DTW-based rolling bearing fault detection method

Abstract

The invention discloses a DTW-based rolling bearing early fault detection method, which comprises the following steps: selecting a sensor and a mounting position thereof, and determining a sampling parameter and a threshold value; acquiring multiple groups of periodic synchronous single-point sampling data X; calculating the DTW distance between two adjacent groups of periodic synchronization single-point sampling data in the multiple groups of periodic synchronization single-point sampling data X to obtain DTW distance data D; calculating an average value M of the DTW distance data D; and judging whether the rolling bearing fails or not. According to the method, fault features are extracted through the DTW algorithm, the influence of working conditions and disturbance noise on a fault detection result is reduced, bearing faults can be detected in real time, the reliability and real-time performance of rolling bearing fault detection are improved, and the fault detection efficiency is improved. According to the method, the fault mode of the rolling bearing can be reliably identified, a control unit with high processing speed does not need to be adopted, the calculation cost is relatively low, and thus the fault detection cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent fault detection, and particularly relates to a rolling bearing fault detection method based on Dynamic Time Warping (DTW). Background Art

[0002] A rolling bearing is a mechanical component used to reduce the friction coefficient of rotational motion. It replaces sliding by rolling rolling elements (such as balls, rollers, etc.) between the inner and outer rings, thereby reducing the friction between the rotating shaft and the bearing housing. Rolling bearings have the advantages of compact structure, small friction coefficient, high efficiency, long life, etc., and are now widely used as key components in various rotating mechanical equipment.

[0003] Due to the complexity of the working environment and long-term high-load operation, rolling bearings will inevitably have various faults, such as wear, shedding, cracks, etc. If the faults of rolling bearings cannot be detected and processed in time, it may lead to the shutdown of rotating mechanical equipment and even cause major safety accidents. In order to improve the working efficiency of rotating machinery and avoid equipment damage caused by rolling bearing faults, it is necessary to detect the faults of rolling bearings, thereby improving the reliability of the operation of rotating mechanical equipment.

[0004] Traditional fault detection methods mainly rely on manual experience and manual signal analysis, and have problems such as high subjectivity and low efficiency. At present, in order to reliably detect rolling bearing faults and avoid their hazards, experts and scholars have proposed various rolling bearing fault detection methods. The invention patent with the application number 202411533814.1 discloses a rolling bearing fault detection method. By acquiring the vibration data of the rolling bearing to be measured, the vibration data is subjected to modal decomposition by using the parameter self-consistent variational mode decomposition method to obtain a set of modal functions; then the effective modal functions are screened, and the correlation-kurtosis values in each mode are calculated; then reconstruction processing and Hilbert demodulation processing are carried out, and finally it is judged whether the rolling bearing has a fault according to the obtained envelope spectrum. Although this method can detect early weak faults, the data processing and calculation process are complex, and the calculation cost is relatively high, resulting in a relatively high cost for rolling bearing fault detection. The invention patent with the application number 202410560772.4 discloses a rolling bearing weak fault early detection method and device. The vibration force value is detected by using a vibration force sensor. If the vibration force value is different from the set value, it is considered that the bearing has a fault. Although this method can detect rolling bearing faults, the vibration force measured by the vibration force sensor is easily affected by the working conditions of rotating machinery, and the reliability of fault detection is low.

[0005] To address the above problems, this invention patent discloses a rolling bearing fault detection method based on DTW. DTW is a machine learning algorithm commonly applied in the field of intelligent speech recognition. Since the vibration patterns of rolling bearings are different under different health states, DTW can also be applied to the field of rolling bearing fault detection. This invention patent uses the DTW algorithm to extract bearing fault features, which not only requires relatively low computational costs but also reduces the impact of working conditions and disturbance noise on the reliability of fault detection. Summary of the Invention

[0006] Object of the Invention: Aiming at the defects existing in the existing methods in the background technology, this invention discloses a rolling bearing fault detection method based on DTW. This inventive method uses the DTW algorithm to extract fault features, which not only reduces the impact of working conditions and disturbance noise on the fault detection results but also can detect bearing faults in real time, improving the reliability and real-time performance of rolling bearing fault detection. This method can reliably identify rolling bearing fault modes without using a control unit with a high processing speed, and the computational cost is relatively low, thus reducing the fault detection cost.

[0007] Technical Solution: A rolling bearing fault detection method based on DTW according to the present invention specifically includes the following steps:

[0008] S1. Select sensors and their installation positions, and determine the sampling parameters and threshold T;

[0009] S2. Obtain multiple groups of periodically synchronized single-point sampling data X;

[0010] S3. Calculate the DTW distances between adjacent two groups of periodically synchronized single-point sampling data in multiple groups of periodically synchronized single-point sampling data X respectively to obtain DTW distance data D;

[0011] S4. Calculate the average value M of the DTW distance data D;

[0012] S5. Determine whether the rolling bearing has a fault.

[0013] Further, the sensors described in S1 include vibration sensors and rotational speed sensors; the vibration sensors can be vibration displacement sensors, vibration velocity sensors, or vibration acceleration sensors; the rotational speed sensors can be optoelectronic rotational speed sensors or Hall-type rotational speed sensors.

[0014] Further, the vibration sensor described in S1 is installed near the bearing to be detected for measuring bearing vibration data; the rotational speed sensor is installed near the rotating shaft for measuring the rotational speed data of the rotating shaft.

[0015] Further, the sampling parameters in S1 include the number of synchronous data points m collected per revolution of the rotating shaft, the data length mL, where L can also be regarded as the number of revolutions of the rotating shaft, and determine the number of synchronous data points m collected per revolution of the rotating shaft, the data length mL, and the number of revolutions L of the rotating shaft; the threshold T is a preset critical value of the fault index.

[0016] Further, the periodic synchronous single-point sampling in S2 means that a single sampling data is synchronously obtained within each revolution period of the rotating shaft.

[0017] Further, since m data points are collected per revolution of the rotating shaft in S2, therefore, m groups of periodic synchronous single-point sampling data X = [X 1 , X 2 , X 3 , …, X i , …, X m T ,

[0018]

[0019] Among them, the i-th group of periodic synchronous single-point sampling data is X i = [x i , x i+m , x i+2m , …, x i+(L-1)m .

[0020] Further, in S3, the DTW algorithm is used to calculate the DTW distance between adjacent two groups of periodic synchronous single-point sampling data respectively, and the distance data D is obtained:

[0021] D = [d 1 d 2 d 3 … d i … d m-1 ,

[0022] Among them, d i represents the DTW distance between the i-th group and the i + 1-th group of periodic synchronous single-point sampling data X i and X i+1 .

[0023] Further, in S4, the mean value M of the distance data D is calculated:

[0024]

[0025] Further, S5 is specifically: regarding the calculated mean value M as the fault index and comparing it with the preset threshold T. If the mean value M is less than the preset threshold T, it is considered that the bearing has no fault. If the mean value M is greater than or equal to the preset threshold T, it is considered that the bearing has a fault.

[0026] Beneficial effects:

[0027] (1) A rolling bearing fault detection method based on DTW disclosed by the method of the present invention can reliably identify the healthy mode of rolling bearings, reduce the influence of working conditions and disturbance noise on the bearing fault detection results, reduce the misdiagnosis rate and missed diagnosis rate of fault detection, and improve the reliability of rolling bearing fault detection;

[0028] (2) A rolling bearing fault detection method based on DTW disclosed by the method of the present invention has relatively low calculation costs and does not require a control unit with a high processing speed, thereby reducing the fault detection cost; Description of the drawings

[0029] Figure 1 is a flow chart of a rolling bearing fault detection method based on DTW disclosed by the present invention;

[0030] Figure 2 are the displacement data and multiple groups of periodically synchronized single-point sampling data when the bearing has no fault;

[0031] Figure 3 are the displacement data and multiple groups of periodically synchronized single-point sampling data when the bearing has a fault;

[0032] Figure 4 are the detection results of the method of the present invention. Specific embodiments

[0033] The implementation process of the method of the present invention will be described below in conjunction with the drawings and specific embodiments.

[0034] The present invention provides a rolling bearing fault detection method based on DTW, and its process is as Figure 1 shown, including the following steps:

[0035] S1. Select a sensor and its installation position, and determine the sampling parameters and threshold T;

[0036] S2. Obtain multiple groups of periodically synchronized single-point sampling data X;

[0037] S3. Calculate the DTW distances between adjacent two groups of periodically synchronized single-point sampling data in multiple groups of periodically synchronized single-point sampling data X to obtain DTW distance data D;

[0038] S4. Calculate the average value M of the DTW distance data D;

[0039] S5. Judge whether the rolling bearing has a fault.

[0040] Further, the sensors described in S1 include a vibration sensor and a rotational speed sensor; the vibration sensor can be a vibration displacement sensor, a vibration velocity sensor, or a vibration acceleration sensor; the rotational speed sensor can be an optoelectronic rotational speed sensor or a Hall rotational speed sensor.

[0041] Further, the vibration sensor described in S1 is installed near the bearing to be detected and is used to measure the bearing vibration data; the rotational speed sensor is installed near the rotating shaft and is used to measure the rotational speed data of the rotating shaft.

[0042] Further, the sampling parameters described in S1 include the number of synchronous data points m collected per revolution of the rotating shaft, the data length mL, and L can also be regarded as the number of revolutions of the rotating shaft; and it is determined that the number of synchronous data points m collected per revolution of the rotating shaft is m = 100, the data length mL is mL = 1000, and the number of revolutions of the rotating shaft L is L = 10; the threshold T = 1.647 is a preset critical value of the fault index.

[0043] Further, the periodic synchronous single-point sampling described in S2 means that a single sampling data is synchronously obtained within each revolution period of the rotating shaft.

[0044] Further, since m = 100 data points are synchronously collected per revolution of the rotating shaft in S2, therefore, 100 groups of periodic synchronous single-point sampling data X = [X 1 , X 2 , X 3 , …, X i , …, X 100 T ,

[0045]

[0046] Among them, the i-th group of periodic synchronous single-point sampling data is X i = [x i , x i+100 , x i+200 , …, x i+900 .

[0047] Further, in S3, the DTW algorithm is used to calculate the DTW distance between adjacent two groups of periodic synchronous single-point sampling data respectively, and the distance data D is obtained:

[0048] D = [d 1 d 2 d 3 … d i … d 99 ,

[0049] Among them, d i represents the DTW distance between the i-th group and the i + 1-th group of periodic synchronous single-point sampling data X i and X i+1 .

[0050] Further, in S4, the mean value M of the distance data D is calculated:

[0051]

[0052] Further, S5 is specifically as follows: The calculated mean value M is regarded as a fault index and compared with a preset threshold T = 1.65; if the mean value M is less than the preset threshold T, it is considered that the bearing has no fault, and the process returns to step S2 to continue fault detection; if the mean value M is greater than or equal to the preset threshold T, it is considered that the bearing has a fault.

[0053] This embodiment uses the vibration data and multiple groups of periodic synchronous single-point sampling data of the rolling bearing at the spindle of the milling machine when it is fault-free and faulty, as shown respectively in Figure 2 and 3 When the above data is processed by the method of the present invention, it can be obtained that the fault indexes when there is no fault are all less than the predetermined threshold T = 1.65; for the vibration data when the bearing is faulty, the fault indexes calculated by the method of the present invention are all greater than the predetermined threshold T = 1.65, and then it can be judged that a fault has occurred.

[0054] In summary, the method of the present invention discloses a rolling bearing fault detection method based on DTW, which overcomes the shortcomings of the existing rolling bearing fault detection methods for rotating machinery being affected by working conditions and disturbance noises, can be applied to the detection of rolling bearing faults in rotating machinery, reduces the misdiagnosis rate and missed diagnosis rate of fault detection, and improves the reliability of rolling bearing fault detection in rotating machinery; the calculation cost of the method of the present invention is relatively low, and there is no need to use a control unit with a high processing speed, thus reducing the detection cost during fault detection.

[0055] The above are only examples of the present invention and are not used to limit the present invention. All equivalent replacements made within the principle of the present invention shall be included in the protection scope of the present invention. The content not elaborated in detail in the present invention belongs to the prior art well-known to those skilled in the art.

Claims

1. A rolling bearing fault detection method based on DTW, characterized in that: The following steps are involved: S1, select the sensor and its installation location, determine the sampling parameters and threshold T; S2, obtaining multiple sets of periodic synchronous single-point sampling data X; S3, respectively calculating the DTW distances of two adjacent groups of periodic synchronous single-point sampling data in the multiple groups of periodic synchronous single-point sampling data X to obtain DTW distance data D; S4, calculating the average value M of the DTW distance data D; S5. Determine whether the rolling bearing is faulty.

2. The rolling bearing fault detection method based on DTW according to claim 1, characterized in that: The vibration sensor described in S1 may be a vibration displacement sensor, a vibration velocity sensor or a vibration acceleration sensor; the rotation speed sensor may be a photoelectric rotation speed sensor or a Hall type rotation speed sensor.

3. The rolling bearing fault detection method based on DTW according to claim 1, characterized in that: The vibration sensor described in S1 is installed near the pre-detection bearing to measure the bearing vibration data; the speed sensor described in S1 is installed near the shaft to measure the shaft speed data.

4. The rolling bearing fault detection method based on DTW according to claim 1, characterized in that: The sampling parameters in S1 include the number of synchronous data points m and data length mL collected per shaft revolution, where L can also be regarded as the number of shaft revolutions; the number of synchronous data points m, data length mL and shaft revolutions L collected per shaft revolution are determined; the threshold T is a preset fault indicator critical value.

5. The rolling bearing fault detection method based on DTW according to claim 1, characterized in that: The periodic synchronous single-point sampling data in S2 refers to the synchronous acquisition of single sampling data within each rotation period of the shaft; since m data points are collected within each rotation of the shaft, m groups of periodic synchronous single-point sampling data X = [X1, X2, X3, ..., X i ,…,X m ] T , Among them, the i-th group of periodic synchronous single-point sampling data is X i =[x i ,x i+m ,x i+2m ,…,x i+(L-1)m ].

6. The rolling bearing fault detection method based on DTW according to claim 1, characterized in that: In S3, the DTW algorithm is used to calculate the DTW distances between two adjacent groups of periodic synchronous single-point sampling data in the m groups of periodic synchronous single-point sampling data X, thereby obtaining the distance data D: <h2 style=";text-align:left;direction:ltr">D = [d1 d2 d3…d<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> …d<h2 style=";text-align:left;direction:ltr"> m-1 <h2 style=";text-align:left;direction:ltr"> ], Among them, d i Indicates the periodic synchronous single-point sampling data X of the i-th group and the i+1-th group i and X i+1 DTW distance.

7. The rolling bearing fault detection method based on DTW according to claim 1, characterized in that: In S4, the mean M of the distance data D is calculated:

8. The rolling bearing fault detection method based on DTW according to claim 1, characterized in that: In S5, the calculated mean value M is regarded as a fault indicator and compared with a preset threshold value T; If the fault indicator M is less than the preset threshold value T, it is considered that the bearing has no fault and the process returns to step S2 to continue fault detection; if the fault indicator M is greater than or equal to the preset threshold value T, it is considered that the bearing has fault.

Images