Patent Application
20200125123
The present invention relates generally to hydraulic positive displacement machines, and more particularly to monitoring the health status of a hydraulic positive displacement machine to predict failure and/or estimate remaining life.
When a hydraulic positive displacement machine, such as a hydraulic pump, fails, often the internal components of the pump are in pieces. Repair costs are significantly higher when a pump reaches this state. Furthermore, an unexpected failure can result in significant downtime. As an example, the unexpected failure of an aircraft pump can lead to a flight delay or cancellation, which can be extremely costly for the airline. Thus, the ability to monitor the health of a hydraulic pump and provide advance warning of an impending pump failure has many advantages.
Due to the costs associated with unexpected pump failures, airlines and aircraft manufacturers have attempted to monitor pump health in a number of different ways. One method of addressing pump failure is to replace all pumps after a certain number of flight hours. This approach assumes that the pump failure rate increases with flight hours, which is not always the case. Sometimes, the failure rate of a particular fielded pump model may decrease with flight hours (infant failures) or remain constant with flight hours. Furthermore, this approach does not monitor the health of the pump, only the statistical likelihood of pump failure as a function of flight hours. With no information on the condition of the pump, this method will necessarily lead to the premature removal of some operationally acceptable pumps, and the failure to identify some operationally unacceptable pumps.
Another method of addressing pump failure is to place temperature indicating tape on the housing of the pump. When the tape indicates a certain temperature has been exceeded, the pump is removed and replaced. A failing pump may be a reason for unusually high pump housing temperatures. However, there are many other factors that affect pump housing temperature, including ambient temperature and duty cycle of the pump. Furthermore, this technique requires visual inspection of the pump to determine whether or not the temperature limit has been exceeded.
Another method of addressing pump failure is to utilize an algorithm that monitors the temperature difference between pump case temperature and reservoir temperature. This is an improvement over simply using temperature indicating tape for monitoring the health of the pump, as it is not as sensitive to ambient temperature. However, there are many other factors that affect this temperature delta, including case drain flow and duty cycle. Additionally, only a limited number of pump failure modes lead to elevated temperatures.
A device and method in accordance with the present invention utilize frequency-domain analysis of dynamic pressure signal(s) to determine the health of a hydraulic machine. This technique provides a more direct means of monitoring the health of the machine, is not dependent on other external variables, and allows many of the most common hydraulic failure modes to be detected well in advance of a catastrophic failure.
According to one aspect of the invention, a system for predicting the health of a hydraulic machine includes: a dynamic pressure sensor configured to measure a dynamic fluid pressure corresponding to the hydraulic machine; and a controller communicatively coupled to the dynamic pressure sensor and operative to receive dynamic pressure data from the dynamic pressure sensor, the controller configured to convert the dynamic pressure data into the frequency domain, identify dynamic pressure amplitudes in the frequency domain, and generate an indication that the hydraulic machine is damaged or worn when dynamic pressure amplitudes in the frequency domain exceed a prescribed threshold and correspond to frequencies other than pumping frequencies.
In one embodiment, the system includes a speed sensor configured to measure a speed of the hydraulic machine, wherein the controller is communicatively coupled to the speed sensor and operative to receive speed data from the speed sensor, and the controller is further configured to determine at least one of a hydraulic fluid pumping frequency corresponding to the hydraulic machine or a shaft operating frequency corresponding to the hydraulic machine based on the speed data, and associate the dynamic pressure amplitudes with one of the hydraulic fluid pumping frequency or the shaft operating frequency.
In one embodiment, the system includes the hydraulic machine, wherein the pressure sensor is fluidically coupled to a fluid port of the hydraulic machine and the speed sensor is operatively coupled to the hydraulic machine.
In one embodiment, the controller is further configured to determine the hydraulic machine is normal when the amplitudes of the dynamic pressure signals at non-pumping frequencies are less than the prescribed thresholds.
In one embodiment, the controller is configured to determine the pumping frequency based on a speed of the hydraulic machine and a number of pumping elements of the hydraulic machine.
In one embodiment, the pressure sensor is in fluid communication with at least one of a discharge, a case or an inlet of the hydraulic machine.
In one embodiment, the pressure sensor is configured to collect pressure samples at a rate of at least 5 kHz.
In one embodiment, the controller is configured to analyze the dynamic pressure amplitudes in combination with hydraulic machine data other than the dynamic pressure amplitudes.
In one embodiment, the hydraulic machine data other than the dynamic pressure amplitudes comprises at least one of hydraulic machine temperature, fluid flow rate, or hydraulic machine efficiency.
In one embodiment, the controller is configured to apply a Fourier transform to the obtained dynamic pressure signals to transform the signals into the frequency domain.
In one embodiment, the controller is configured to: trend the amplitudes of the dynamic pressure signals at non-pumping frequencies over time; compare the trended values to a threshold level; and conclude the hydraulic machine is damaged or worn when the trended dynamic pressure amplitudes exceed a prescribed threshold level.
In one embodiment, the hydraulic machine comprises a hydraulic pump or a hydraulic motor.
According to another aspect of the invention, a method for predicting the health of a hydraulic machine includes: obtaining, using a pressure sensor, dynamic pressure signals corresponding to operation of the hydraulic machine; analyzing, in the frequency domain, dynamic pressure amplitudes in the obtained dynamic pressure signals; and concluding the hydraulic machine is damaged or worn when the amplitudes of the dynamic pressure signals at non-pumping frequencies exceed the prescribed thresholds.
In one embodiment, the method includes concluding the hydraulic machine is normal when the amplitudes of the dynamic pressure signals at non-pumping frequencies are less than the prescribed thresholds.
In one embodiment, a pumping frequency corresponds to the operating speed of the hydraulic machine in revolutions per second multiplied by the number of pumping elements of the hydraulic machine multiplied by any positive integer.
In one embodiment, obtaining dynamic pressure signals comprises obtaining the signals of at least one of a discharge, a case or an inlet.
In one embodiment, obtaining the pressure signals is performed during periods of steady-state hydraulic machine operation.
In one embodiment, obtaining the pressure signals comprises obtaining the pressure signals at a rate of at least 5 kHz.
In one embodiment, analyzing comprises analyzing the dynamic pressure amplitudes in combination with hydraulic machine data other than the dynamic pressure amplitudes.
In one embodiment, the hydraulic machine data other than the dynamic pressure amplitudes comprises at least one of hydraulic machine temperature, fluid flow rate, or hydraulic machine efficiency.
In one embodiment, the analyzing and concluding steps are performed in real time.
In one embodiment, the analyzing and concluding steps are performed time-shifted relative to the obtaining step.
In one embodiment, transforming comprises applying a Fourier transform to the obtained dynamic pressure signals.
In one embodiment, the method includes: trending the amplitudes of the dynamic pressure signals at non-pumping frequencies over time; comparing the trended values to threshold levels; and concluding the hydraulic machine is damaged or worn when the dynamic pressure amplitudes trend over prescribed threshold levels.
In one embodiment, the hydraulic machine comprises a hydraulic pump or a hydraulic motor.
To the accomplishment of the foregoing and related ends, the invention, then, comprises the features hereinafter fully described and particularly pointed out in the claims. The following description and the annexed drawings set forth in detail certain illustrative embodiments of the invention. These embodiments are indicative, however, of but a few of the various ways in which the principles of the invention may be employed. Other objects, advantages and novel features of the invention will become apparent from the following detailed description of the invention when considered in conjunction with the drawings.
Embodiments of this invention will now be described in further detail with reference to the accompanying drawings.
Embodiments of the present invention will now be described with reference to the drawings, wherein like reference numerals are used to refer to like elements throughout. It will be understood that the figures are not necessarily to scale.
The present invention has applicability to hydraulic machines (e.g., hydraulic pumps, hydraulic motors), such as swash-plate-type axial piston hydraulic pumps utilized in aircraft, and therefore will be described chiefly in this context. It should be appreciated, however, that aspects of the invention are applicable to any positive displacement machine and/or application in which it is desired to predict the health of such positive displacement machine.
Referring to
As the drive shaft 28 rotates, this effects rotation of the cylinder block 20. The rotation also causes reciprocating motion of each piston 24, where an angle of the swashplate 26 relative to a longitudinal axis of the drive shaft 28 determines the travel of each piston 24 within its respective bore 22 (and thus the fluid displacement of the pump 10). As each piston 24 moves outward, fluid is drawn into the respective bore 22 via the inlet port 22 and slot 18a of the valve body 12. With further rotation the piston 24 begins to move inward, expelling the fluid through the slot 18b of the valve body 12 and into the outlet port 16.
For an ideal axial piston pump, each pumping element (each piston 24 and corresponding bore 22 in the exemplary pump of
In accordance with the present invention, frequency-domain analysis of dynamic pressure signal(s) from the hydraulic machine taken during periods of steady-state operation are analyzed to determine the health of the hydraulic machine. The pressure signals may be obtained, for example, from the outlet port 16, within the case, or the inlet port 14. It should be appreciated, however, that the pressure signals can be collected anywhere along the discharge circuit, case circuit, or inlet circuit. For example, in aircraft, an “outlet” pressure transducer is typically located in the filter manifold, which may be fifteen feet downstream of the pump. Further, frequency-domain indicators from the dynamic pressure signals may be monitored alone or in combination with other signals from operation of the hydraulic machine, such as for example temperature, flow and machine efficiency, to assess the health of the hydraulic machine.
The device and method in accordance with the invention monitors the health of the hydraulic machine by analyzing, in the frequency-domain, dynamic pressure signal(s) produced by the hydraulic machine during periods of steady-state operation. Imbalance can be detected by analyzing certain frequency-domain indicators in the dynamic pressure data.
An example of a time-frequency domain representation of the dynamic pressure signal from a healthy hydraulic pump is shown in
The pumping frequencies may also be determined by analyzing the dynamic pressure measurements without knowledge of the rotational speed of the machine. For example, the first pumping frequency may be determined based on the number of pressure peaks in the dynamic pressure measurement over a prescribed time period.
An example of a time-frequency domain representation of the dynamic pressure signal from a damaged pump running at the same operating condition is shown in
To use this type of analysis for monitoring pump health, the dynamic pressure amplitudes of undesirable non-pumping frequencies can be trended over time during periods where the pump is operating at the same steady-state condition. One way to do this is to simply sum all dynamic pressure energy that exists at non-pumping frequencies, and trend this value. Borrowing the term “signal-to-noise ratio” from signal processing, where the pumping frequencies can be thought of as the signal, and all other frequencies can be thought of as noise, this summation of dynamic pressure amplitudes at non-pumping frequencies can be referred to as “noise power”. The trended values then can be compared to a threshold level such as, for example, a first threshold level indicative of “pump wear” or a second threshold level indicative of “imminent failure”. Based on the comparison, it can be concluded that the hydraulic machine is worn or damaged when the trended dynamic pressure amplitudes exceed the prescribed threshold level.
An example of the “noise power” trended over time for a failing pump can be seen in
When performing these calculations, care must be taken to differentiate between imbalance and control circuit instability. Control circuit instability can appear in the dynamic pressure data as a non-pumping frequency. However, instability is not an indication of pump wear/damage by itself. A simple way to avoid mistaking control circuit instability for imbalance is to apply a high pass filter to the dynamic pressure data to filter out any instability frequencies before calculating the noise power.
Another way to monitor the pump health using dynamic pressure data would be to track the amplitude of individual shaft order frequencies, without summing all non-pumping frequencies. By monitoring the amplitude of individual shaft frequencies, one may gain more insight into the specific failure mode, but at a cost of added complexity.
The setting of thresholds is the simplest way to use this type of analysis to monitor the health of a pump. For example, a warning can be set if the dynamic pressure noise power exceeds a certain level, or if the amplitude of any individual shaft order exceeds a certain level. This approach alone can be quite effective in predicting pump failure. However, more complexity can be added if other signals are available. By combining dynamic pressure data with temperature, flow, or efficiency data, the fidelity of this method may be improved. For example, a high dynamic pressure noise power reading in combination with an elevated temperature reading may be indicative of one failure mode, while a high noise power reading without elevated temperature may be indicative of a different failure mode.
This method is most effective when data are available from a large number of pumps, where thresholds can be determined by computing summary statistics of the field data. As more data are available, more sophisticated algorithms can be utilized. A pump health monitoring approach based on machine learning would utilize dynamic pressure noise power and/or shaft order amplitudes as features used to train a machine learning model.
Referring now to
In addition to the pressure sensor 56, the system 50 may optionally include a temperature sensor 60 for monitoring a temperature of the pump 10, a speed sensor 62 for monitoring a speed of the pump 10, and/or a flow sensor 64 for monitoring a fluid flow from the pump 10. The temperature sensor 60, speed sensor 62 and flow sensor 64 may be communicatively coupled to the controller 58 to provide the temperature, speed and flow measurements thereto. This additional data may be used in combination with the pressure data to refine the determination of pump wear/failure.
The exemplary controller 58 includes a processor 66 coupled to memory 68 and an I/O module 70 via a bus 72. The I/O module 70 receives data from the respective sensors and makes the data available to the processor 66. Application instructions may be stored in the memory 68 and executed by the processor 66 to monitor the health of a hydraulic pump 10 as described herein. Further, it is noted that the illustrated controller is merely exemplary, and other configurations are possible. For example, instead of a processor that executes code stored in memory the controller 58 may be implemented using an application-specific integrated circuit (ASIC), dedicated hardware, or the like.
Referring now to
Beginning at step 102, dynamic pressure signals corresponding to operation of the hydraulic machine are obtained. Such signals may be obtained, for example, from pressure sensor 56, which monitors fluid pressure in the system 50 at one of a machine outlet, a machine case or a machine inlet. In one embodiment, the dynamic pressure is obtained during periods of steady-state operation. The dynamic pressure signals obtained by the pressure sensor 56 may be sampled at a rate of 5 kHz, and more preferably at a rate of 10 kHz. It should be appreciated, however, that the required sample rate will depend on the speed of the pump (e.g., a pump running at a lower speed would not require as high of a sampling rate as a pump running at a higher speed).
Next at step 104, the obtained dynamic pressure signals are transformed into the frequency domain. Such transformation may be implemented using conventional techniques, such as a Fourier Transform. Once in the frequency domain, the dynamic pressure amplitudes are analyzed as indicated at step 106. In this regard, analyzing may include determining if the dynamic pressure amplitudes correspond to non-pumping frequencies, as indicated at step 108. Analyzing the dynamic pressure amplitudes may also include analyzing the dynamic pressure amplitudes in combination with pump data other than the dynamic pressure amplitudes. Such other data can include, for example, at least one of pump temperature, fluid flow rate, or pump efficiency.
At step 110, the non-pumping frequency dynamic pressure amplitudes may be compared to a threshold level as indicated at step 110. If the non-pumping frequency dynamic pressure amplitudes do not exceed the threshold level (or the only amplitudes that are present correspond to pumping frequencies), then at step 112 it can be concluded that the pump 10 is normal and the method moves back to step 102 and repeats. However, if the non-pumping frequency dynamic pressure amplitudes do exceed the threshold level, then at step 114 it can be concluded that the pump 10 is abnormal (e.g., damaged or worn) and an alarm maybe generated. The method then moves back to step 102 and repeats.
In one embodiment, the analyzing step is performed in real time by the controller 58. Thus, the entire method may be implemented in a single controller. In another embodiment, the analyzing and concluding steps are performed time-shifted relative to the obtaining step. For example, the controller 58 can collect the data and provide the data to an offline controller that performs the analysis (e.g., after each flight).
Accordingly, by monitoring the dynamic pressure amplitudes in the frequency domain, an accurate indication of hydraulic pump/motor condition can be ascertained. This enables maximum use of the pump/motor and minimizes costs for maintaining systems that utilize such hydraulic pumps/motors.
Although the invention has been shown and described with respect to a certain embodiment or embodiments, it is obvious that equivalent alterations and modifications will occur to others skilled in the art upon the reading and understanding of this specification and the annexed drawings. In particular regard to the various functions performed by the above described elements (components, assemblies, devices, compositions, etc.), the terms (including a reference to a “means”) used to describe such elements are intended to correspond, unless otherwise indicated, to any element which performs the specified function of the described element (i.e., that is functionally equivalent), even though not structurally equivalent to the disclosed member which performs the function in the herein illustrated exemplary embodiment or embodiments of the invention. In addition, while a particular feature of the invention may have been described above with respect to only one or more of several illustrated embodiments, such feature may be combined with one or more other features of the other embodiments, as may be desired and advantageous for any given or particular application.
This application claims the benefit of U.S. Provisional Application No. 62/747,257 filed Oct. 18, 2018, the contents of which are incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| 62747257 | Oct 2018 | US |
