistical parameter using a normal distribution during a training period after the sensor (8) is implemented in the machine (1). - A second determining component (12) is used to determine a detection threshold (Sd) based at least on a normal distribution during the training period. - A third determining component (13) is used to determine, at least based on a first value of the Hjorth complexity parameter (P1) and a normal distribution modeled on the value of the first statistical parameter, a score (Sc) during normal operation of the machine (1), and - A comparison component (14) is used to compare the score (Sc) with the detection threshold (Sd) and to determine the failure of the bearing (4) based on the result of the comparison.