Systems and methods for monitoring and diagnosing engine health using a snapshot-CEOD based approach

Abstract

Systems and methods are provided for monitoring and diagnosing the health of an engine. In one aspect, the system receives continuous operational data (COD) associated with an asset. The COD includes parameter values of one or more parameters during a collection time period. The system generates synthetic snapshot data at least in part based on the COD. The synthetic snapshot data includes one or more synthetic snapshots, each synthetic snapshot including parameter values of one or more parameters at a given time point during the collection time period. The system also receives snapshot data associated with the asset. The snapshot data includes one or more snapshots, each snapshot including parameter values of one or more parameters at a given time point. The system generates an output indicative of the health of the asset or one or more of its components at least in part based on the snapshot data and the synthetic snapshot data.

Description

Technical Field

[0001] The present subject matter generally relates to systems and methods for monitoring and diagnosing engine health. Background Art

[0002] Entities desire that their assets operate in an optimal manner while online, that they be offline for as short a time as possible, that all repairs / overhauls be scheduled (no unscheduled events / maintenance), and that failure events be avoided. For example, aircraft operators desire that their engines operate with high efficiency and high performance, spend as little time as possible outside the wing or off the aircraft, that all repairs and service visits be scheduled, and that major failure events of engine components be avoided. Understanding the health state of an engine and / or one or more of its components can help achieve these goals.

[0003] Understanding the health state of an engine and / or one or more of its components presents many challenges. For example, an aircraft gas turbine engine typically has a limited number of on-board sensors to measure or sense parameter values that may indicate the health state of the engine / component. Generally, the health state of an engine and / or its components is based on a limited number of data snapshots captured at various time points during the operation of the engine. Each snapshot includes the captured values of various parameters. The captured parameter values are fed into a health state module, and the output is the health state of the engine and / or one or more of its components. While the health state of a conventional module can provide insights into the engine health state, the health state may be based on only a limited number of snapshots. This limited number of snapshots may not provide the level of granularity required to reduce unscheduled engine removals (UERs) and major events, as well as planned / targeted repairs / maintenance for engine health issues to maximize flight time (TOW).

[0004] Accordingly, systems and methods that address one or more of the above challenges would be useful. Summary of the Invention

[0005] Aspects of the present disclosure relate to a distributed control system and method for controlling a turbine. Aspects and advantages of the present invention will be set forth in part in the description that follows, or may be apparent from the description, or may be learned by practice of the present invention.

[0006] In one aspect, a system is provided. The system includes one or more memory devices and one or more processors. The one or more processors are configured to: receive continuous operational data associated with an asset, the continuous operational data including parameter values of one or more parameters during a collection time period; generate synthetic snapshot data at least in part based on the continuous operational data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot containing parameter values of one or more parameters at a given point in time during the collection time period; receive snapshot data associated with the asset, the snapshot data including one or more snapshots, each snapshot containing parameter values of one or more parameters at a given point in time during the operation of the asset; and generate an output indicative of the health of the asset or one or more of its components at least in part based on the snapshot data and the synthetic snapshot data.

[0007] In another aspect, a method is provided. The method includes receiving, by one or more processors of a system, continuous operational data associated with an asset, the continuous operational data including parameter values of one or more parameters during a collection time period. Additionally, the method includes generating, by the one or more processors, synthetic snapshot data at least in part based on the continuous operational data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot containing parameter values of one or more parameters at a given point in time during the collection time period. Further, the method includes receiving, by the one or more processors, snapshot data associated with the asset, the snapshot data including one or more snapshots, each snapshot containing parameter values of one or more parameters at a given point in time during the operation of the asset. The method also includes generating, by the one or more processors, an output indicative of the health of the asset or one or more of its components at least in part based on the snapshot data and the synthetic snapshot data.

[0008] In a further aspect, a method is provided. The method includes receiving, by one or more processors of a system, continuous engine operation data associated with an aero gas turbine engine, the continuous engine operation data including parameter values of one or more parameters during a collection time period. The method further includes generating, by the one or more processors, synthetic snapshot data at least partially based on the continuous engine operation data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot including parameter values of one or more parameters at a given time point during the collection time period. The method further includes creating, by the one or more processors, one or more new snapshots by applying a machine learning model that utilizes one or more COD-snapshot transfer functions that associate the one or more synthetic snapshots with historical snapshot data associated with the aero gas turbine engine. Additionally, the method includes receiving, by the one or more processors, snapshot data associated with the gas turbine engine, the snapshot data including one or more snapshots, each snapshot including parameter values of one or more parameters at a given time point during the collection time period. Further, the method includes adding, by the one or more processors, the one or more new snapshots to the snapshot data. The method also includes generating, by the one or more processors, an output indicative of the health of the aero gas turbine engine or one or more of its components at least partially based on the one or more snapshots, new snapshots, and synthetic snapshots.

[0009] These and other features, aspects, and advantages of the present invention will become better understood with reference to the following description and appended claims. The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] A complete and enabling disclosure of the present invention, including the best mode thereof, to one of ordinary skill in the art, is set forth in the specification, which makes reference to the accompanying drawings, in which:

[0011] Figure 1 A schematic cross-sectional view of an aero gas turbine engine in accordance with an exemplary embodiment of the present subject matter is provided;

[0012] Figure 2 A block diagram of a system in accordance with an exemplary embodiment of the present subject matter is provided;

[0013] Figure 3 A block diagram of snapshot data in accordance with an exemplary embodiment of the present subject matter is provided;

[0014] Figure 4 A graph depicting the altitude of an aircraft having an asset installed thereon during flight as a function of time is provided and also shows snapshots captured during flight, where the snapshots are Figure 3 part of the snapshot data;

[0015] Figure 5 Provides a block diagram of continuous operation data (COD) according to an example embodiment of the present subject matter;

[0016] Figure 6 Provides a graph depicting the altitude of an aircraft with an asset installed thereon as a function of time during a flight of Figure 4 and also shows the time period for capturing the Figure 5 COD;

[0017] Figure 7 Shows composite snapshots generated at various time points during the COD collection time period of the flight depicted in Figure 4 and Figure 6 ;

[0018] Figure 8 Provides a block diagram of a snapshot creator module of a system depicting Figure 2 creating a new snapshot;

[0019] Figure 9 Provides a block diagram of a snapshot data health indicator module of a system depicting Figure 2 generating an alert score for various alerts indicating the health of an asset of Figure 2 or its components;

[0020] Figure 10 Provides a block diagram of a composite snapshot data health indicator module of a system depicting Figure 2 generating an alert score for various alerts indicating the health of an asset of Figure 2 or its components;

[0021] Figure 11 Provides a flowchart of an example method for monitoring and diagnosing the health of an asset according to an example embodiment of the present subject matter; and

[0022] Figure 12 Provides a schematic diagram of an example computing system for implementing one or more aspects of the present subject matter. Detailed Description

[0023] Reference will now be made in detail to the present embodiments of the invention, one or more examples of which are illustrated in the accompanying drawings. The detailed description uses numerical and alphabetical labels to refer to features in the drawings. In the drawings and the description, like or similar labels have been used to refer to like or similar parts of the invention. As used herein, the terms "first", "second", and "third" may be used interchangeably to distinguish one component from another, and are not intended to denote the position or importance of each component. The terms "upstream" and "downstream" refer to the relative flow direction with respect to the fluid flow in the fluid path. For example, "upstream" refers to the flow direction from which the fluid flows, and "downstream" refers to the flow direction towards which the fluid flows.

[0024] Aspects of the present disclosure are directed to systems and methods for monitoring and diagnosing the health of an asset, such as an aero gas turbine engine. The systems and methods provided herein utilize a method based on snapshot continuous operation data to determine the health of the asset. Conventionally, continuous operation data has not been used to generate a health status estimate.

[0025] During the operation of the asset, two types of data are captured, including snapshot data and continuous operation data (COD). Snapshot data is captured at different time points during the operation of the asset. That is, at a specific time point, a "snapshot" of the operating conditions of the asset is captured. The snapshot includes the values of various parameters at a specific time point during the operation of the asset. Continuous operation data is captured continuously during the operation of the asset. In particular, continuous operation data can be collected during a collection period (e.g., from the start to the end of a flight). The continuous operation data can include a large amount of data, which includes the captured values of various parameters during the collection period. One or more sensors associated with the asset can sense or measure the parameter values of the two types of data.

[0026] In one aspect, the system receives continuous operation data associated with the asset. The continuous operation data includes the parameter values of the parameters during the collection period. The system generates synthetic snapshot data at least in part based on the continuous operation data. The synthetic snapshot data includes one or more synthetic snapshots, each synthetic snapshot containing the parameter values of one or more parameters at a given time point during the collection period. In some embodiments, the system creates one or more new snapshots by applying a machine learning model that utilizes one or more COD-snapshot transfer functions that relate one or more synthetic snapshots to historical snapshot data associated with the aero gas turbine engine. In some embodiments, the system can create one or more new snapshots by applying a set of rules. In addition to receiving the continuous operation data, the system also receives snapshot data associated with the gas turbine engine, the snapshot data including one or more snapshots, each snapshot containing the parameter values of one or more parameters at a given time point during the collection period. The system adds the one or more new snapshots to the snapshot data.

[0027] The system applies one or more time series pattern recognition techniques to the snapshot data to determine at least one alert score associated with the snapshot data. The alert score associated with the snapshot data is determined at least in part based on one or more detected features associated with parameter values of one or more parameters of the snapshot data. Additionally, the system applies one or more time series pattern recognition techniques to the synthetic snapshot data to determine at least one alert score associated with the synthetic snapshot data. The alert score associated with the synthetic snapshot data is determined at least in part based on one or more detected features associated with parameter values of one or more parameters of the synthetic snapshot data. The system aggregates at least one alert score associated with the snapshot data and at least one alert score associated with the synthetic snapshot data into an aggregated alert score via a probability aggregation technique. The system generates an output indicative of the health of the asset or one or more of its components based at least in part on one or more snapshots, new snapshots, and synthetic snapshots, or more specifically, at least in part based on the aggregated alert score. In this way, the output indicative of the health of the asset is based on the received snapshot data and the received COD data.

[0028] Figure 1 A schematic cross-sectional view of an aero gas turbine engine according to an exemplary embodiment of the present subject matter is provided. In particular, Figure 1 An aero high bypass turbofan engine is provided, referred to herein as "turbofan engine 10". Figure 1 The turbofan engine 10 can be mounted to an aircraft (e.g., a fixed-wing aircraft) and can generate thrust for propelling the aircraft. For reference, the turbofan engine 10 defines an axial direction A, a radial direction R, and a circumferential direction. Additionally, for reference purposes, the turbofan engine 10 defines an axial centerline or longitudinal axis 12 that extends along the axial direction A. Generally, the axial direction A extends parallel to the longitudinal axis 12, the radial direction R extends outwardly and inwardly from the longitudinal axis 12 in a direction perpendicular to the axial direction A, and the circumferential direction extends 360 degrees (360°) around the longitudinal axis 12.

[0029] The turbofan engine 10 includes a core gas turbine engine 14 and a fan section 16 positioned upstream thereof. The core engine 14 includes a tubular outer casing 18 that defines an annular core inlet 20. The outer casing 18 further encloses and supports a booster or low-pressure compressor 22 that is operative to pressurize air entering the core engine 14 through the core inlet 20. A high-pressure, multi-stage, axial-flow compressor 24 receives the pressurized air from the LP compressor 22 and further increases the pressure of the air. The pressurized air flows downstream to a combustor 26 where fuel is injected into the pressurized air stream and ignited to increase the temperature and energy level of the pressurized air. The high-energy combustion products flow downstream from the combustor 26 to a high-pressure turbine 28 for driving the high-pressure compressor 24 via a high-pressure shaft 30 or a second rotatable member. The high-energy combustion products then flow to a low-pressure turbine 32 for driving the LP compressor 22 and the fan section 16 via a low-pressure shaft 34 or a first rotatable member. In this exemplary embodiment, the LP shaft 34 is coaxial with the HP shaft 30. After driving each of the turbines 28 and 32, the combustion products exit the core engine 14 through an exhaust nozzle 36 to produce a propulsive thrust.

[0030] The fan section 16 includes a rotatable axial-flow fan rotor 38 that is surrounded by an annular fan casing 40. The fan casing 40 is supported by the core engine 14 via a plurality of substantially radially-extending, circumferentially-spaced outlet guide vanes 42. In this manner, the fan casing 40 encloses the fan rotor 38 and a plurality of fan blades 44. A downstream section 46 of the fan casing 40 extends over an outer portion of the core engine 14 to define a bypass passage 48. The air passing through the bypass passage 48 provides a propulsive thrust, which will be further explained hereinafter. In some alternative embodiments, the LP shaft 34 may be connected to the fan rotor 38 in an indirect drive or gear drive configuration via a reduction device (such as a reduction gearbox). Such a reduction device may be included between any suitable shafts / spindles within the turbofan engine 10 as needed or required.

[0031] During operation of the turbofan engine 10, an initial or inlet air flow represented by arrow 50 enters the turbofan engine 10 through an inlet 52 defined by the fan casing 40. The air flow 50 passes through the fan blades 44 and divides into a first air flow (represented by arrow 54) that moves through the bypass passage 48 and a second air flow (represented by arrow 56) that enters the LP compressor 22 through the core inlet 20.

[0032] The pressure of the second air stream 56 is gradually increased by the LP compressor 22 and then enters the HP compressor 24, as shown by arrow 58. The discharged pressurized air stream flows downstream towards the burner 26, where fuel is introduced to produce combustion gases or products. The combustion products 60 leave the burner 26 and flow through the HP turbine 28. The combustion products 60 then flow through the LP turbine 32 and leave the exhaust nozzle 36 to produce thrust. Additionally, as described above, a portion of the incoming air stream 50 flows through the bypass passage 48 and through the outlet nozzle defined at the downstream section 46 of the fan casing 40 between the fan casing 40 and the outer casing 18. In this way, a large propulsive thrust is generated.

[0033] As Figure 1 Further shown therein, the burner 26 defines an annular combustion chamber 62 that is generally coaxial with the longitudinal centerline axis 12, as well as an inlet 64 and an outlet 66. The burner 26 receives an annular pressurized air stream from the high-pressure compressor discharge outlet 69. A portion of this compressor discharge air (“CDP” air) flows into a mixer (not shown). Fuel is injected from the fuel nozzle 68 to mix with the air and form a fuel-air mixture, which is provided to the combustion chamber 62 for combustion. Ignition of the fuel-air mixture is accomplished by a suitable igniter, and the resulting combustion gases 60 flow and enter the annular first-stage turbine nozzle 72 in the axial direction A. The nozzle 72 is defined by an annular flow passage including a plurality of radially extending, circumferentially spaced nozzle vanes 74 that turn the gases such that they flow at an angle and impinge on the first-stage turbine blades of the HP turbine 28. For this embodiment, the HP turbine 28 rotates the HP compressor 24 via the HP spool 30, and the LP turbine 32 drives the LP compressor 22 and the fan rotor 38 via the LP spool 34.

[0034] Although the turbofan engine 10 has been described and shown in Figure 1 as representing an example gas turbine engine, the subject matter of the present disclosure can be applied to other suitable types of engines and turbines. For example, the subject matter of the present disclosure can be applied to other suitable turbine engines (such as steam and other gas turbine engines) or in combination with other suitable turbine engines. Example gas turbine engines can include, but are not limited to, turbojet engines, turboprop engines, turboshaft engines, aero-derivative engines, auxiliary power units, etc. Additionally, as will be explained below, a gas turbine engine (such as the turbofan engine 10) is an example of an asset that can be monitored and diagnosed by the systems and methods described herein.

[0035] Figure 2 A block diagram of a system 100 according to an example embodiment of the present subject matter is provided. The system 100 can be used to monitor and diagnose the health of an asset 300 or a number of assets. In Figure 2In the illustrated embodiment, the asset 300 is Figure 1 a turbofan engine 10. However, it should be understood that the system 100 can be used to monitor and diagnose the health of other assets (e.g., any suitable machine, device, system, etc. that requires health monitoring). As an example, the asset can be the landing gear of an aircraft. As another example, the asset can be the main rotor of a rotary-wing aircraft. As yet another example, the asset can be the drill bit of a drill string used for oil and gas exploration. These examples are not intended to be limiting; other examples can be envisioned.

[0036] The asset 300 can include one or more associated sensors 302 for measuring or sensing, for example, the operating conditions of the asset 300 during operation of the asset 300. In particular, the sensors 302 can measure or sense the values of one or more parameters indicative of the operating conditions of the asset 300. Example parameters that can be recorded for a gas turbine engine include, but are not limited to, the low-pressure spool speed N1, the high-pressure or core spool speed N2, the compressor inlet pressure P2 and temperature T2, the compressor discharge pressure P3, and / or the temperature T3 at the inlet or the temperature T45 at the outlet of the combustor. Other example parameters can include altitude, airspeed, ambient temperature, weather conditions, etc. The values of other parameters can also be sensed.

[0037] Generally, one or more sensors 302 of the asset 300 capture two types of data, including snapshot data and continuous operation data (COD). COD is also referred to as continuous engine operation data (CEOD) in the field of aero gas turbine engines. These two data types will be further explained below.

[0038] Generally, snapshot data includes one or more snapshots or data snapshots captured at a given point in time or "time instant". Each snapshot contains the parameter values of one or more parameters at the given point in time. For example, referring to Figure 3 and Figure 4 , Figure 3 provides a block diagram of snapshot data 200 according to an example embodiment of the present subject matter. Figure 4 provides a graph depicting the altitude of an aircraft equipped with the asset 300 during flight as a function of time. Figure 4 Also shown is a snapshot captured during flight, where the snapshot is Figure 3A portion of the snapshot data 200. As shown, various snapshots are captured during the operation of the asset 300. Specifically, the snapshot data 200 includes a first snapshot S1 captured at time t1, a second snapshot S2 captured at time t2, and a third snapshot S3 captured at time t3. It should be understood that the snapshot data 200 may also include other data snapshots, represented by the Nth snapshot captured at time tN. The snapshot data 200 may include any suitable number of snapshots. As used herein, N represents any suitable integer.

[0039] Each of the snapshots S1, S2, S3, S N includes parameter values of one or more parameters at a given point in time. For example, the first snapshot S1 captured at time t1 includes parameter values PV1-1, PV2-1, PV3-1, PVN-1 corresponding to a first parameter P1, a second parameter P2, a third parameter P3, and an Nth parameter, respectively. All the parameter values of the first snapshot S1 are captured at time t1. Similarly, the second snapshot S2 captured at time t2 includes parameter values PV1-2, PV2-2, PV3-2, PVN-2 corresponding to the first parameter P1, the second parameter P2, the third parameter P3, and the Nth parameter PN, respectively. All the parameter values of the second snapshot S2 are captured at time t2. Additionally, the third snapshot S3 captured at time t3 includes parameter values PV1-3, PV2-3, PV3-3, PVN-3 corresponding to the first parameter P1, the second parameter P2, the third parameter P3, and the Nth parameter PN, respectively. All the parameter values of the third snapshot S3 are captured at time t3. Other snapshots may also include parameter values of the parameters, as shown in the Nth snapshot.

[0040] Snapshots can be captured based on a trigger condition. As an example, snapshots can be captured at a predetermined time interval. As another example, snapshots can be captured when the aircraft on which the asset 300 is installed reaches a predetermined altitude. For example, as Figure 4 shown, the aircraft operates in various flight phases (FPs) including takeoff FP1, climb FP2, cruise FP3, and descent and landing FP4. As shown, a predetermined altitude is set in this example such that the first snapshot S1 is captured during takeoff FP1, the second snapshot S2 is captured during climb FP2, and the third snapshot S3 is captured during cruise FP3. As described above, other snapshots can be captured. For example, a snapshot can be captured after the aircraft performs a step climb SC during cruise FP3 or when the aircraft reaches the predetermined altitude a second time (e.g., at some point during descent and landing FP4). In summary, the snapshot data 200 includes one or more data snapshots, each capturing the operating conditions of the asset 300 or some of its components at a given point in time.

[0041] Return Figure 2, as shown in the figure, the asset 300 may include or be associated with a communication unit 304. The sensed snapshot or snapshot data may be routed to the communication unit 304. The communication unit 304 may be a digital data link system or a communication management unit (CMU) of the aircraft on which the asset 300 is installed. Thus, in some embodiments, the communication unit 304 may be, for example, an aircraft communication, addressing, and reporting system (ACARS). The communication unit 304 may be directly mounted to the asset 300 or may be located remotely from the asset 300, for example, within the fuselage of the aircraft on which the asset 300 is installed. The communication unit 304 may include one or more processors, one or more memory devices, and a communication interface for communicating messages, alerts, etc. to a remote station. As an example, the snapshot data may be received and stored by the communication unit 304. The communication unit 304 may then transmit the snapshot data to a remote station, such as a ground station or another aircraft. Any suitable transmission technology and protocol may be used to transmit the snapshot data to the remote station. As will be further explained herein, the system 100 may receive the snapshot data and perform a health assessment of the asset 300 based on the received snapshot data.

[0042] In addition to the snapshot data, one or more sensors 302 of the asset 300 may capture COD, or in this example, capture CEOD. Generally, as the name implies, COD is captured continuously over a period of time (e.g., a COD collection period spanning from a start point to an end point). COD includes the parameter values captured for one or more parameters during that period. The parameter values may be captured at different capture rates (e.g., once per millisecond, once per second, once every three seconds, etc.). The parameter values of the parameters are captured in the form of frames or capture frames.

[0043] For example, referring to Figure 5 and Figure 6 , Figure 5 a block diagram of a COD 210 according to an example embodiment of the present subject matter is provided. Figure 6 A graph depicting the altitude of an aircraft on which the asset 300 is installed during flight as a function of time is provided. Figure 6 Also shown is the period of time during which the Figure 5 COD is captured. As Figure 5 shown, the COD 210 includes time-sequenced capture frames, including a first capture frame CF1, a second capture frame CF2, a third capture frame CF3, and so on up to an Nth capture frame CFN. The COD 210 may include any suitable number of capture frames.

[0044] Each capture frame CF1, CF2, CF3, CFN of COD 210 includes parameter values captured for one or more parameters. For example, the first capture frame CF1 has captured parameter values V1-1, V2-1, V3-1, VN-1 corresponding to the first parameter P1, the second parameter P2, the third parameter P3, and the Nth parameter, respectively. As the asset continues to operate, additional capture frames of parameter values are captured. Specifically, the second capture frame CF2 has captured parameter values V1-2, V2-2, V3-2, VN-2 corresponding to the first parameter P1, the second parameter P2, the third parameter P3, and the Nth parameter PN, respectively. In addition, the third capture frame CF3 has captured parameter values V1-3, V2-3, V3-3, VN-3 corresponding to the first parameter P1, the second parameter P2, the third parameter P3, and the Nth parameter PN, respectively. As will be appreciated, COD 210 may include more captured data frames than those represented by the Nth capture frame as shown in Figure 5 Some of the parameters for which values are captured for snapshot data and the parameters for which values are captured for COD data may be the same or overlapping. Thus, some of the values captured for snapshot data and the values captured for COD data may be the same or overlapping.

[0045] COD 210 may be captured over a period of time as described above. The COD collection period may span between a start point and an end point. As Figure 6 shown, in some embodiments, the start point may correspond to the time when the aircraft on which the asset 300 is installed takes off or just before takeoff. COD 210 may begin to be sensed and recorded when the asset 300 is powered on or spooling up, during takeoff roll, or at another suitable time point (e.g., just before or after takeoff). COD 210 may be sensed and recorded continuously for the entire flight or a portion of the flight as needed. For the Figure 6 flight depicted in

[0046] Returning again Figure 2, Asset 300 may include or be associated with a recorder 306 that records the COD 210. For example, an engine monitoring unit or “black box” of a vehicle on which Asset 300 is installed may record the sensed COD 210. The sensed COD 210 may be stored in one or more memory devices of the recorder 306. The recorder 306 may be installed on Asset 300 (e.g., installed on or under a fairing) or may be located remotely from Asset 300 (e.g., in an avionics bay of an aircraft on which Asset 300 is installed). Once recorded and stored in the recorder 306, the COD 210 is transmitted, routed, or otherwise moved to a remote station, e.g., via a wired communication link, wirelessly, or by some other method. The COD 210 may be moved to the remote station automatically or manually. Any suitable transmission technology and protocol may be used to transmit the COD 210 to the remote station. As will be further explained herein, System 100 may receive COD 210 in addition to snapshot data and perform a health assessment associated with Asset 300 based on the received snapshot data and COD data.

[0047] System 100 will now be described in detail. Generally, as described above, System 100 is operable to monitor and diagnose the health of Asset 300 or its components. System 100 may include, for example, one or more processing devices and one or more memory devices embodied in one or more computing devices and data storage devices. The one or more memory devices may store data and instructions accessible by the one or more processors, including computer-readable instructions executable by the one or more processors. The instructions may be any set of instructions that, when executed by the one or more processors, cause the one or more processors to perform operations (e.g., the operations described herein for monitoring and diagnosing the health of Asset 300).

[0048] As Figure 2As shown, system 100 receives snapshot data 200 associated with asset 300. The received snapshot data 200 is stored in a data storage device (such as snapshot data storage device 110). Additionally, system 100 receives COD 210 associated with asset 300. The received COD 210 is stored in a data storage device (such as COD data storage device 120). As COD 210 is stored, COD 210 is accessed, and at parameter calculation module 130, one or more processors of system 100 determine one or more values of additional parameters associated with asset 300 at least in part based on COD 210. As an example, as described above, COD 210 may include sensed parameter values of various parameters (such as pressures and temperatures at different stations of an engine, ambient temperature, shaft speed of an axis, etc.). The sensed values of these parameters can be used to calculate or determine values of other parameters associated with asset 300. For example, parameter values of parameters (such as exhaust gas temperature (EGT), engine pressure ratio, stall margin, various mass flows, various efficiencies, etc.) can be determined at least in part based on the sensed values. The calculated parameter values can be used to monitor and evaluate the health of asset 300. The calculated values of the parameters can be added to or otherwise included in COD 210.

[0049] At synthetic snapshot generator module 140, COD 210 including sensed and calculated values of parameters associated with asset 300 is processed. Specifically, one or more processing devices of system 100 can generate synthetic snapshot data 220 at least in part based on COD 210. The generated synthetic snapshot data may include one or more synthetic snapshots. Each synthetic snapshot may contain sensed parameter values and / or calculated parameter values of one or more parameters at a given time point within a time period (i.e., the time period spanning from the start point of COD collection to the end point). For example, Figure 7 Synthetic snapshots generated at various time points during the COD collection period are shown. Specifically, the first synthetic snapshot SN1 is generated at time point t1-SN, the second synthetic snapshot SN2 is generated at time point t2-SN, and the third synthetic snapshot SN3 is generated at time point t3-SN. By generating synthetic snapshots, the large amount of data contained in COD 210 can be broken down into manageable and more easily processable data points that can be used to monitor and diagnose the health of asset 300 or its components.

[0050] Synthetic snapshots can be generated at any suitable time point within the COD collection period. As an example, a synthetic snapshot can be generated at a time point that is midway between the times of snapshots captured as part of the snapshot data. As another example, synthetic snapshots can be generated at time points such that at least one synthetic snapshot is separated from each snapshot captured as part of the snapshot data by a predetermined time (e.g., 15 seconds). As yet another example, synthetic snapshots can be generated at time points corresponding to the maximum or minimum value of a particular parameter in one, some, or all of the flight phases. For example, for each flight phase, a synthetic snapshot can be generated at a time point corresponding to the maximum pressure at the outlet of the compressor of asset 300 during takeoff, a synthetic snapshot can be generated at a time point corresponding to the maximum pressure at the outlet of the compressor of asset 300 during climb, a synthetic snapshot can be generated at a time point corresponding to the maximum pressure at the outlet of the compressor of asset 300 during cruise, and so on. In some embodiments, multiple synthetic snapshots can be generated at time points within a given flight phase, some flight phases, or all flight phases. For example, a first synthetic snapshot corresponding to the maximum or minimum value of a first parameter in a given one flight phase can be generated at a first time point, and a second synthetic snapshot corresponding to the maximum or minimum value of a second parameter in the same flight phase can be generated at a second time point. Synthetic snapshots can also be generated at time points within the COD collection period based on other criteria and / or considerations.

[0051] In some embodiments, one or more of the synthetic snapshots can include parameter values from a single captured frame. In some embodiments, one or more of the synthetic snapshots are a given or multiple captured frames that are close to each other in time. For example, in certain cases, the parameter values of each parameter required to create the synthetic snapshot are sensed in a given captured frame. To obtain the parameter values of the required parameters, the parameter values from temporally adjacent or temporally nearest captured frames can be utilized to generate the synthetic snapshot.

[0052] As Figure 2 shown, for this embodiment, the synthetic snapshot data generated including the synthetic snapshots is routed to two modules, including the snapshot creator module 150 and the synthetic health module 170. In some alternative embodiments, the system 100 does not include the snapshot creator module 150. Thus, in such alternative embodiments, the synthetic snapshot data is routed to the synthetic health module 170 instead of the snapshot creator module 150.

[0053] The snapshot creator module 150 is used as a feature generation tool. Specifically, the synthetic snapshot data 220 generated at the synthetic snapshot generator module 140 is input into the snapshot creator module 150. The snapshot creator module 150 then creates one or more new snapshots using the synthetic snapshot data 220. The one or more new snapshots can be added to the snapshot data 200 stored in the snapshot data storage device 110. In this way, the new snapshots can enhance or augment the snapshot data. The increased number of data points can increase the confidence of the health alerts provided by the system 100.

[0054] The snapshot creator module 150 creates new snapshots by applying one or more machine learning models that utilize one or more COD-snapshot transfer functions that associate one or more synthetic snapshots with historical snapshot data associated with an asset. In some embodiments, the historical snapshot data includes snapshot data for the most recent flight or operation cycle received at the snapshot data storage device 110. The snapshot creator module 150 can include instructions, models, functions, etc. One or more processors of the system 100 can execute the instructions to implement the models, functions, etc., and ultimately create new snapshots.

[0055] As an example, Figure 8 A block diagram depicting the creation of new snapshots by the snapshot creator module 150 of the system 100 is provided. As shown, the synthetic snapshot data 220 is input into the snapshot creator module 150. The synthetic snapshot data 220 includes a first synthetic snapshot SN1, a second synthetic snapshot SN2, a third synthetic snapshot SN3, and so on up to an Nth synthetic snapshot SNN. As described above, each synthetic snapshot SN1, SN2, SN3, SNN can include parameter values at a given point in time. As Figure 8As shown, synthetic snapshots SN1, SN2, SN3, SNN are fed into machine learning model 152. In particular, synthetic snapshots SN1, SN2, SN3, SNN are fed into the COD-snapshot transfer function of machine learning model 152. In this example, one or more COD-snapshot transfer functions include a first COD-snapshot transfer function TF1, a second COD-snapshot transfer function TF2, a third COD-snapshot transfer function TF3, and so on up to an Nth COD-snapshot transfer function. One or more COD-snapshot transfer functions TF1, TF2, TF3, TFN associate one or more synthetic snapshots SN1, SN2, SN3, SNN with historical snapshot data associated with asset 300. In this way, the input (in this example, the input is the parameter values of synthetic snapshots SN1, SN2, SN3, SNN) can be used by transfer functions TF1, TF2, TF3, TFN to generate an output (in this example, the output is new snapshot data 230), where new snapshot data 230 includes a first new snapshot NS1, a second new snapshot NS2, a third new snapshot NS3, and so on up to an Nth new snapshot NSN.

[0056] Machine learning model 152, or more specifically the COD-transfer function, can be trained based on historical snapshot data points. The COD-transfer function can be trained based on COD and snapshot data obtained by system 100 and can be periodically retrained when new data is obtained. In some cases, snapshot data from flight can be used to train / retrain the transfer function, generating the COD for generating synthetic snapshot data from that flight before creating new data points. In this way, the transfer function can be trained using the latest data. Additionally, machine learning model 152 can be trained at least in part based on historical snapshot data associated with asset 300 and fleet historical snapshot data associated with other assets that have the same model as asset 300. For example, fleet historical snapshot data can include snapshot data captured by other aero gas turbine engines during their respective flights. In some alternative embodiments, the snapshot creator module 150 creates new snapshots by applying a set of rules rather than on a trained machine learning model.

[0057] Return Figure 2, in the case where a new snapshot is created, the new snapshot is added to the snapshot data 200 stored in the snapshot data storage device 110. The snapshot data 200, including the snapshots captured or calculated from the sensors 302 of the asset 300 and the new snapshots created by the snapshot creator module 150, is input into the snapshot health module 160. And as described above, the synthetic snapshot data 220 generated at the synthetic snapshot generator module 140 is input into the synthetic health module 170. Generally, the snapshot health module 160 and the snapshot health module 160 apply one or more time series pattern recognition techniques or anomaly detection techniques, which use the data they each receive to output alert scores for various alerts indicating the health of the asset 300 or one or more of its components.

[0058] As an example, Figure 9 A block diagram is provided depicting the snapshot health module 160 of the system 100 generating alert scores for various alerts indicating the health of the asset 300 or its components. As depicted, the snapshot data 200 is input into the snapshot health module 160. The snapshot data 200 includes snapshots (e.g., Figure 3 S1, S2, S3, SN as shown) and new snapshots (e.g., Figure 8 NS1, NS2, NS3, NSN as shown). One or more time series pattern recognition techniques 164 can be applied or utilized to detect certain features in the input snapshot data 200. For example, one or more time series pattern recognition techniques 164 can be used to detect trends, offsets, changes, or otherwise detect anomalies and other features in certain sensed or calculated parameter values in the input snapshot data 200. Any suitable number of features can be considered.

[0059] For example, in Figure 9 , the features 162 include a first feature F1, a second feature F2, a third feature F3, a fourth feature F4, and so on up to an Nth feature FN. As an example, the first feature F1 can be associated with detecting an offset of a first parameter, the second feature F2 can be associated with detecting a trend of the first parameter, the third feature F3 can be associated with detecting an offset of a second parameter, and the fourth feature F4 can be associated with detecting a trend of the second parameter. Other features can be associated with detecting offsets and / or trends in other parameters of the snapshot data.

[0060] As described above, one or more time series pattern recognition techniques 164 or anomaly detection techniques can be applied to detect features 162 in the received snapshot data 200. In particular, the applied time series pattern recognition technique 164 can be used to determine whether a given feature (e.g., an offset or trend associated with a parameter) exceeds a predetermined threshold. For example, the predetermined threshold can be set or determined based on historical data. The predetermined threshold can be any suitable type or combination of thresholds. For example, the predetermined threshold can be a rate of change threshold, an offset threshold, a maximum and / or minimum threshold, a trend threshold, etc.

[0061] An alert score for an alert can be generated at least in part based on whether one or more features associated with the alert exceed their respective thresholds. One or more features can be associated with a given alert. An alert score can be generated for each alert. In some embodiments, the alert score can be a binary score. For example, when one or more features associated with an alert exceed their respective thresholds, an alert score of "1" can be generated for the alert. When one or more features associated with an alert do not exceed their respective thresholds, an alert score of "0" can be generated for the alert.

[0062] As Figure 9 shown, for example, alert scores are generated for four alerts, including a first alert (Alert 1), a second alert (Alert 2), a third alert (Alert 3), and a fourth alert (Alert 4). Although four alerts are shown in Figure 9 , any suitable number of alerts can be considered and thus have generated alert scores. The alerts can indicate the health of the asset 300 or its components. As shown, for this example, an alert score of "1" has been output for Alert 1 because its associated features have exceeded their respective thresholds, an alert score of "1" has been output for Alert 2 because its associated features have exceeded their respective thresholds, an alert score of "0" has been output for Alert 3 because its associated features have not exceeded their respective thresholds, and an alert score of "1" has been output for Alert 4 because its associated features have exceeded their respective thresholds. Thus, three of the four alerts have output a score of "1", indicating that their respective one or more associated features have exceeded their respective thresholds. The alert score output for an alert can be forwarded to the Figure 2 aggregator 180 shown in

[0063] Figure 10 A block diagram is provided depicting the synthetic health module 170 of the system 100 generating alert scores for various alerts indicating the health of the asset 300 or its components. As shown, synthetic snapshot data 220 is input into the synthetic health module 170. The synthetic snapshot data 220 includes synthetic snapshots, such as Figure 8SN1, SN2, SN3, SNN shown in [figure]. One or more time series pattern recognition techniques 174 can be applied or utilized to detect certain features in the input synthetic snapshot data 220. For example, one or more time series pattern recognition techniques 174 can be used to detect trends, offsets, changes, or otherwise detect anomalies and other features in certain sensed or computed parameter values in the input synthetic snapshot data 220. Any suitable number of features can be considered.

[0064] For example, in Figure 10 the features 172 include a first feature F1, a second feature F2, a third feature F3, a fourth feature F4, and so on up to an Nth feature FN. As an example, the first feature F1 can be associated with detecting an offset of a first parameter, the second feature F2 can be associated with detecting a trend of the first parameter, the third feature F3 can be associated with detecting an offset of a second parameter, and the fourth feature F4 can be associated with detecting a trend of the second parameter. Other features can be associated with detecting offsets and / or trends in other parameters of the snapshot data. The features of the synthetic health module 170 can be the same features as those used in the Figure 9 snapshot health module 160 of [[figure]].

[0065] As described above, one or more time series pattern recognition techniques 174 or anomaly detection techniques can be applied to detect the features 172 in the received synthetic snapshot data 220. Specifically, the applied time series pattern recognition technique 174 can be used to determine whether a given feature (e.g., an offset or trend associated with a parameter) exceeds a predetermined threshold. For example, the predetermined threshold can be set or determined based on historical data. The predetermined threshold can be any suitable type of threshold or combination of thresholds. For example, the predetermined threshold can be a rate of change threshold, an offset threshold, a maximum and / or minimum threshold, a trend threshold, etc. One or more time series pattern recognition techniques 174 applied to the features 172 can be the same techniques as those applied to the Figure 9 features 162 of the snapshot health module 160 shown in [[figure]].

[0066] In addition, as described above, an alert score for an alert can be generated at least in part based on whether one or more features associated with the alert exceed their respective thresholds. One or more features can be associated with a given alert. An alert score can be generated for each alert. In some embodiments, the alert score can be a binary score. For example, when one or more features associated with an alert exceed their respective thresholds, an alert score of "1" can be output for the alert. When one or more features associated with an alert do not exceed their respective thresholds, an alert score of "0" can be output for the alert.

[0067] As Figure 10As shown, for example, alert scores are generated for four alerts, including a first alert (Alert 1), a second alert (Alert 2), a third alert (Alert 3), and a fourth alert (Alert 4). Although four alerts are shown in Figure 10 , any suitable number of alerts may be considered. The alerts may indicate the health of the asset 300 or its components. As shown, for this example, an alert score of "0" has been output for Alert 1 because its associated features have not exceeded their respective thresholds, an alert score of "1" has been output for Alert 2 because its associated features have exceeded their respective thresholds, an alert score of "0" has been output for Alert 3 because its associated features have not exceeded their respective thresholds, and an alert score of "1" has been output for Alert 4 because its associated features have exceeded their respective thresholds. Thus, two of the four alerts have a score of "1", indicating that one or more of their respective associated features have exceeded their respective thresholds, and two of the four alerts have a score of "0", indicating that one or more of their respective associated features have not exceeded their respective thresholds. The alert scores generated for the alerts may be forwarded to Figure 2 the aggregator 180 shown in

[0068] As Figure 2 shown, the aggregator 180 receives the alert scores generated by the snapshot health module 160 and the alert scores generated by the synthetic health module 170. The aggregator 180 may utilize probability aggregation techniques to generate an output indicating the health of the asset 300 or its components, at least in part, based on the alert scores generated by the snapshot health module 160 and the alert scores generated by the synthetic health module 170. In particular, the aggregator 180 may aggregate the alert scores associated with the snapshot data and the alert scores associated with the synthetic snapshot data into an aggregated alert score via probability aggregation techniques. The output indicating the health of the asset or one or more of its components may be generated, at least in part, based on the aggregated alert score. The aggregator 180 may assign a weight value to the alert scores and may generate an output based on the alert scores and the weight values assigned to the alerts. For example, one of the alerts may be a primary indicator of health and may thus carry more weight in determining the output. In some embodiments, the aggregator 180 may generate an output indicating the health as one or more time trend graphs that trend or predict certain parameters.

[0069] The generated output indicating the health status of asset 300 can be used by system 100 or some other system to perform control actions. As an example, an electronic engine controller (EEC) of a gas turbine engine (asset) can control the gas turbine engine at least in part based on the health status of components of the engine or its modules (e.g., a compressor). For example, the EEC can control the gas turbine engine to operate more or less aggressively based on the health status. As another example, a maintenance system can receive the health status and schedule service visits automatically at least in part based on the output health status associated with the asset. Other control actions are envisioned; the examples provided above are not intended to be limiting.

[0070] Figure 11 A flowchart of an example method (400) for monitoring and diagnosing the health status of an asset in accordance with an example embodiment of the present subject matter is provided. Figure 11 The method (400) can be implemented using, for example, system 100 and / or its components described herein. For purposes of illustration and discussion, Figure 11 acts are depicted as being performed in a particular order. Those of ordinary skill in the art using the disclosure provided herein will understand that the various acts disclosed herein can be modified in various ways without departing from the scope of the present disclosure.

[0071] At (402), the method (400) includes receiving, by one or more processors of a system, continuous operation data associated with an asset, the continuous operation data including parameter values of one or more parameters during a collection time period. For example, the asset can be a gas turbine engine, such as an aircraft gas turbine engine. Thus, for example, the continuous operation data can be continuous engine operation data. The system can be an engine monitoring and health diagnostic system. The continuous operation data can be collected during a collection time period spanning from a start point to an end point, e.g., as Figure 6 shown. In this way, continuous operation data for an entire flight can be collected. In some cases, continuous operation data for a partial flight can be collected. The continuous operation data can be collected continuously during the collection time period. Values of various parameters associated with the engine can be sensed and recorded. For example, a recorder associated with the engine can record values of one or more parameters frame by frame (or in other words, per captured frame). The parameter values can be captured in each frame, or at time intervals in different captured frames. Figure 5 An example block diagram of the continuous operation data is provided. The system can receive the continuous operation data and can store the continuous operation data in a data storage device accessible by one or more processors of the system.

[0072] In some embodiments, method (400) includes determining, by one or more processors, one or more values of additional parameters associated with an asset, the one or more values of the additional parameters being determined using continuous operational data. For example, as Figure 2 shown, the continuous operational data stored in the COD data store 120 can be accessed and one or more processors of the system 100 can execute the parameter calculation module 130 to determine one or more values of the additional parameters using the continuous operational data. For example, the parameter calculation module 130 can include computer-readable instructions and one or more physics-based models. The one or more values of the additional parameters of the continuous operational data can be used to generate synthetic snapshot data at (404).

[0073] At (404), method (400) includes generating, by one or more processors, synthetic snapshot data at least in part based on the continuous operational data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot including parameter values of one or more parameters at a given point in time within a collection time period. For example, synthetic snapshots can be generated at Figure 7 the time points shown. Any suitable number of synthetic snapshots can be generated. One or more processors can execute the synthetic snapshot generator module 140( Figure 2 ) to generate the synthetic snapshots. By generating the synthetic snapshots, the large amount of data included in the continuous operational data can be broken down into manageable and more easily processed data points that can be used to monitor and diagnose the health of an asset (such as an aero gas turbine engine) or its components. In some embodiments, at least one of the one or more synthetic snapshots is created at a point in time within a defined operational phase (e.g., the climb phase of a flight) of the collection time period corresponding to a maximum or minimum value of a given parameter among the one or more parameters.

[0074] At (406), method (400) includes receiving, by one or more processors, snapshot data associated with an asset, the snapshot data including one or more snapshots, each snapshot containing parameter values of one or more parameters at a given point in time during operation of the asset. In particular, for each operating cycle of the asset, one or more sensors associated with the asset may capture a "snapshot" of the operating conditions at a specific point in time or timestamp during operation. Each captured snapshot may include values of various parameters (such as pressure, temperature, speed, etc.). As an example, the snapshot data may include at least one snapshot in each predefined operating phase of the asset, e.g., at least one snapshot in each flight phase. In some embodiments, one, some, or all of the snapshots may be captured during a collection period for recording COD. In other embodiments, one, some, or all of the snapshots may be captured during an operating period of the asset that is not during the collection period. Additionally, in some cases, all time points associated with the snapshots may be different from the time points associated with the synthetic snapshots. However, in other cases, one or more of the synthetic snapshots may correspond in time to one of the snapshots. Thus, the snapshots and the synthetic snapshots may contain parameter values of one or more parameters at the same point in time. In this regard, in some embodiments, the accuracy of the synthetic snapshot or a general synthetic snapshot may be checked against the actual snapshots.

[0075] In some embodiments, method (400) includes creating, by one or more processors, one or more new snapshots by applying a machine learning model that utilizes one or more COD-snapshot transfer functions that associate one or more synthetic snapshots with historical snapshot data associated with the asset. For example, as Figure 8 shown, synthetic snapshot data including the generated synthetic snapshots may be fed or input into a machine learning model that utilizes one or more COD-snapshot transfer functions that associate one or more synthetic snapshots with historical snapshot data associated with the asset. In particular, the COD-snapshot transfer function may utilize the parameter values of the synthetic snapshot and create a new data snapshot. The machine learning model may be trained at least in part based on historical snapshot data associated with the asset and fleet historical snapshot data associated with other assets that have the same model as the asset. Additionally, method (400) may include adding, by one or more processors, one or more new snapshots to the snapshot data, where the one or more new snapshots are added to the snapshot data before generating an output indicative of the health of the asset or one or more of its components.

[0076] At (408), method (400) includes generating, by one or more processors, an output indicative of the health of an asset or one or more of its components based at least in part on snapshot data and synthetic snapshot data. Thus, the health of the engine is generated using a snapshot-COD based method. That is, the output indicative of health is generated based on snapshot data and synthetic snapshot data, the synthetic snapshot data being generated based on COD. Further, in embodiments where a new snapshot is created based at least in part on the generated synthetic snapshot data, the output indicative of the health of an aero gas turbine engine or one or more of its components is based at least in part on one or more snapshots, the new snapshot, and the synthetic snapshot.

[0077] In some embodiments, to ultimately generate the output, method (400) includes applying, by one or more processors, one or more time series pattern recognition techniques to the snapshot data to determine at least one alert score associated with the snapshot data, the at least one alert score associated with the snapshot data being determined based at least in part on one or more detected features associated with parameter values of one or more parameters of the snapshot data. Further, method (400) includes applying, by one or more processors, one or more time series pattern recognition techniques to the synthetic snapshot data to determine at least one alert score associated with the synthetic snapshot data, the at least one alert score associated with the synthetic snapshot data being determined based at least in part on one or more detected features associated with parameter values of one or more parameters of the synthetic snapshot data. In such embodiments, the output indicative of the health of an asset or one or more of its components is generated based at least in part on the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data.

[0078] Further, in some embodiments, method (400) includes aggregating, by one or more processors, the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data into an aggregated alert score via a probability aggregation technique. In such embodiments, the output indicative of the health of an asset or one or more of its components is generated based at least in part on the aggregated alert score.

[0079] In some embodiments, system (e.g., Figure 2System 100) or a second system associated with the system can perform control actions at least in part based on the output indicating the health of the asset or one or more of its components. As an example, the EEC of an engine can control a gas turbine engine at least in part based on the health of the engine or its modular components (e.g., a compressor). For example, the EEC can control the gas turbine engine to operate more or less aggressively based on the health condition. As another example, a maintenance system can receive the health condition and automatically schedule a service visit at least in part based on the outputted health condition associated with the asset.

[0080] The systems and methods disclosed herein provide many technical and commercial advantages and benefits. For example, the systems and methods disclosed herein can provide an automatic health assessment of an asset or one or more components of an asset (e.g., a compressor). Such automatic assessment can be provided in real time or near real time. Additionally, by using COD in combination with snapshot data, the health state of an asset or one or more of its components can be improved compared to conventional methods and systems. Further, the output indicating the health state can provide a basis for optimal asset utilization and planning engine disassembly / maintenance activities. Additionally, the systems and methods disclosed herein provide a non-invasive technique for determining the health state. Additionally, the systems and methods disclosed herein provide an opportunity to identify asset-specific maintenance requirements in service and reduce UER and major failure events while enabling asset missions and maximizing TOW. Additionally, the systems and methods disclosed herein provide an opportunity to modify asset usage to maximize the service value of the asset.

[0081] Figure 12 A block diagram of an example computing system 500 according to an example embodiment of the present subject matter is provided. The example computing system 500 can be used to implement the methods and systems described herein. The computing system 500 is one example of a suitable computing system for the computing elements implementing the system 100 described herein.

[0082] As Figure 12 shown, the computing system 500 includes one or more computing devices 502. The one or more computing devices 502 can include one or more processors 504 and one or more memory devices 506. The one or more processors 504 can include any suitable processing device, such as a microprocessor, a microcontroller, an integrated circuit, a logic device, or other suitable processing device. The one or more memory devices 506 can include one or more computer-readable media, including but not limited to non-transitory computer-readable media or media, RAM, ROM, hard disk drives, flash drives, and other memory devices, such as one or more buffer devices.

[0083] One or more memory devices 506 may store information accessible by one or more processors 504, including computer-readable instructions 508 executable by one or more processors 504. The instructions 508 may be any set of instructions that, when executed by one or more processors 504, cause one or more processors 504 to operate. The instructions 508 may be software written in any suitable programming language or may be implemented in hardware. The instructions 508 may be any of the computer-readable instructions described herein. Each module mentioned herein may include associated computer-readable instructions.

[0084] The memory device 506 may further store data 510 accessible by the processor 504. For example, the data 510 may include received data, such as COD and snapshot data. According to an example embodiment of the present subject matter, the data 510 may include one or more tables, functions, algorithms, models, equations, etc.

[0085] One or more computing devices 502 may further include a communication interface 512 for communicating, for example, with other components or systems such as maintenance systems, aircraft systems, etc. The communication interface 512 may include any suitable components for interfacing with one or more networks, including, for example, transmitters, receivers, ports, controllers, antennas, or other suitable components.

[0086] The techniques discussed herein refer to computer-based systems, actions taken by computer-based systems, information sent to computer-based systems, and information from computer-based systems. It should be understood that the inherent flexibility of computer-based systems allows for many possible configurations, combinations, and divisions of tasks and functions among and within components. For example, the processing discussed herein may be implemented using a single computing device or multiple computing devices working in combination. Databases, memories, instructions, and applications may be implemented on a single system or distributed across multiple systems. Distributed components may operate sequentially or in parallel.

[0087] Although specific features of various embodiments may be shown in some figures and not in others, this is merely for convenience. Any feature of any figure may be referenced and / or claimed in combination with any feature of any other figure in accordance with the principles of the present disclosure.

[0088] The written description uses examples to disclose the invention, including the best mode, and also enables any person skilled in the art to practice the invention, including making and using any device or system and performing any combined method. The scope of the patent for the invention is defined by the claims, and may include other examples that occur to those skilled in the art. If these other examples include structural elements that are not different from the literal language of the claims, or if they include equivalent structural elements that do not have a substantial difference from the literal language of the claims, then these other examples are intended to fall within the scope of the claims.

[0089] A further aspect of the invention is provided by the subject matter of the following clauses:

[0090] 1. A system, comprising: one or more memory devices; and one or more processors configured to: receive continuous operational data associated with an asset, the continuous operational data including parameter values of one or more parameters during a collection time period; generate synthetic snapshot data at least in part based on the continuous operational data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot including the parameter values of the one or more parameters at a given point in time during the collection time period; receive snapshot data associated with the asset, the snapshot data including one or more snapshots, each snapshot including parameter values of the one or more parameters at a given point in time during operation of the asset; and generate an output indicative of the health of the asset or one or more of its components at least in part based on the snapshot data and the synthetic snapshot data.

[0091] 2. The system according to any of the preceding clauses, wherein the one or more processors are further configured to: create one or more new snapshots by applying a machine learning model that utilizes one or more COD-snapshot transfer functions that relate the one or more synthetic snapshots and historical snapshot data associated with the asset.

[0092] 3. The system according to any of the preceding clauses, wherein the one or more processors are further configured to: add the one or more new snapshots to the snapshot data, wherein the one or more new snapshots are added to the snapshot data before generating the output indicative of the health of the asset or one or more of its components.

[0093] 4. The system according to any of the preceding clauses, wherein the machine learning model is trained at least in part based on the historical snapshot data associated with the asset and fleet historical snapshot data associated with other assets having the same model as the asset.

[0094] 5. The system according to any of the preceding clauses, wherein the one or more processors are further configured to: apply one or more time series pattern recognition techniques to the snapshot data to determine at least one alert score associated with the snapshot data, the at least one alert score associated with the snapshot data being determined at least in part based on one or more detected features associated with the parameter values of the one or more parameters of the snapshot data; and apply one or more time series pattern recognition techniques to the synthetic snapshot data to determine at least one alert score associated with the synthetic snapshot data, the at least one alert score associated with the synthetic snapshot data being determined at least in part based on one or more detected features associated with the parameter values of the one or more parameters of the synthetic snapshot data, and wherein the output indicating the health condition of the asset or one or more of its components is generated at least in part based on the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data.

[0095] 6. The system according to any of the preceding clauses, wherein the one or more processors are further configured to: aggregate the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data into an aggregated alert score via a probability aggregation technique, and wherein the output indicating the health condition of the asset or one or more of its components is generated at least in part based on the aggregated alert score.

[0096] 7. The system according to any of the preceding clauses, wherein at least one of the one or more synthetic snapshots is created at a time point within a defined operating phase of the collection period corresponding to the maximum or minimum value of a given parameter among the one or more parameters.

[0097] 8. The system according to any of the preceding clauses, wherein the system or a second system associated with the system performs a control action at least in part based on the generated output indicating the health condition of the asset or one or more of its components.

[0098] 9. The system according to any of the preceding clauses, wherein the asset is an aero gas turbine engine.

[0099] 10. A method includes: receiving, by one or more processors of a system, continuous operation data associated with an asset, the continuous operation data including parameter values of one or more parameters during a collection time period; generating, by the one or more processors, synthetic snapshot data at least in part based on the continuous operation data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot including the parameter values of the one or more parameters at a given time point during the collection time period; receiving, by the one or more processors, snapshot data associated with the asset, the snapshot data including one or more snapshots, each snapshot including the parameter values of the one or more parameters at a given time point during operation of the asset; and generating, by the one or more processors, an output indicative of the health of the asset or one or more of its components at least in part based on the snapshot data and the synthetic snapshot data.

[0100] 11. The method according to any of the preceding clauses, further comprising: creating, by the one or more processors, one or more new snapshots by applying a machine learning model that utilizes one or more COD-snapshot transfer functions that associate the one or more synthetic snapshots with historical snapshot data associated with the asset.

[0101] 12. The method according to any of the preceding clauses, further comprising: adding, by the one or more processors, the one or more new snapshots to the snapshot data, wherein the one or more new snapshots are added to the snapshot data before generating the output indicative of the health of the asset or one or more of its components.

[0102] 13. The method according to any of the preceding clauses, wherein the machine learning model is trained at least in part based on the historical snapshot data associated with the asset and fleet historical snapshot data associated with other assets having the same model as the asset.

[0103] 14. The method according to any of the preceding clauses further comprises: applying, by the one or more processors, one or more time series pattern recognition techniques to the snapshot data to determine at least one alert score associated with the snapshot data, wherein the at least one alert score associated with the snapshot data is determined at least in part based on one or more detected features associated with the parameter values of the one or more parameters of the snapshot data; applying, by the one or more processors, one or more time series pattern recognition techniques to the synthetic snapshot data to determine at least one alert score associated with the synthetic snapshot data, wherein the at least one alert score associated with the synthetic snapshot data is determined at least in part based on one or more detected features associated with the parameter values of the one or more parameters of the synthetic snapshot data, and wherein the output indicating the health condition of the asset or one or more of its components is generated at least in part based on the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data.

[0104] 15. The method according to any of the preceding clauses further comprises: aggregating, by the one or more processors, the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data into an aggregated alert score via a probability aggregation technique, and wherein the output indicating the health condition of the asset or one or more of its components is generated at least in part based on the aggregated alert score.

[0105] 16. The method according to any of the preceding clauses, wherein at least one of the one or more synthetic snapshots is created at a time point within a defined operating phase of the collection period corresponding to a maximum or minimum value of a given parameter among the one or more parameters.

[0106] 17. The method according to any of the preceding clauses, wherein the system or a second system associated with the system performs a control action at least in part based on the generated output indicating the health condition of the asset or one or more of its components.

[0107] 18. The method according to any of the preceding clauses, wherein the asset is an aero gas turbine engine.

[0108] 19. The method according to any of the preceding clauses further comprises: determining, by the one or more processors, one or more values of additional parameters associated with the asset, determining the one or more values of the additional parameters using the continuous operation data, and wherein the continuous operation data used to generate the synthetic snapshot data comprises the one or more values of the additional parameters associated with the asset.

[0109] 20. A method comprising: receiving, by one or more processors of a system, continuous engine operation data associated with an aero gas turbine engine, the continuous engine operation data comprising parameter values of one or more parameters during a collection time period; generating, by the one or more processors, synthetic snapshot data at least in part based on the continuous engine operation data, the synthetic snapshot data comprising one or more synthetic snapshots, each synthetic snapshot comprising the parameter values of the one or more parameters at a given time point during the collection time period; creating, by the one or more processors, one or more new snapshots by applying a machine learning model utilizing one or more COD-snapshot transfer functions that associate the one or more synthetic snapshots with historical snapshot data associated with the aero gas turbine engine; receiving, by the one or more processors, snapshot data associated with the gas turbine engine, the snapshot data comprising one or more snapshots, each snapshot comprising the parameter values of the one or more parameters at a given time point during the collection time period; adding, by the one or more processors, the one or more new snapshots to the snapshot data; and generating, by the one or more processors, an output indicative of the health of the aero gas turbine engine or one or more of its components at least in part based on the one or more snapshots, new snapshots, and synthetic snapshots.

Claims

1. A system, characterized in that, Comprising: One or more memory devices; And One or more processors configured to: Receive continuous operation data associated with an asset, the continuous operation data including parameter values of one or more parameters within a collection time period; Generate synthetic snapshot data at least in part based on the continuous operation data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot containing the parameter values of the one or more parameters at a given time point within the collection time period; Receive snapshot data associated with the asset, the snapshot data including one or more snapshots, each snapshot containing the parameter values of the one or more parameters at a given time point during the operation of the asset; And Generate an output indicating the health status of the asset or one or more of its components at least in part based on the snapshot data and the synthetic snapshot data.

2. The system according to claim 1, characterized in that Wherein the one or more processors are further configured to: Create one or more new snapshots by applying a machine learning model utilizing one or more COD-snapshot transfer functions that associate the one or more synthetic snapshots with historical snapshot data associated with the asset.

3. The system according to claim 2, wherein Wherein the one or more processors are further configured to: Add the one or more new snapshots to the snapshot data, wherein the one or more new snapshots are added to the snapshot data before generating the output indicating the health status of the asset or one or more of its components.

4. The system according to claim 2, wherein Wherein the machine learning model is trained at least in part based on the historical snapshot data associated with the asset and fleet historical snapshot data associated with other assets having the same model as the asset.

5. The system according to claim 1, wherein Wherein the one or more processors are further configured to: Apply one or more time series pattern recognition techniques to the snapshot data to determine at least one alert score associated with the snapshot data, the at least one alert score associated with the snapshot data being determined at least in part based on one or more detected features associated with the parameter values of the one or more parameters of the snapshot data; And Apply one or more time series pattern recognition techniques to the synthetic snapshot data to determine at least one alert score associated with the synthetic snapshot data, the at least one alert score associated with the synthetic snapshot data being determined at least in part based on one or more detected features associated with the parameter values of the one or more parameters of the synthetic snapshot data, Wherein the output indicating the health status of the asset or one or more of its components is generated at least in part based on the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data.

6. The system according to claim 5, wherein, Wherein the one or more processors are further configured to: Aggregating, via a probability aggregation technique, the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data into an aggregated alert score, and wherein the output indicating the health of the asset or one or more of its components is generated at least in part based on the aggregated alert score.

7. The system according to claim 1, characterized in that, wherein at least one of the one or more synthetic snapshots is created at a time point within a defined operating phase of the collection period corresponding to a maximum or minimum value of a given one of the one or more parameters.

8. The system according to claim 1, wherein wherein the system or a second system associated with the system performs a control action at least in part based on the generated output indicating the health of the asset or one or more of its components.

9. The system according to claim 1, characterized in that, wherein the asset is an aero gas turbine engine.

10. A method, characterized in that, Comprising: Receiving, by one or more processors of a system, continuous operating data associated with an asset, the continuous operating data including parameter values of one or more parameters within a collection period; Generating, by the one or more processors, synthetic snapshot data at least in part based on the continuous operating data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot including the parameter values of the one or more parameters at a given time point within the collection period; Receiving, by the one or more processors, snapshot data associated with the asset, the snapshot data including one or more snapshots, each snapshot including the parameter values of the one or more parameters at a given time point during operation of the asset; And Generating, by the one or more processors, an output indicating the health of the asset or one or more of its components at least in part based on the snapshot data and the synthetic snapshot data.

11. The method according to claim 10, wherein Further comprising: Creating, by the one or more processors, one or more new snapshots by applying a machine learning model utilizing one or more COD-snapshot transfer functions that associate the one or more synthetic snapshots with historical snapshot data associated with the asset.

12. The method according to claim 11, wherein Further comprising: Adding, by the one or more processors, the one or more new snapshots to the snapshot data, wherein the one or more new snapshots are added to the snapshot data before generating the output indicating the health of the asset or one or more of its components.

13. The method according to claim 11, wherein wherein the machine learning model is trained at least in part based on the historical snapshot data associated with the asset and fleet historical snapshot data associated with other assets having the same model as the asset.

14. The method according to claim 10, wherein Further comprising: Applying, by the one or more processors, one or more time series pattern recognition techniques to the snapshot data to determine at least one alert score associated with the snapshot data, the at least one alert score associated with the snapshot data being determined at least in part based on one or more detected features associated with the parameter values of the one or more parameters of the snapshot data; The one or more processors apply one or more time series pattern recognition techniques to the synthetic snapshot data to determine at least one alert score associated with the synthetic snapshot data, the at least one alert score associated with the synthetic snapshot data being determined at least in part based on one or more detected features associated with the parameter values of the one or more parameters of the synthetic snapshot data, and wherein the output indicating the health of the asset or one or more of its components is generated at least in part based on the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data.

15. The method according to claim 14, characterized in that, Further comprising: the one or more processors aggregate the at least one alert score associated with the snapshot data and the at least one alert score associated with the synthetic snapshot data into an aggregated alert score via a probability aggregation technique, and wherein the output indicating the health of the asset or one or more of its components is generated at least in part based on the aggregated alert score.

16. The method according to claim 10, wherein wherein at least one of the one or more synthetic snapshots is created at a time point within a defined operating phase of the collection period corresponding to a maximum or minimum value of a given parameter of the one or more parameters.

17. The method according to claim 10, wherein wherein the system or a second system associated with the system performs a control action at least in part based on the generated output indicating the health of the asset or one or more of its components.

18. The method according to claim 10, wherein wherein the asset is an aero gas turbine engine.

19. The method according to claim 10, wherein Further comprising: the one or more processors determine one or more values of additional parameters associated with the asset, the one or more values of the additional parameters being determined using the continuous operation data, and wherein the continuous operation data used to generate the synthetic snapshot data includes the one or more values of the additional parameters associated with the asset.

20. A method, characterized in that, Comprising: one or more processors of the system receive continuous engine operation data associated with an aero gas turbine engine, the continuous engine operation data including parameter values of one or more parameters within a collection period; the one or more processors generate synthetic snapshot data at least in part based on the continuous engine operation data, the synthetic snapshot data including one or more synthetic snapshots, each synthetic snapshot including the parameter values of the one or more parameters at a given time point within the collection period; the one or more processors create one or more new snapshots by applying a machine learning model utilizing one or more COD-snapshot transfer functions that associate the one or more synthetic snapshots with historical snapshot data associated with the aero gas turbine engine; the one or more processors receive snapshot data associated with the gas turbine engine, the snapshot data including one or more snapshots, each snapshot including the parameter values of the one or more parameters at a given time point within the collection period; Adding, by the one or more processors, the one or more new snapshots to the snapshot data; and Generating, by the one or more processors, an output indicative of the health of the aero gas turbine engine or one or more components thereof, at least in part based on the one or more snapshots, new snapshots, and synthetic snapshots.

Images