Bearing monitoring apparatus and method

Abstract

A bearing monitoring apparatus including at least one interface for receiving strain signals detected by at least one strain sensor mounted on a bearing and at least one data processing unit configured to monitor the bearing by detecting bearing defects based on the strain signals is provided. The data processing unit is configured to subject the strain signal to a high pass filtering. A limit frequency of the high pass filter is set higher than a typical ball pass frequency of the bearing to be monitored, to calculate a quantity representative of the amplitude of the envelope signal of the high-pass filtered strain signal, to compare the peak hold of the envelope signal with an envelope threshold value, to classify the bearing using the result of the envelope comparison and to issue at least one warning signal if the bearing is classified as defective.

Claims

1. A bearing monitoring apparatus comprising: at least one interface for receiving strain signals detected by at least one strain sensor mounted in at least one circumferential groove in an outer ring of a bearing and at least one data processing unit configured to monitor the bearing by detecting bearing defects based on the strain signals, wherein the data processing unit is configured to: a subject the strain signal to a high pass filtering, wherein a limit frequency of the high pass filter is set higher than a typical ball pass frequency of the bearing to be monitored; b. calculate a quantity representative of an amplitude of an envelope signal of the high-pass filtered strain signal; c. compare the quantity representative of the amplitude of the envelope signal with an envelope threshold value; d. classify the bearing using the result of the envelope comparison; and e. issue at least one warning signal if the bearing is classified as defective; and wherein the data processing unit is further configured to: f. determine at least one fluctuation signal representative for a fluctuation width of the high-pass filtered strain signal; g. compare the at least one fluctuation signal with a fluctuation threshold value; and h. use the result of the fluctuation comparison in said classification of the bearing.

2. The bearing monitoring apparatus according to claim 1, wherein the data processing unit is further configured to: a. increment a defect detection hit counter if a peak hold of the envelope signal exceeds the envelope threshold value within a predetermined time span; b. increment the defect detection hit counter if the at least one fluctuation signal exceeds the fluctuation threshold value within a predetermined time span; and c. issue at least one warning signal if the defect detection hit counter exceeds at least one counter threshold value.

3. The bearing monitoring apparatus according to claim 1, wherein the quantity representative of the amplitude of the envelope signal data is a peak hold of the envelope signal, wherein a peak value is retained for a predetermined time corresponding to at least one rotation of a cage of the bearing to be monitored.

4. The bearing monitoring apparatus according to claim 1, wherein the at least one fluctuation signal representative for the fluctuation width of the high-pass filtered strain signal is the result of the application of an RMS filter to the high-pass filtered strain signal.

5. The bearing monitoring apparatus according to claim 4, wherein the averaging time of the RMS filter is set to a value corresponding to multiple rotations of a cage of the bearing to be monitored.

6. The bearing monitoring apparatus according to claim 1, wherein the data processing unit is further configured to compare a defect detection hit counter with multiple counter threshold values in order to classify a bearing defect status and to issue different warning signals as a function of the result of the classification.

7. The bearing monitoring apparatus according to claim 1, wherein at least one of the envelope threshold value and the fluctuation threshold value are set to correspond to at least 1.3 times the levels obtained for normally operating bearings.

8. The bearing monitoring apparatus according to claim 1, wherein the data processing unit is further configured to detect at least one out of the rotation speed and the load of the bearing to be monitored and to set at least one of the envelope threshold value and the fluctuation threshold value as a function of at least one of the load and the speed.

9. The bearing monitoring apparatus according to claim 1, wherein the at least one circumferential groove in the outer ring of the bearing comprises a first circumferential groove and a second circumferential groove that is lateral to the first circumferential groove.

10. The bearing monitoring apparatus according to claim 9, wherein each of the first and second circumferential grooves receive corresponding optical fibers wound around a circumference of the bearing.

11. The bearing monitoring apparatus according to claim 9, wherein the at least one strain sensor comprises a plurality of first sensors and a plurality of second sensors respectively housed in the first and second circumferential grooves.

12. The bearing monitoring apparatus according to claim 11, wherein a measuring length of each of the plurality of first sensors and the plurality of second sensors is 5 mm.

13. The bearing monitoring apparatus according to claim 11, wherein a spacing of the plurality of first sensors and the plurality of second sensors is a circumferential distance between the bearing and a rolling element.

14. A system comprising: a bearing, at least one strain sensor mounted in at least one circumferential groove in an outer ring of the bearing, and a bearing monitoring apparatus having; at least one interface for receiving strain signals detected by the at least one strain sensor and at least one data processing unit configured to monitor the bearing by detecting bearing defects based on the strain signals, wherein the data processing unit is configured to: a. subject the strain signal to a high pass filtering, wherein a limit frequency of the high pass filter is set higher than a typical ball pass frequency of the bearing to be monitored; b. calculate a quantity representative of an amplitude of an envelope signal of the high-pass filtered strain signal; c. compare the quantity representative of the amplitude of the envelope signal with an envelope threshold value; d. classify the bearing using the result of the envelope comparison; and e. issue at least one warning signal if the bearing is classified as defective; and wherein the data processing unit is further configured to: f. determine at least one fluctuation signal representative for a fluctuation width of the high-pass filtered strain signal; g. compare the at least one fluctuation signal with a fluctuation threshold value; and h. use the result of the fluctuation comparison in said classification of the bearing, receiving the signals of the at least one strain sensor via the interface.

15. The system according to claim 14, wherein the at least one strain sensor further comprises a fiber Bragg grating (FBG) strain gauge in an optical fiber connected to the interface of the bearing monitoring apparatus.

16. The system according to claim 14, further comprising multiple strain sensors distributed over a circumference of the bearing.

17. The system according claim 14, wherein the bearing is mounted in a deep sea water pump and the monitoring apparatus is located at or above sea level, wherein the sensor data are transmitted via at least one optical fiber.

18. A method for monitoring a bearing using strain signals detected by at least one strain sensor mounted in at least one circumferential groove in an outer ring of a bearing, the method comprising the steps of: a. subjecting the strain signal to a high pass filtering, wherein a limit frequency of the high pass filter is set higher than a typical ball pass frequency of the bearing to be monitored; b. calculating a quantity representative of an amplitude of an envelope signal of the high-pass filtered strain signal; c. comparing the quantity representative of the amplitude of the envelope signal with an envelope threshold value; d. classifying the bearing using the result of the envelope comparison; and e. issuing at least one warning signal if the bearing is classified as defective; f. determining a fluctuation signal representative for a fluctuation width of the high-pass filtered strain signal; g. comparing the fluctuation signal with a fluctuation threshold value; and h. using the result of the fluctuation comparison in said classification of the bearing.

Description

BRIEF DESCRIPTION OF THE FIGURES

(1) FIG. 1 is a schematic illustration of a system including a bearing monitoring apparatus and multiple bearings according to the invention.

(2) FIG. 2 is a schematic illustration of a deep-sea pump with bearings monitored by the bearing monitoring apparatus according to the invention.

(3) FIG. 3a is a schematic illustration of an outer ring of a bearing for use in a system according to the invention.

(4) FIG. 3b is a sectional view of a sensor region in a groove in the outer ring of FIG. 3a.

(5) FIG. 4 is a flow diagram of a defect detection method implemented in the bearing monitoring apparatus according to the invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

(6) FIG. 1 illustrates a system including a deep sea pump 22 with bearings 18 (FIG. 2) equipped with multiple sensors including strain sensors 14 and a bearing monitoring apparatus 10 located in a floating oil rig at sea level. The sensor signals are transmitted from the deep sea pump 22 to the bearing monitoring apparatus 10 by means of optical fibers 26 in a so-called umbilical 28 connecting the sea-level part of the system with the deep sea unit.

(7) At the sea-level side, the fibers 26 are connected to a data acquisition unit 24 forming the core part of the interface 12. Optoelectronic units are provided in the data acquisition unit 24 that illuminate the sensor network and record the optical reflection (wavelength change) from each discrete sensor 14 and output them as strains and temperatures.

(8) Stainless-steel tubes protect the fibers inside the pump 22. In the actual application, the optical fibers 26 are guided together with power supply lines and many control signals.

(9) The pump 22 is formed as multiphase twin screw pump 22 as schematically illustrated in FIG. 2. A mixture of oil, water and gas enters the pump 22 at two inlets at both sides of the rotors. Subsequently, the flow passes through a chamber created with interlocking screws 30a, 30b that moves the mixture along the length of the screws to the outlet. The outlet is situated at the top of the pump 22. The pump 22 is designed for subsea operation, i.e., several km below sea level. Therefore high boost pressures and consequently high radial bearing loads are produced.

(10) The bearing monitoring apparatus 10 installed in the sea level part of the system includes at least one interface 12 for receiving strain signals detected by the multiple strain sensor 14 mounted on the bearings 18 respectively. Further, the bearing monitoring apparatus 10 includes a data processing unit 20 configured to monitor the bearing 18 by detecting bearing defects based on the strain signals.

(11) An outer ring 32 of the bearing 18 adapted to be used in combination with a bearing monitoring apparatus 10 is illustrated in FIG. 3a and FIG. 3b.

(12) The outer ring 32 of the bearing 18 is provided with circumferential grooves 34a, 34b receiving glass fibers 26 provided with Fiber Bragg Gratings (FBG) equipping the bearings 18 with sensors 14. As illustrated in FIG. 3b, the optical fibers 26 are wound around the circumference of the bearing in the respective groove 34a, 34b. FBG sensors 18 are provided in both of the 2 lateral channels/grooves 34a, 34b of the bearing 18. The measuring length of each FBG strain sensor 14 set to 5 mm and the spacing of the sensors is equal to the circumferential distance between 2 rolling elements, in this case 37.9°. Though this is not illustrated in FIG. 3b, the glass fibres 26 may be guided from the outside into the grooves via a single lateral bore in the outer ring 32.

(13) The middle sensor is placed at the top dead centre (TDC), which is marked on the side of the outer ring of the bearing 18 in order to facilitate the assembly in the correct orientation.

(14) Further, a temperature sensor is placed at TDC in each raceway. The grooves 34a, 34b are sufficiently deep so that no component installed in the seat of the outer ring 32 in the channels/grooves 34a, 34b protrudes above the radially outer surface of the bearing 18. The components are therefore protected from damage.

(15) The sensors 14 forming strain gauges and the temperature sensors are connected to the optical fiber 26, which is in turn connected to the interface 12 of the bearing monitoring apparatus 10.

(16) The data processing unit 20 is configured to execute a data processing as illustrated in FIG. 4.

(17) In a step S1, a limit frequency of a subsequent high pass filter is set so as to be higher than a typical ball pass frequency of the bearing 18 to be monitored.

(18) The limit frequency is a multiple of the roller pass frequency detected by the data processing unit 20. The roller pass frequency is the most prominent frequency peak in the spectrum of the strain signal, which is accompanied by a set of harmoics of the fundamental roller pass frequency. Therefore the cut-off frequency of the high pass filter is set far above the harmonics. The factor between the roller pass frequency and the limit frequency of the high pass filter depends on how many harmonics are significant in the specific application.

(19) In general, at 20× the fundamental roller pass frequency, it is expected that no significant amount of basic roller pass signal is passed through.

(20) In step S2, the signal is subjected to the high-pass filter with a limit frequency as determined in step S1. The high-pass filter may be a 4th order Butterworth filter.

(21) After having passed the high-pass filter, the processing splits into two branches, e.g. by copying the signal. In a first branch illustrated on the left-hand side in FIG. 4, the data processing unit 20 calculates in step S3a a peak hold as a value representative of the amplitude of the envelope signal of the high-pass filtered strain signal. The peak hold of the envelope signal is then compared with an envelope threshold value in step S4a.

(22) In the second branch, the data processing unit 20 applies in a step S3b a root-mean-square (RMS)-filter to determine a fluctuation signal representative for the fluctuation width of the high-pass filtered strain signal. The averaging time of the RMS filter is set to a value corresponding to at least one rotation of a cage of the bearing 18 to be monitored, e.g. to 2 seconds, which is determined based on the roller pass frequency extracted in step S1. Then, this fluctuation signal is compared in step S4b with a fluctuation threshold value. As described further below, the result of the fluctuation comparison will be used in a classification of the bearing 18.

(23) In step S5, which might be combined with step S1, the data processing unit 20 calculates the threshold values to be used in steps S4a and S4b as a function of the load on the bearing and of the rotation speed. The data processing unit 20 reads the threshold values for the peak hold and for the fluctuation signal as a function of the load and/or of the speed from a storage unit or by using a predetermined characteristic. The rotation speed and the load of the bearing 18 to be monitored are calculated by the data processing unit 20 from a low-frequency part of the strain signal.

(24) In step S6, if in steps S4a and S4b one either the fluctuation signal or the envelope signal exceeds the pertinent threshold within a predetermined time span amounting e.g. to 2 seconds, the data processing unit 20 increments a defect detection hit counter. The hit counter is automatically decremented or set to zero if no hits are detected in a certain period.

(25) In step S7, the defect detection hit counter is then compared to several counter threshold values as described further below.

(26) The above predetermined time spans are chosen such that a reliable detection for the most relevant kinds of defects is ensured. While for an outer ring defect, if the outer ring is instrumented with strain gauges, a holding time corresponding to one rotation of the cage is enough, this may not be sufficient for the reliable detection for the other inner ring and roller defects.

(27) The detection of a defect relies on the alignment of the defect with the contact on inner or outer raceway and the sensor element, as the strain sensors 14 measure only very localized deformations. As the probability of the alignment decreases, the hold time needs to be longer.

(28) In the embodiment illustrated, the data processing unit 20 employs two different counter threshold values and uses a warning signal with green, yellow and red light similar to a traffic light.

(29) In step S8, if the value of the hit counter is below the first counter threshold value, the data processing unit 20 generates a signal corresponding to the green light indicating that everything is in good order and that the bearing 18 need not be replaced. If the value of the hit counter is greater or equal to the first counter threshold value and smaller than the second threshold value, the data processing unit 20 generates a signal corresponding to the yellow light indicating that there might be a defect and that a replacement of the bearing 18 is recommended, but not mandatory. If the value of the hit counter is greater or equal to the second counter threshold value, the data processing unit 20 generates a signal corresponding to the red lights indicating that there is a defect and that the bearing 18 is to be replaced. Depending on the field of application, the bearing monitoring apparatus 10 might trigger an emergency stop of the device or system using the bearing 18.

(30) The above comparison of the hit counter is a classification the bearing 18 using the result of the envelope comparison and of the fluctuation comparison. The signals corresponding to the different colors of light are warning signals. The red light signalizes to the user that bearing 18 is classified as defective.