SYSTEM AND METHOD FOR VALVE EVENT DETECTION AND CONTROL

Abstract

System, comprehensive: an engine control system configured to control an internal combustion engine (10), the engine control system comprising a processor (72) configured to: to receive a vibration signal that is detected by a knock sensor (23) arranged in the internal combustion engine (10); to correlate the vibration signal with a fingerprint (200) that has at least one ADSR envelope (182) indicating the operating event of a valve train (70) of the internal combustion engine (10), to analyze the vibration signal with a statistical valve train model (460) or a combination thereof; based on the correlation of the vibration signal with the fingerprint (200), based on the analysis of the vibration signal with a statistical valve train model (460) or a combination thereof, to detect whether the operational event has occurred; and to control the valve train (70) based on the operational event; wherein the processor (72) is configured to derive the ADSR envelope (182) from a basic data acquisition noise signal indicating the operational event and to plot the ADSR envelope and operational event indicator data over time to derive a location of the ADSR envelope (182) where the operational event occurs; wherein the location of the ADSR envelope (182) where the operational event occurs lies within a mid-range of a decay vector (184) of the ADSR envelope (182).

Description

BACKGROUND

[0001] The subject of the disclosure concerns valve event detection and control.

[0002] Internal combustion engines typically burn a carbonaceous fuel, such as natural gas, gasoline, diesel, and the like, and use the associated expansion of gases at high temperature and pressure to exert force on specific engine components, such as the piston located in a cylinder, to move the components over a distance. Each cylinder may contain one or more valves that open and close in correlation with the combustion of the carbonaceous fuel. For example, an intake valve may direct an oxidizer, such as air, into the cylinder, which is then mixed with fuel and burned. Combustion fluids, such as hot gases, may then be directed out of the cylinder by means of an exhaust valve. Accordingly, the carbonaceous fuel is converted into mechanical motion that can be used to drive a load.For example, the load could be a generator producing electrical power. Improving valve event detection and control would be advantageous.

[0003] US 2005 / 0027433A1 discloses an engine control system and method for controlling an internal combustion engine with a control device configured to receive a vibration signal detected by a knock sensor located in the internal combustion engine, to correlate the detected vibration signal with a fingerprint, to calculate the impact velocity when an engine valve strikes a valve seat or the energy consumed in the process, to detect operational anomalies such as excessive friction or decreasing spring force, and to modify the control of the actuation of the engine valve based on this.

[0004] US Patent 2016 / 0298553A1 discloses a computer-implemented system and method for controlling an engine using knock intensity data. Knock intensities from multiple combustion cycles are used to estimate a statistical distribution of knock intensities. This distribution is used to determine a descriptive statistic that can represent the engine's tuning state. A calculated descriptive statistic is compared to a desired descriptive statistic, and any deviation between the calculated and desired knock intensity distribution is used to adjust an engine control parameter.

[0005] US 2009 / 0048729 A1 discloses a method for determining whether a variable valve actuation (VFA) device or subsystem is operating in a non-compliant mode, in which an engine or powertrain controller monitors and evaluates an output signal from a knock sensor, acquired during a predefined sampling window defined to include a valve closing event when the VFA device is operating correctly. The acquired knock sensor output signal is processed to detect the presence (or absence) of a valve closing event. The absence of a valve closing event, when one is expected, indicates a malfunction of the VFA device. SHORT DESCRIPTION

[0006] Based on this, it is an object of the present invention to create a control system and method for an internal combustion engine and a tangible, non-transient, computer-readable medium that makes it possible to improve the valve event detection and control for valves of the internal combustion engine.

[0007] This problem is solved by the features of the independent patent claims. Advantageous embodiments of the invention are described in the dependent claims.

[0008] Certain aspects and embodiments that correspond to the scope of protection of the originally claimed invention are summarized below. These embodiments are not intended to limit the scope of protection of the claimed invention, but rather to provide only a brief summary of possible forms of the invention. In fact, the invention may comprise a variety of forms that are similar to or different from the embodiments described below.

[0009] In a first aspect of the invention, a system comprises an engine control system configured to control an internal combustion engine. The engine control system comprises a processor configured to receive a vibration signal detected by a knock sensor located in the internal combustion engine. The processor is further configured to correlate the vibration signal with a fingerprint that includes at least one ADSR envelope indicating the operating event of a valve train of the internal combustion engine, to analyze the vibration signal with a statistical valve train model, or a combination thereof.The processor is also configured to detect whether the operational event has occurred based on the correlation of the vibration signal with the fingerprint (200), based on the analysis of the vibration signal with a statistical valve train model, or a combination thereof, and to control the valve train based on the operational event. The processor is configured to derive the ADSR envelope from a baseline noise signal indicating the operational event and to plot the ADSR envelope and operational event indicator data over time to derive a location on the ADSR envelope where the operational event occurs, where the location of the ADSR envelope where the operational event occurs lies within a mid-range of a vector of the ADSR envelope's decay.

[0010] In each embodiment of the system, it may be advantageous for the processor to be configured to control the valve train by controlling a valve adjusting device, a variable valve timing, or a combination thereof.

[0011] In each embodiment of the system, it may be advantageous for the valve adjusting device to have a variable valve tappet and to include controlling a variable valve timing, controlling a valve actuator, the variable valve tappet, or a combination thereof.

[0012] In each embodiment of the system, it may be advantageous for the processor to be configured to detect operating event characteristics for the operating event, with the engine control system being configured to control the valve train based on the operating event and the operating event characteristics.

[0013] In each embodiment of the system, it may be advantageous for the operating event characteristics to include a valve clearance distance, a valve clearance consumption, a valve timing drift measurement, a valve seat velocity, or a combination thereof.

[0014] In each embodiment of the system, it may be advantageous for the processor to be configured to: receive a crankshaft signal detected by a crankshaft angle sensor located in the engine, wherein the crankshaft angle signal represents an engine crankshaft angle; and monitor valve timing by deriving a cylinder head acceleration measurement using the vibration signal received from the knock sensor, wherein the control system monitors the valve timing by deriving a valve clearance based on the vibration signal, the engine crankshaft angle, and a threshold valve clearance model included in the statistical valve train model.

[0015] In a second aspect of the invention, a method comprises receiving a vibration signal detected by a knock sensor arranged in an internal combustion engine and correlating the vibration signal with a fingerprint that includes at least one ADSR envelope indicating the operating event of a valve train of the internal combustion engine, and analyzing the vibration signal with a statistical valve train model or a combination thereof. The method further comprises detecting whether the operating event has occurred, based on correlating the vibration signal with the fingerprint, based on analyzing the vibration signal with a statistical valve train model or a combination thereof, and controlling the valve train based on the operating event.According to the inventive method, the ADSR envelope is derived from a basic data acquisition noise signal indicating the operational event, and the ADSR envelope and operational event indicator data are plotted over time to derive a location of the ADSR envelope where the operational event occurs, wherein the location of the ADSR envelope where the operational event occurs lies within a central region of a vector of the fall of the ADSR envelope.

[0016] In each embodiment of the method, it may be advantageous for the control of the valve train based on the operating event to include the control of a valve adjusting device, a variable valve timing or a combination thereof.

[0017] In each embodiment of the method, it may be advantageous for the method to include the acquisition of operating event characteristics for the operating event, wherein the control of the valve train includes the control of the valve train based on the operating event and the operating event characteristics, wherein the operating event characteristics include a valve clearance distance, a valve clearance consumption, a valve timing drift measurement, a valve seat velocity, or a combination thereof.

[0018] In each embodiment of the method, it may be advantageous for the method to comprise: receiving a crankshaft signal detected by a crankshaft angle sensor arranged in the engine, wherein the crankshaft signal represents an engine crankshaft angle; and monitoring a valve timing by deriving a cylinder head acceleration measurement from the vibration signal received by the knock sensor, wherein the control system is configured to monitor the valve timing by deriving a valve clearance based on the vibration signal, the engine crankshaft angle, and a threshold valve clearance model included in the statistical valve train model.

[0019] In a third aspect of the invention, a tangible, non-transitory, computer-readable medium storing code is provided to cause a processor to receive a vibration signal detected by a knock sensor arranged in an engine, and to correlate the vibration signal with a fingerprint having at least one ADSR envelope indicating the operating event of a valve train of the internal combustion engine, to analyze the vibration signal with a statistical valve train model, or a combination thereof.The tangible, non-volatile, computer-readable medium that stores code is configured to additionally instruct the processor to detect whether the operational event has occurred, based on the correlation of the noise signal with the fingerprint, based on the analysis of the vibration signal with a statistical valve train model, or a combination thereof, and to control the valve train based on the operational event. The code is further configured to instruct the processor to derive the ADSR envelope from a baseline noise signal indicating the operational event and to plot the ADSR envelope and operational event indicator data over time to derive a location on the ADSR envelope where the operational event occurs, the location of the ADSR envelope where the operational event occurs being within a mid-range of a vector of the ADSR envelope's decay.

[0020] In any embodiment of the computer-readable medium, it may be advantageous that, when executed, it is further configured to cause the processor to control the valve train by controlling a valve adjusting device, a variable valve timing device, or a combination thereof.

[0021] In each embodiment of the computer-readable medium, it may be advantageous that, when executed, it is further configured to cause the processor to acquire operational event characteristics for the operational event, wherein the code causes the processor to control the valve train based on the operational event and the operational event characteristics, and wherein the operational event characteristics include a valve clearance distance, a valve clearance consumption, a valve timing drift measurement, a valve seat velocity, or a combination thereof.

[0022] In each embodiment of the computer-readable medium, it may be advantageous that, when executed, it is further configured to cause the processor to: receive a crankshaft signal detected by a crankshaft angle sensor arranged in the engine, wherein the crankshaft signal is representative of an engine crankshaft angle; and monitor valve timing by deriving a cylinder head acceleration measurement using the vibration signal received from the knock sensor, wherein the control system is configured to monitor the valve timing by deriving a valve clearance based on the vibration signal, the engine crankshaft angle, and a threshold valve clearance model included in the statistical valve train model. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] These and other features, aspects and advantages of the present invention will be better understood when the following detailed description is read with reference to the accompanying drawings, in which the same reference numerals represent the same sections throughout the drawings, wherein: Fig. Figure 1 is a block diagram of an embodiment of a section of a motor-driven power generation system according to aspects of the present disclosure; Fig. 2 is a side cross-sectional view of an embodiment of a piston arrangement in a cylinder of the in Fig. 1 reciprocating engine shown according to aspects of the present disclosure; Fig. Figure 3 is a block diagram of an embodiment of a valve train with a valve clearance and a specific valve timing control and valve adjusting devices; Fig. 4 is an embodiment of an engine noise diagram of data generated by the Fig. 2 knock sensor shown, measured according to aspects of the present disclosure; Fig. Figure 5 is an embodiment of a scaled version of the in Fig. 4 exemplary engine noise diagram shown according to aspects of the present disclosure; Fig. 6 is an embodiment of a in Fig. 5 shown exemplary scaled engine noise representation with four main parameters of an attack-decay-hold-release (ADSR) envelope superimposed according to aspects of the present disclosure; Fig. Figure 7 is an embodiment of a scaled engine noise diagram and an ADSR envelope, which is shown in Fig. Figure 6 shows the extracted tones superimposed according to aspects of the present revelation; Fig. Figure 8 is a flowchart showing an embodiment of a process for characterizing a noise according to aspects of the present disclosure; Fig. Figure 9 is an embodiment of a scaled engine noise diagram corresponding to an engine operating event, an operating event indicator associated with the engine operating event, and an ADSR envelope associated with the engine operating event, according to the process of Fig. 8 and other aspects of the present revelation; Fig. Figure 10 is a flowchart showing one embodiment of a process for identifying a fingerprint, which is in Fig. 8 is shown, according to aspects of the present revelation; Fig. Figure 11 is a flowchart of an embodiment of a process suitable for processing engine noise in order to derive certain engine operating events; Fig. Figure 12 illustrates an embodiment of diagrams showing valve lift and speed (top diagram) and cylinder head vibration signal (valve noise) (bottom diagram); Fig. Figure 13 illustrates an embodiment of diagrams that represent derived valve clearance values ​​versus valve seat velocity and valve noise; Fig. Figure 14 illustrates an embodiment of diagrams showing the correlation between the valve clearance in operation and the closing crank angle recorded by the valve lift measurement (left diagram) and the crank angle triggered at a knock sensor on a cylinder head (right diagram); Fig. Figure 15 represents an embodiment of a diagram used for condition-based maintenance of the engine of Fig. 1 is suitable; Fig. Figure 16 illustrates an embodiment of a second diagram used for condition-based maintenance of the engine of Fig. 1 is suitable; and Fig. Figure 17 illustrates an embodiment of a process suitable for detecting valve events and applying valve control based on the detected valve events. DETAILED DESCRIPTION

[0024] One or more specific embodiments of the present invention are described below. In an attempt to provide a concise description of these embodiments, not all features of an actual implementation can be explained in the description. It should be understood that in developing any such actual implementation, as in any engineering or design project, many implementation-specific decisions must be made to achieve the developer's specific objectives, such as conformity with system-related or business-related conditions, which may vary from one implementation to another.Furthermore, it should be understood that such a development effort may be complex and time-consuming, but would nevertheless be a routine undertaking of design, fabrication and manufacturing for experts in the field who would benefit from this revelation.

[0025] When elements from different embodiments of the present invention are introduced, the articles "a," "the," and "this" are intended to indicate the presence of one or more of these elements. The expressions "having," "containing," and "with" are intended to be inclusive and mean that additional elements or elements other than those listed may be present.

[0026] The techniques described here involve the use of one or more knock sensor systems and methods that can detect specific valve events and conditions and then respond by actively controlling valve timing and / or valve lift. For example, knowledge or information derived from knock sensor signal processing, as described herein, to detect an incorrectly set valve clearance can be used to correct the problem via variable valve timing and / or valve lift. For example, a corrective action might involve adjusting the lift height of a hydraulic tappet by modifying the supply pressure to the tappet. Similarly, electronic valve actuation can be implemented to dynamically adjust the valve timing. Corrective actions would enable the engine to operate as intended.Accordingly, engine knocking, excessive imbalance in peak cylinder pressure, and other problems resulting from incorrectly adjusted valves can be corrected. By adjusting to specific valve states, engine operation can be made more efficient and maintenance schedules improved.

[0027] It may first be useful to describe the use of knock sensors to detect certain non-knocking engine events. When a knock sensor is used to monitor an internal combustion engine, the knock sensor system occasionally records a noise, such as an abnormal or undesirable sound that may not be identified at that time. Alternatively, the knock sensor may record a noise that is normal or desired, where the noise has been previously identified and characterized. For example, noises emitted by the internal combustion engine during various engine operations may initially be characterized during an in-house baseline data acquisition process. Noise signals for specific operating events and conditions (e.g.,Valve closures, valve openings and peak firing pressure, valve blockage, valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation, and valve leakage can be processed and stored in a database as relating to one or more operating event characteristics during the basic data acquisition process. During normal operation of the internal combustion engine, data stored in the database relating to the operating event characteristics characterized during the basic data acquisition process can be accessed to determine whether the operating noise corresponds to the operating events characterized during the basic data acquisition process.

[0028] In one embodiment, the techniques described herein can generate a noise “fingerprint” of specific engine tones or noises. The fingerprint (e.g., the profile, the comparator, and / or the reference signal) can be developed during the basic data acquisition process, as described above, and the fingerprint can correspond to a specific operational event (e.g., valve closing) tested during the basic data acquisition process. Other valve events for the fingerprint include valve blockage, valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation, valve leakage, and so on. It should be noted that the basic data acquisition process can be performed during full operation of the internal combustion engine or while only specific components (e.g.,The components relating to the operating event(s) whose baseline values ​​are determined are implemented. For example, in some embodiments, different baseline values ​​for operating events of the internal combustion engine can be determined in the factory during partial or full operation.

[0029] During full operation of the internal combustion engine (e.g., after basic data acquisition), the knock sensor can detect the noise, and the noise signal can be processed and compared with various fingerprints (e.g., profiles, signatures, comparators, reference signals, unique identifiers, unique representations, etc.) related to the internal combustion engine. If the fingerprint and the processed noise signal match or correlate (e.g., "match"), the signal can be confirmed as belonging to the operating event associated with the fingerprint. The noise signal can also be processed to determine time-sensitive information regarding the operating event that corresponds to the matching fingerprint and noise signal.For example, if the noise signal matches a fingerprint corresponding to the closing of an exhaust valve, the noise signal can be plotted against time (or crank angle) to determine when the exhaust valve closed.

[0030] As described in more detail below, systems and methods for identifying and classifying noise are provided via an attack-decay-hold-release (ADSR) envelope and / or joint time-frequency techniques, where the ADSR envelope can correspond to at least one segment of the aforementioned fingerprint. The joint time-frequency techniques can include cepstrum techniques, quefrence techniques, chirplet techniques, and / or wavelet techniques to develop an acoustic model or fingerprint of the noise, as described in more detail below.

[0031] The techniques described herein further include the use of one or more knock sensor systems and methods capable of detecting a dynamic cylinder head response caused by intake and exhaust valve seat excitation. Valve excitation can occur when a valve, such as a conical or circular valve, "seats" or otherwise closes a cylinder chamber with certain valve sections seated in front of others. Advantageously, the techniques described herein include retrofitting existing systems, such as upgrading an existing engine control unit (ECU) or engine control module (ECM), to utilize existing knock valve systems to obtain intake and / or exhaust valve states, including deriving variations in valve timing for each cylinder of an internal combustion or reciprocating engine.Such data can be used to identify specific valve conditions, such as stuck valves, separated valve stems, structural defects in the valve train, etc., and can thus be used to improve engine maintenance and overall operation. Accordingly, valve timing can be monitored by measuring cylinder head acceleration, for example, via a knock sensor for remote diagnostics of the valve train.

[0032] In one embodiment, the techniques described herein can detect a time drift of the cylinder head response due to valve seat excitation in order to infer a drift in the valve closing phase due to, for example, valve clearance (e.g., play or gap in a valve train between the camshaft and the valve) or valve stem separation. Time drift detection can involve the use of statistical techniques, as described in more detail below, which are useful in analyzing knock sensor data using a variety of sensors, including standard knock sensors positioned to detect engine knock. Accordingly, the retrofit can apply a software update (e.g., a flash update) and may not involve any hardware modifications. Accelerometers can measure a dynamic cylinder head response due to valve seat excitation.A trigger crank angle (CA) of the signal determines the actual valve closing event and can be correlated with the valve clearance during operation. The variation of the trigger CA allows for the determination of the variation in the actual valve closing time. This determination can be used to detect at least two types of failure modes: 1) clearance variations during operation, resulting in a slow temporal drift of the trigger CA, e.g., valve wear progression (wear = clearance setting - clearance during operation), loosening of an adjusting screw, variation in the thermal expansion of valve train components; and 2) valve train failure, resulting, for example, in a sudden temporal variation of the valve timing, independent of the valve clearance during operation, including valve slippage, connecting rod breakage, and so on. The techniques described herein provide results independent of the type of accelerometer used (e.g.,piezoelectric, charge accelerometer) and a position of the accelerometer on a cylinder head (e.g. a sensor can be used in one or more cylinder head bolts).

[0033] Accordingly, the techniques described herein can provide for the remote and local detection of certain undesired valve events, such as valve train events. The techniques described herein can additionally include systems and procedures for controlling engine operating modes once the undesired valve events (e.g., valve blockage, valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation) have been detected.

[0034] With reference to Fig. Figure 1 shows a block diagram of an embodiment of a section of an engine-driven power generation system 8. As described in detail below, the system 8 comprises an engine 10 (e.g., a reciprocating internal combustion engine) with one or more combustion chambers 12 (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 18, 20 or more). An upper section of the combustion chamber 12 may be formed by a cylinder head 14. Although Fig. Figure 1 shows an internal combustion engine 10; it should be understood that any reciprocating device can be used. An air supply is provided to deliver a pressurized oxidizer 16, such as air, oxygen, oxygen-enriched air, oxygen-reduced air, or any combination thereof, to each combustion chamber 12. The combustion chamber 12 is also configured to receive a fuel 18 (e.g., a liquid and / or gaseous fuel) from a fuel supply 19 and ignites and burns a fuel-air mixture within each combustion chamber 12. The hot, pressurized combustion gases cause a piston 20 adjacent to each combustion chamber 12 to move linearly within a cylinder 26, converting the pressure exerted by the gases into a rotary motion that causes a shaft 22 to rotate.Furthermore, the shaft 22 can be connected to a load 24, which is driven by the rotation of the shaft 22. For example, the load 24 can be any suitable device capable of generating power via the rotational power of the system 10, such as an electric generator. Although the following description refers to air as the oxidizer 16, any suitable oxidizer can additionally be used with the disclosed embodiments. Similarly, the fuel 18 can be any suitable gaseous fuel, such as natural gas, associated petroleum gas, propane, biogas, sewage gas, landfill gas, or coal mine gas.

[0035] The system 8 disclosed herein can be adapted for use in stationary applications (e.g., in industrial power-generating machines) or in mobile applications (e.g., in cars or aircraft). The engine 10 can be a two-stroke, three-stroke, four-stroke, five-stroke, or six-stroke engine. The engine 10 can also comprise any number of combustion chambers 12, pistons 20, and associated cylinders (e.g., 1-24). For example, in certain embodiments, the system 8 can comprise a large-scale industrial reciprocating engine with 4, 6, 8, 10, 16, 24, or more pistons 20 moving back and forth in cylinders. In some such cases, the cylinders and / or the pistons 20 can have a diameter between approximately 13.5 and 34 centimeters (cm). In some embodiments, the cylinders and / or the pistons 20 may have a diameter between approximately 10 to 40 cm, 15 to 25 cm or approximately 15 cm.System 10 can generate a power output in the range of 10 kW to 10 MW. In some embodiments, the motor 10 can operate at less than approximately 1800 revolutions per minute (rpm). In some embodiments, the motor 10 can operate at less than approximately 2000 rpm, 1900 rpm, 1700 rpm, 1600 rpm, 1500 rpm, 1400 rpm, 1300 rpm, 1200 rpm, 1000 rpm, 900 rpm, or 750 rpm. In some embodiments, the motor 10 can operate between approximately 750 and 2000 rpm, 900 and 1800 rpm, or 1000 and 1600 rpm. In some embodiments, the motor 10 can operate at approximately 1800 rpm, 1500 rpm, 1200 rpm, 1000 rpm, or 900 rpm. Exemplary motors 10 may include, for example, Jenbacher motors from General Electric Company (e.g., Jenbacher Type 2, Type 3, Type 4, Type 6, or J920 FleXtra) or Waukesha motors (e.g., Waukesha VGF, VHP, APG, 275GL).

[0036] The driven power generation system 8 can include one or more knock sensors 23 capable of detecting engine knock. The knock sensor 23 can be any sensor configured to detect vibrations caused by the engine 10, such as pre-ignition and / or pinging. Multiple knock sensors 23 are used, just as there are multiple pistons 20 and cylinders 26. The knock sensor(s) 23 is / are communicatively connected to a control unit, the engine control unit (ECU) 25. During operation, signals from the knock sensor 23 are displayed and transmitted to the ECU 25 to determine whether knocking conditions (e.g., pinging) exist. The ECU 25 can then adjust certain parameters of the engine 10 to mitigate or eliminate the knocking conditions. For example, the ECU 25 can adjust the ignition timing and / or the boost pressure to eliminate knocking.As further described herein, the knock sensor 23 can additionally deduce that certain vibrations should be further analyzed and categorized in order to detect, for example, undesirable engine conditions. The ECU 25 can then actuate certain devices, such as a valve adjusting device 27 and / or a camless valve actuating device 31, which may be included in a valve train 70 described below, to adjust valve actuations.

[0037] Fig. Figure 2 is a side cross-sectional view of an embodiment of a piston arrangement 29 with a piston 20 arranged in a cylinder 26 (e.g., an engine cylinder) of the reciprocating machine 10. The cylinder 26 has an inner annular wall 28 that defines a cylindrical cavity 30 (e.g., bore). The piston 20 can be defined by an axial axis or direction 34, a radial axis or direction 36, and a circumferential axis or direction 38. The piston 20 includes an upper section 40 (e.g., an upper web). The upper section 40 generally blocks the fuel 18 and air 16, or a fuel-air mixture 32, from exiting the combustion chamber 12 during the reciprocating motion of the piston 20.

[0038] As shown, the piston 20 is attached to a crankshaft 54 ​​via a connecting rod 56 and a pin 58. The crankshaft 54 ​​converts the linear reciprocating motion of the piston 24 into a rotary motion. As the piston 20 moves, the crankshaft 54 ​​rotates to actuate the load 24 (in Fig. (1 shown) with energy, as discussed above. As shown, the combustion chamber 12 is positioned adjacent to the upper web 40 of the piston 24. A fuel injection device 60 supplies the combustion chamber 12 with fuel 18, and an inlet valve 62 controls the supply of air 16 to the combustion chamber 12. An exhaust valve 64 controls the discharge of exhaust from the engine 10. However, it should be understood that any suitable elements and / or techniques may be used to supply fuel 18 and air 16 to the combustion chamber 12 and / or to discharge exhaust, and that in some embodiments no fuel injection is used. In operation, the combustion of the fuel 18 with the air 16 in the combustion chamber 12 causes the piston 20 to move back and forth (e.g., forward and backward) in the axial direction 34 within the cavity 30 of the cylinder 26.

[0039] When the piston 20 is at its highest point in the cylinder 26 during operation, it is in a position known as top dead center (TDC). When the piston 20 is at its lowest point in the cylinder 26, it is in a position known as bottom dead center (BDC). As the piston 20 moves from top to bottom or bottom to top, the crankshaft 54 ​​rotates half a revolution. Each movement of the piston 20 from top to bottom or bottom to top is called a stroke, and the embodiments of the engine 10 can include two-stroke, three-stroke, four-stroke, five-stroke, and six-stroke engines, or engines with more strokes.

[0040] During the operation of engine 10, a typical sequence occurs that includes an intake process, a compression process, a power process, and an exhaust process. The intake process allows a combustible mixture, such as fuel and air, to be drawn into cylinder 26, thus opening the intake valve 62 and closing the exhaust valve 64. The compression process compresses the combustible mixture into a smaller space, so that both the intake valve 62 and the exhaust valve 64 are closed. The power process ignites the compressed fuel-air mixture, which may involve spark ignition by a spark plug system and / or compression ignition by the heat of compression. The pressure resulting from combustion then pushes the piston 20 to BDC. The exhaust process typically returns the piston 20 to TDC while the exhaust valve 64 remains open.The exhaust process thus expels the used fuel-air mixture through the exhaust valve 64. It should be noted that more than one intake valve 62 and one exhaust valve 64 can be used per cylinder 26.

[0041] The depicted engine 10 also includes a crankshaft sensor 66, a knock sensor 23, and the engine control unit (ECU) 25, which comprises a processor 72 and memory 74. The crankshaft or crank angle sensor 66 detects the position and / or rotational speed of the crankshaft 54. Accordingly, crank angle or crank timing information can be derived from the crankshaft sensor 66. That is, when monitoring internal combustion engines, timing is often expressed as the angle of the crankshaft 54. For example, a complete cycle of a four-stroke engine 10 can be measured as a 720° cycle. The knock sensor 23 can be a piezoelectric accelerometer, a microelectromechanical system (MEMS) sensor, a Hall effect sensor, a magnetostrictive sensor, and / or any other sensor configured to detect vibration, acceleration, sound, and / or motion.In other embodiments, the sensor 23 may not be a knock sensor, but any sensor that can detect noise, vibration, pressure, acceleration, deflection and / or movement.

[0042] Due to the percussive nature of the engine 10, the knock sensor 23 can detect signatures even when mounted on the outside of the cylinder 26. However, the knock sensor 23 can be located at various points in or around the cylinder 26. Furthermore, in some embodiments, a single knock sensor 23 can be shared, for example, with one or more adjacent cylinders 26. In other embodiments, each cylinder 26 can include one or more knock sensors 23. The crankshaft sensor 66 and the knock sensor 23 are shown in electronic communication with the engine control unit (ECU) 25. The ECU 25 includes a processor 72 and a memory 74. The memory 74 can store computer instructions that can be executed by the processor 72.The ECU 25 monitors and controls the operation of the engine 10, for example by adjusting the combustion timing, the timing of the valves 62, 64, the valve clearance, the adjustment of the supply of fuel and oxidizer (e.g. air) and so on.

[0043] With reference to Fig. Figure 3 shows an embodiment of a valve train 70. The valve train 70 comprises one of the valves 62, 64 with a valve stem 72. The valve stem 72 (e.g., tappet or "bolt") can be mechanically connected to a rocker arm 74, which can move (e.g., open or close) the valve 62, 64 during operation of the engine 10. A rod 76 is also shown mechanically connected to the rocker arm 74 and is suitable for connecting the rocker arm to the engine 10, thereby transmitting a driving force to the valve train 70. A spring 78 is also shown, which provides a preload force for the valve 62, 64 to assist in opening and / or closing the valve 62, 64. A valve clearance (e.g., gap or distance) 80 between the valve stem 72 and the rocker arm 74 is shown. During operation, the valve clearance can “drift” or otherwise become larger or smaller.The techniques described herein can use the knock sensor(s) 23 to derive and analyze the drift of the valve clearance 80 over time and to derive and analyze other features of the valve train 70, and can provide a dynamic control approach for the valve train 70 where the valve train 70 can be adjusted, for example, based on a derived clearance.

[0044] In one embodiment, the valve clearance 80 can be adjusted via the valve adjusting device 27, such as a variable valve tappet. The variable valve tappet can be a discrete valve tappet. The valve adjusting device 27 and the camless valve actuator 31 can be operatively connected to the ECU 25 via lines 81 and 87 (e.g., wired line, wireless line). The valve adjusting device 27 can dynamically change the length of the rod 76 by "raising" and / or moving it downwards in directions 83, 85. Lengthening the rod 76 via the valve adjusting device 27 shortens the valve clearance 80, while shortening the rod 76 increases the valve clearance 80. Actuating the camless valve actuator 31 can assist in delaying the valve 62, 64 and / or increasing the valve timing.

[0045] Advantageously, it was found that the states of the valve train 70 (e.g., incorrect valve clearance, excessive wear of the valve train 70, valve leakage, etc.) can be derived based on signals from the knock sensor(s) 23. Two sets of techniques suitable for deriving the states of the valve train 70 are described here. A first set of techniques, which are described below with reference to the Fig. The techniques described in more detail in sections 4-11 are aimed at identifying and classifying noise via a rise-fall-hold-release (ADSR) envelope and / or joint time-frequency techniques, where the ADSR envelope can correspond to at least one segment of the aforementioned fingerprint. The joint time-frequency techniques can include cepstrum techniques, quefrency techniques, chirplet techniques, and / or wavelet techniques to develop an acoustic model or fingerprint of the noise, as described in more detail below.

[0046] A second set of techniques, described in more detail with reference to the Fig. The method described in Sections 12-16 is aimed at a specific statistical analysis of the data from the knock sensor 23. For example, a dynamic response of the cylinder head 14, caused by the seat excitation of the intake valve 62 and / or the exhaust valve 64, can be detected by the knock sensor(s) 23 and used to derive a variety of states of the valve train 70, including a drift of the valve gap 80. In one example, a quantity (e.g., |F|, where F is a frequency range) and a phase (e.g., Φ(F)) of a dynamic response of the cylinder head 14, measured by one or more accelerometers in a crank angle range (e.g., between -720° and 720°), include information regarding the seat excitation force and the timing of the valves 62 and 64, respectively.The angle of the triggered crank 54 of the absolute acceleration signal can be robust information correlated with the timing of the valves 62, 64. This timing can be monitored during operation to detect long-term drift of the valve clearance 80 (e.g., drift after approximately 100, 500, 1000, 2000, 2500, 3000 or more operating hours) and sudden valve train events, such as a stem break 72 (e.g., between two successive cycles). Several model types can then be derived to detect states of the valve train 70 (e.g., valve drift 80, valve stem break 72).

[0047] The models can include one or more valve noise models that analyze valve knock sensor signals without considering triggered (e.g., opening) crank angle information. The models can additionally include triggered crank angle models that, in addition to knock sensor signals, contain triggered crank angle information models. The models (e.g., noise models and triggered crank angle modes) can be used individually or in combination to derive valve train states. Both types of techniques, e.g., ADSR techniques and statistical techniques, can be used individually or in combination to detect valve train states (e.g., valve blockage, valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation, valve leakage) and / or to derive control measures.The control measures include modulating or otherwise controlling the valve adjusting device 27 and / or the camless valve actuator 31, applying variable valve timing techniques, applying ignition timing control techniques, and so on.

[0048] Fig. Figures 4-7 and 9 illustrate data that can be subjected to data processing, for example, through a process or processes related to the Fig. 8 and Fig. 10 will be described in more detail. The data for Fig. 4-7 and 9 can include data transmitted via the knock sensor 23 and the crankshaft sensor 66. Fig. Figure 4 is, for example, an embodiment of a raw machine noise diagram 175, which is derived (e.g., by the ECU 25) from noise data measured by the knock sensor 23, in which the x-axis 176 is the position of the crankshaft 54 ​​(e.g., a crank angle) correlated with time. According to the present embodiments, the noise data can correspond to a specific operating event or action of the engine 10. For example, the noise data can correspond to the opening or closing of a valve of the engine 10, such as the exhaust valve 64. Alternatively, the noise data can correspond to a peak firing pressure, which describes the highest pressure in the combustion chamber 12 during combustion.

[0049] Diagram 175 is generated when the ECU 25 combines the data received from the knock sensor 23 and the crankshaft sensor 66 during engine 10 operation. In the illustrated embodiment, an amplitude curve 177 of the knock sensor 23 signal with an amplitude axis 78 is shown. That is, the amplitude curve 177 contains amplitude measurements of vibration data (e.g., noise, sound data) acquired by the knock sensor 23, plotted against the crankshaft angle. It should be understood that this is merely a graphical representation of a sample data set (e.g., corresponding to the closing of the exhaust valve 64) and is not intended to limit curves generated by the ECU 25. The curve 177 can then be scaled for further processing, as shown in Fig. 5 shown.

[0050] Fig. Figure 5 is an embodiment of a scaled engine noise diagram 179, which can be derived from the ECU 25. In the scaled diagram 179, the raw engine noise was derived from the Fig. The amplitude recording 175 shown in Figure 4 is scaled to derive a scaled amplitude curve 180. In this case, a single multiplier was applied to each data point such that the maximum positive value of the scaled amplitude curve 180 is 1. Note that the multiplier applied to each point of the curve 180 to produce a maximum positive value of 1 can result in negative values ​​less than or greater than -1. That is, the maximum negative value can be, for example, -0.5, or it can be -1.9, as in the example shown in Figure 4. Fig. The scaled engine noise diagram shown in Figure 5 is 179.

[0051] Fig. Figure 6 is an embodiment of a scaled engine noise diagram 181 with four principal parameters of a rise-fall-hold-release (ADSR) envelope 182 superimposed on the diagram. The ADSR envelope 182 is typically used in music synthesizers to emulate the sound of musical instruments. Advantageously, the techniques described herein apply the ADSR envelope 182 to provide knock sensor 23 data more quickly and efficiently for a specific noise analysis, as described below. For example, the scaled curve 180 may be characteristic (or include features) of a particular operating event (e.g., opening / closing of valves 62, 64 or peak firing pressure in combustion chamber 12) of the engine 10, and the ADSR envelope 182 developed for the scaled curve 180 may be used for future analysis of the operating event during the operation of the engine 10.

[0052] The four main parameters of the ADSR envelope are the attack (183), the decay (184), the hold (185), and the release (186). The attack (183) occurs from the beginning of the noise up to a peak amplitude (187) of the scaled curve (180). The decay (184) occurs during the downward movement from the peak amplitude down to a specified hold (185) level, which can be a certain percentage of the maximum amplitude. It should be understood that the order of the four parameters does not have to be attack, decay, hold, release. For some noises, the order might be attack, hold, decay, release. In such cases, an ADSR would be used instead of an ADSR envelope. For the sake of simplicity, this is referred to as an "ADSR envelope," but it should be understood that the term applies to a noise regardless of the order of the parameters.The Hold 185 level is the main level during the duration of the noise. In some embodiments, the Hold level may occur at 55% of the maximum amplitude. In other embodiments, the Hold level may be at least equal to or greater than 35%, 40%, 45%, 50%, 60%, or 65% of the maximum amplitude. A user or the ECU 25 can verify that the Hold level is as desired by determining whether the Hold level is maintained for at least 15% of the signature duration. If the Hold 185 level lasts for more than 15% of the signature duration, the Hold 185 level is set as desired. The Release 186 level occurs during the fall-down from the Hold 185 level back to zero. It should be noted that in some embodiments, the noise signal (e.g.,The scaled amplitude curve (180) can be filtered using a high-pass, low-pass, or band-pass filter to attenuate sections of the signal with frequencies that are not characteristic of the operational event. The specific filter applied to the noise signal can depend on the operational event being monitored. For example, if events (e.g., openings and closings) of valve 62, 64 are being monitored, a high-pass filter (e.g., greater than 10 kilohertz (kHz)) or a band-pass filter (e.g., between 10 and 20 kilohertz (kHz)) can be applied to the noise signal. If combustion events (e.g., peak fire pressure) are being monitored, a low-pass filter (e.g., less than 2 kilohertz (kHz)) can be applied to the noise signal.

[0053] Fig. Figure 7 shows the same scaled engine noise diagram 179, which is in the Fig. 5 and Fig. Figure 6 shows this with certain superimposed (e.g., layered) tones. After applying the ADSR envelope 182, the ECU 25 can extract three to five of the strongest frequencies in the noise and convert them into musical tones. For example, a lookup table mapping frequency ranges to musical tones can be used. Additionally or alternatively, equations based on the observation that pitch is typically perceived as the logarithm of frequency for equal temperament tuning systems, or equations for other musical temperament systems, can be used. In other embodiments, more or fewer frequencies can be extracted. In curve 181, shown in Fig. As shown in Figure 7, the three most prominent (e.g., extracted) notes are C#5, E4, and B3. However, it should be understood that these three notes are merely examples of possible notes and are not intended to limit which notes may be present in a recorded noise.

[0054] Fig. Figure 8 is a flowchart illustrating an embodiment of process 188 for characterizing a noise, such as a noise detected by the knock sensor 23. Characterizing the noise allows it to be plotted and sorted for analysis, including future analysis and / or real-time analysis. For example, in some embodiments, process 188 can be used to characterize a noise related to a specific operational event or action of the engine 10, such as peak firing pressure or the opening / closing of the inlet or outlet valves 62, 64. Furthermore, the noise can first be characterized during a baseline data acquisition process (e.g., an in-house baseline data acquisition process) before the engine 10 is implemented for normal or full-time operation, e.g., before it is sold, deployed to a site, implemented at a site, etc.For example, prior to normal engine operation, 10 different operating events (e.g., peak firing pressure, intake / exhaust opening / closing) can be tested by analyzing the noise detected during the operating event(s) (and by the knock sensor 23), whereby the noise signals or ADSR envelopes 182 of the noise signals can be matched with the fingerprints in relation to the operating events under test, thus generating a baseline. The noise can further be characterized during baseline data acquisition to detect valve blockage, valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation, valve leakage, and so on. It should be noted that the process 188 (e.g.,The basic data acquisition process can be used when the motor 10 is not operating at full capacity to simplify the processing of the noise signal. For example, process 188 can be used while only one valve (e.g., the exhaust valve 64 or the intake valve 82) is opening or closing to characterize the noise according to the opening or closing of the valve (e.g., the exhaust valve 64 or the intake valve 82). In other embodiments, process 188 can be used during partial or full operation of the motor 10.

[0055] In the illustrated embodiment, process 188 can be implemented as computer instructions or executable code stored in memory 74 and executable by processor 72 of the ECU 25. In block 190, a data sample is taken using the knock sensor 23 and the crankshaft sensor 66. For example, sensors 66 and 23 collect data from an operational event (e.g., closing of the exhaust valve 174) during baseline data acquisition and subsequently transmit the data to the ECU 25. As previously described, process 188 can be a baseline data acquisition process and can be executed while only certain components of the engine 10 are in operation. For example, process 188 can be executed while the exhaust valve 64 (or the intake valve 62) is opening and / or closing, so that the noise emitted, for example, during the closing of the exhaust valve 64 can be easily processed. Certain states (e.g.,Valve blockage, valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation, and valve leakage can be intentionally set for the basic data acquisition of the conditions. The ECU 25 then logs the angles of the crankshaft 54 ​​at the beginning and end of the data acquisition, as well as the time and / or crankshaft angle at the maximum (e.g., amplitude 187) and minimum amplitude. In fact, the angle of the crankshaft 54 ​​can be continuously plotted during the basic data acquisition process, enabling continuous recording of noise data across the angle of the crankshaft 54.

[0056] In block 192, the ECU 25 prepares the data from the knock sensor 23. This block 192 includes recording the raw data from the knock sensor 23 about the position or angle of the crankshaft 54 ​​(or, in some embodiments, about time). An exemplary recording of a raw engine noise was referred to as the amplitude recording 175 in Fig. 3 shown. Block 192 also includes scaling of the raw engine noise data. To scale the data, ECU 25 determines a multiplier that would result in a maximum amplitude of positive 1. It should be noted that the maximum negative value has no effect on the selection of the multiplier. ECU 25 then multiplies each data point (e.g., a data point in amplitude curve 177) by the multiplier to derive the scaled amplitude curve 180, as shown in Fig. Figure 5 is shown. It should be clear that the scaled engine noise diagram is 179 in Fig. Figure 5, which shows the scaled amplitude curve 180, is merely an example and is not intended to limit the scope of this disclosure to diagrams that look the same or similar to the scaled engine noise diagram 179.

[0057] In block 194, the ECU 25 applies the ADSR envelope 182 to the engine noise signal. The processing in this block was described in the section on Fig. 6. The ADSR envelope 182 is used to divide a noise data set into four different parameters or phases (rise 183, fall 184, hold 185, release 186). As discussed previously, it should be understood that the order of the four parameters need not be rise, fall, hold, release. For example, the order for some noises may be rise, hold, fall, and release, or any other possible sequence. For the sake of simplicity, this is referred to as an "ADSR envelope," but it should be understood that the term applies to a noise regardless of the order of the parameters. Traditionally, the ADSR envelope 182 is used in the process of reproducing a musical sound, such as that of a trumpet.However, in the techniques described herein, the ADSR envelope can be used to categorize and characterize noises so that they can be cataloged and sorted for later analysis, real-time analysis, or another purpose. The four main parameters of the ADSR envelope 182 are the rise 183, the fall 184, the hold 185, and the release 186. The rise 183 occurs from the beginning of the noise until the peak amplitude 187. The fall 184 occurs during the downward slope from the peak amplitude 187 to a defined hold 185 level, which is a specific percentage of the maximum amplitude. The hold 185 level is the main level during the duration of the noise. In some embodiments, the hold 185 level may occur at 55% of the maximum amplitude. In other embodiments, the hold-185 level can be at least equal to or greater than 35%, 40%, 45%, 50%, 60% or 65% of the maximum amplitude.A user or the ECU 25 can check if the hold level is as desired by determining whether the hold level (185) is maintained for at least 15% of the signature duration. If the hold level (185) lasts for more than 15% of the signature duration, the hold level (185) is set as desired. The release level (186) occurs during the downward movement from the hold level (185) back to zero. In block 194, the ECU 25 measures the time from zero to the maximum amplitude (187) (the maximum amplitude should have a value of 1). The ECU 25 then measures the downward movement duration from the maximum amplitude (187) to the set hold level (185). The ECU 25 then measures the level and the time during which the noise is maintained. Finally, ECU 25 measures the time it takes for the noise to fall from the holding level 185 to zero. ECU 25 then logs the ADSR vectors or segments that define the ADSR envelope 182.

[0058] In block 196, the ECU derives 25 tone information (e.g., musical tones) from the data. This block was described in the description of Fig. 7 is discussed. During this block, the ECU extracts 25 tonal information from the data, identifying, for example, the three to five loudest tones in the data. In another embodiment, any number of tones can be identified, e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 or more tones. Fig. Figure 7 shows three tones derived from signals C#5, E4, and B3. The ECU 25 can derive five or more tones from the data. Although Fig. Figure 7 shows the tones C#5, E4, and B3; it should be clear that these tones are examples and that the ECU 25 can derive any tones from the data. The ECU 25 then logs the derived tone information, which may include the frequency of the derived fundamental tones (i.e., the tones with the lowest frequency), the sequence of the derived fundamental tones, the frequency of the derived harmonics (i.e., the tones with a frequency that is an integer multiple of the fundamental frequency), the sequence of the derived harmonics, and any other relevant tonal information.

[0059] In block 198, the ECU 25 generates a fingerprint 200 based on the ADSR envelope 182 and the tonal information derived in blocks 194 and 196. Fingerprint 200 includes a characterization of the noise, breaking it down into its components (e.g., components 183, 184, 185, and 186 of the ADSR envelope 182, which can help identify valve opening / closing events and / or peak firing pressure) and quantifying these sections so that the noise can be cataloged, categorized, and sorted. At this point in the process, fingerprint 200 is primarily based on the ADSR envelope in block 194 and the tonal information derived in block 196.

[0060] In Block 202, fingerprint 200 is identified and verified. Using a number of techniques described later, fingerprint 200 can be modified or added to and then re-verified. It should be noted that, as previously described, fingerprint 200 can be logged in relation to a specific operational event or action of the engine 10. For example, Process 188 can correspond to a baseline data acquisition process that characterizes noise signals in relation to specific operational events that may occur during the operation of the engine 10. In particular, the noise signal captured during Process 188 can relate to a peak firing pressure, the closing or opening of the exhaust valve 64, the closing or opening of the inlet valve 62, or a combination thereof. Fingerprint 200 can be used as the specific event to be tested (e.g.,The operating event or condition associated with the baseline-determined event or condition is stored in memory 74 of the ECU 25.

[0061] In some embodiments, the process 188 (e.g., the basic data acquisition process) may include one or more additional steps that further process the noise signal or the ADSR envelope 182 to provide additional information regarding the operational event (e.g., opening or closing of the exhaust or intake valve 64, 62) or an action of the engine 10. For clarification, Fig. 9 an embodiment of a scaled engine noise diagram 181 with a scaled amplitude curve 180 belonging to an engine operating event (e.g. valve opening or closing event), an operating event indicator 203 belonging to the engine operating event and an ADSR envelope 182 belonging to the engine operating event according to the process of Fig. 8. It should be noted that, as previously described, the machine operating event and the corresponding diagram 181 in Fig. 9 can be tested (e.g., via the basic data acquisition process 188) while the motor 10 is not fully operating. Thus, the fluctuations in the depicted amplitude curve 180 occur at predictable times corresponding to the operational event and allow for simpler processing. In other words, in some embodiments, no components or operational events of the motor 10 can emit noise other than the components or operational events being tested (e.g., basic data). Additionally or alternatively, the noise signal can be filtered using a high-pass filter, a low-pass filter, or a band-pass filter to attenuate sections of the signal with frequencies that are uncharacteristic of the operational event. The specific filter applied to the noise signal can depend on the operational event being monitored. For example, if events of valves 62, 64 (e.g.,When monitoring openings and closings, a high-pass filter (e.g., greater than 10 kilohertz (kHz)) or a band-pass filter (e.g., between 10 and 20 kilohertz (kHz)) can be applied to the noise signal. When monitoring combustion events (e.g., peak fire pressure), a low-pass filter (e.g., less than 2 kilohertz (kHz)) can be applied to the noise signal.

[0062] With reference to process 188, which took place in Fig. As shown in 8, the fingerprint 200, which contains the information of the ADSR envelope 182, can be used. Fig. Figure 9 (e.g., with rise 183, fall 184, hold 185, and release 186) includes additional information relating to the operational event identified by fingerprint or for which the baseline is determined. For example, during process 188 (e.g., the baseline data acquisition process), an operational event indicator 203 may also be plotted over the scaled / normalized motor noise diagram 181. The operational event indicator 203 may, for example, be a diagram provided by a switch (e.g., a limit switch) that modulates between high and low to indicate the operational event during the baseline data acquisition process (e.g., process 188). For example, the switch may be actuated each time the operational event occurs during the baseline data acquisition process.In general, however, the switch cannot be included in the engine 10 during normal operation, as including both the switch and the knock sensor 23 can be redundant and expensive. Therefore, the switch and the corresponding operating event indicator 203 can be used during the basic data acquisition process (e.g., process 188) to more precisely determine a location in the ADSR envelope 182 where the operating event specifically occurs (e.g., depending on the operating event, within 2-4 degrees of the crankshaft 54 ​​angle). This allows the ADSR envelope 182 to be stored in the ECU 25 and later used during normal engine 10 operation to determine an angle or time of the crankshaft 54 ​​at which the operating event specifically occurs within the ADSR envelope 182.

[0063] In the illustrated embodiment, the operating event is the closing of the outlet valve 64, which is located in Fig. Figure 2 shows that when the exhaust valve 64 closes, the switch is actuated, causing the limit switch to move from low (e.g., low voltage) to high (e.g., high voltage). The switch transmits a signal from the operating event indicator 203 to the ECU 25, which can record the operating event indicator 203 on the scaled engine noise diagram 181. An intersection point 205 between the operating event indicator 203 and the ADSR envelope 182 can be stored along with the fingerprint 200 associated with the operating event (e.g., the closing of the exhaust valve 64). In the illustrated embodiment, the intersection point 205 is located approximately in the middle of the drop vector 184 (e.g., within 5–10 percent of the length of the drop vector 184 from its center point). In general, the closing of the outlet valve 64 occurs at the midpoint or middle region of the waste vector 184 (e.g.(if the middle area is an area defined by 5-10 percent of the length of the vector of the fall 184 on each side of the midpoint of the vector of the fall 184) and the coordinates of the midpoint of the vector of the fall 184 can be calculated using a geometric midpoint relation, e.g. P. i = [(X1 + X2) / 2, (Y1 + Y2) / 2], where P i The midpoint (and thus the intersection point 105) is given, X1 and X2 are the X-coordinates along axis 176 at each end of the drop vector 184, and Y1 and Y2 are the Y-coordinates along axis 178 at each end of the drop vector 184. It should be noted that in the illustrated embodiment, axis 178 encompasses time, but in another embodiment, axis 178 can contain information about the angle (e.g., the crank angle) of the crankshaft 54 ​​from the crankshaft sensor 66, which is correlated with time.

[0064] After determining the intersection point 205 (which in the illustrated embodiment relating to the closing of the outlet valve 64 is the midpoint of the drop vector 184 of the ADSR envelope 182), the fingerprint 200 (e.g. with the information of the ADSR envelope 182 and the information of the intersection point 205) can be stored for later analysis.

[0065] In some embodiments, it may be advantageous to check the fingerprint 200 to ensure that the fingerprint 200 is accurate, and it can be used to identify operational events during the normal operation of the internal combustion engine 10. Fig. Figure 10, for example, is a flowchart showing further details of an embodiment of process 202, which is described in Fig. Fingerprint 200, as depicted in Figure 8, is identified and verified. Process 202 can be implemented as computer instructions or executable code stored in memory 74 and executed by processor 72 of ECU 25. In decision 204, ECU 25 determines whether the noise signal is modulated (i.e., changing from one tone to another). If the signal is not modulated (decision 204), then ECU 25 proceeds to block 212 and attempts to find a suitable wavelet. A wavelet, effectively a segment or component of a wave, is a wave-like oscillation with an amplitude that starts at zero, increases, decreases, or both, and then returns to zero. Wavelets can be modified by adjusting the frequency, amplitude, and duration, making them very useful for signal processing.For example, with continuous wavelet transformations, a given signal can be reconstructed by integrating over the various modified frequency components. Exemplary "mother" wavelets include Meyer, Morlet, and Mexican Hat wavelets. However, new wavelets can also be generated if the mother wavelets are unsuitable.

[0066] When the tone is modulated (decision 204), the ECU 25 proceeds to decision 208 and determines whether the noise signal matches a chirplet. A chirp is a signal whose frequency increases or decreases over time. Just as a wavelet is a segment of a wave, a chirplet is a segment of a chirp. Similar to wavelets, the characteristics of a chirplet can be modified, and then multiple chirplets can be combined (i.e., a chirplet transformation) to approximate a signal. A chirplet can modulate up or down (i.e., change its frequency). In decision 208, the ECU 25 can adjust the modulation of chirplets to match them to the noise signal.If, after setting the chirplet modulation, ECU 25 can adjust the chirplets to match the noise signal, it logs whether a chirplet matched the signal. If so, it records the first frequency of the chirplet, the second frequency of the chirplet, and the chirplet modulation rate in frequency / crank angle or frequencies per second. ECU 25 then proceeds to block 210, where it shifts the phase of the noise signal to check fingerprint 200. In block 210, ECU 25 generates a noise signal based on the vectors of ASDR envelope 182 or other components, extracts tonal information, and matches a chirplet or wavelet. ECU 25 then phase-shifts (block 210) the generated signal, for example, by 180 degrees. If the characterization of the noise signal is correct, the phase-shifted generated noise signal should cancel out the noise signal.

[0067] If the noise signal does not match a chirplet (decision 108), ECU 25 proceeds to block 212 and attempts to match a wavelet to the noise signal. In block 212, ECU 25 selects one or more wavelets that may match the noise signal. The selected wavelet or wavelets may be a Meyer wavelet, a Morlet wavelet, a Mexican hat wavelet, or another suitable wavelet. In decision 214, ECU 25 determines whether the selected wavelet or wavelets match the noise signal. If the selected wavelet matches (decision 214), ECU 25 logs that a wavelet match occurred, the parent wavelet type, the first scaling range of the wavelet, and the second scaling range of the wavelet. If the wavelet fits (decision 214), the ECU 25 proceeds to block 210, in which the ECU 25 shifts the phase of the noise signal to check the fingerprint 200.If one of the selected wavelets does not match the noise signal (decision 214), the ECU 25 can proceed to block 216 and generate a wavelet. In decision 218, the ECU 25 determines whether the newly generated wavelet matches the noise signal. If the generated wavelet matches (decision 218), the ECU 25 logs that a wavelet match occurred, the first scaling range of the wavelet, and the second scaling range of the wavelet. If the generated wavelet matches the noise signal (decision 218), the ECU 25 proceeds to block 210, where the ECU shifts the phase of the noise signal to check fingerprint 200. If the new wavelet does not match (decision 218), the ECU 25 proceeds to block 220, where it characterizes the noise signal as broadband noise.

[0068] Returning to Block 210, if ECU 25 finds a chirplet or wavelet that matches the noise signal, it can verify the match by attempting noise suppression. Accordingly, in Block 210, ECU 25 generates a noise signal based on the vectors or other components of ASDR envelope 182, extracted tone information, and chirplet or wavelet matches. ECU 25 then shifts (Block 210) the generated signal by 180 degrees. ECU 25 then determines (Decision 222) whether the shifted signal cancels out the original noise signal within a desired residual tolerance.If the shifted signal cancels out the original noise signal within a desired residual tolerance (decision 222), the ECU 25 determines that fingerprint 200 is a "good" fingerprint 226 and proceeds to block 228, where the ECU 25 records the coefficients and associated data, which may include the root mean square (RMS) value of the signal or the RMS error.The ECU 25 can also log other data, including but not limited to crankshaft angles at the beginning or end of the signal, vectors of the ASDR envelope 182 or other ADSR components, spectral fundamentals, spectral harmonics, order of spectral tones, order of harmonics, whether a chirplet fits, the first chirplet frequency, the second chirplet frequency, the rate of chirplet modulation, whether a wavelet fits, the parent wavelet type, the first scale range of the wavelet, the second scale range of the wavelet, the maximum amplitude value and maximum time, the minimum amplitude value and minimum time, the RMS value of the signal, the RMS error of the signal relative to the generated signal, and whether the noise is classified as broadband noise or not. Furthermore, as previously described, the ECU 25 can record the intersection point 205 on the ADSR envelope 182, as shown in . Fig. Figure 9 shows that this logged data, and other data logged by the ECU 25, enables the ECU 25 to characterize and categorize known noises (e.g., according to certain operational events described in this disclosure) so that these noises can be stored on the memory component 74 of the ECU 25, perhaps transferred to another storage device, and then logged and sorted in a database for future analysis. On the other hand, if the ECU 25 determines (decision 222) that the shifted signal does not cancel out the original noise signal within a residual tolerance, the ECU 25 proceeds to block 224, where the noise signal is characterized as broadband noise.

[0069] It should be noted that, depending on the implementation, process 202 in Fig. 10 cannot be used according to the basic data determination procedure (e.g., process 188). For example, in some embodiments, it can be determined that fingerprint 200 is a “good fingerprint” 226 without using process 202. In any case, verified fingerprints 230 (e.g., fingerprint 200 and / or good fingerprint 226) can be stored in a database 232 for later access during an engine monitoring process 234, as in one embodiment of process 234 in Fig. Figure 11 shows that, for example, during the illustrated process 234, a noise from the motor 10 is detected (e.g., recorded or applied) (Block 236). As previously described, the noise can be detected via the knock sensor 23 or another sensor configured to detect noise or vibration from the motor 10. The noise signal can be processed according to the description of the Fig. 4-6 can be preconditioned (e.g., scaled, normalized, and / or filtered). The crankshaft sensor 66 can also detect, measure, or record the position of the crankshaft 54 ​​(e.g., in crank angles). Accordingly, the noise signal (e.g., the preconditioned noise signal) can be plotted against the position of the crankshaft 54 ​​via the ECU 25. As described above, in certain embodiments, the noise signal can be plotted against time instead of the position of the crankshaft 54.

[0070] Process 234 further includes accessing the fingerprints 230 in database 232 (block 238). For example, the ECU 25 can access fingerprint 230, which relates to a specific operating event monitored via process 234. Depending on the embodiment, the operating event (or operating state) may be a peak firing pressure, opening of the inlet valve 62, closing of the inlet valve 62, opening of the exhaust valve 64, closing of the exhaust valve 64, or any other operating event (or operating state) of the engine 10.

[0071] After accessing fingerprint 230, which corresponds to the operational event (or condition) monitored by ECU 25 via process 234, ECU 25 can correlate fingerprint 230 and the noise signal (e.g., the preconditioned noise signal) to determine whether the noise signal includes a segment that matches fingerprint 230. As previously described, fingerprint 230 may, for example, have ADSR envelope 182, which relates to the operational event monitored and generated during the basic data acquisition process (e.g., process 188). The ADSR envelope 182 of fingerprint 230 can be shifted or dragged along the time or position axis of the crankshaft 54 ​​of the noise signal (e.g., preconditioned noise signal) to determine whether fingerprint 230 matches any segment of the noise signal.For example, the ADSR envelope 182 of the fingerprint 230 can be directly compared or merged with sections of the noise signal, or one or more operational ADSR envelopes can be generated for sections of the noise signal (e.g. according to the descriptions of the . Fig. 6 and Fig. 7) to compare it with the ADSR envelope 182 of fingerprint 230. Furthermore, the operational event may generally have occurred within a known time range or positions (e.g., in crank angles) of the crankshaft 54. Thus, the portion of the noise signal processed by the ECU 25 to determine whether a portion of the noise signal matches fingerprint 230 can be reduced to the known range of time or positions of the crankshaft 54. It should be noted that the match between fingerprint 230 and the noise signal may not be an exact match. For example, fingerprint 230 may substantially match a portion of the noise signal and may be evaluated by a percentage of the accuracy of the match. A threshold value (e.g.,The information stored in memory 74 of ECU 25 enables ECU 25 to determine whether the percentage accuracy of the match between fingerprint 230 and the sound signal is sufficient to consider that a match exists between fingerprint 230 and the sound signal. The threshold can be at least equal to or greater than 75% match, 80% match, 85% match, 90% match, 95% match, 97% match, 98% match, 99% match, or 100% match.

[0072] In decision 242, the ECU 25 determines whether fingerprint 230 matches any segment of the noise signal (e.g., preconditioned noise signal) from block 236. If the correlation in block 240 is a match in decision 242, the monitored operational event is verified. Furthermore, as shown in block 244, the specific location of the operational event (e.g., with respect to time or crank angle of the crankshaft 54) can be determined. For example, as previously described, the operational event can be located at the intersection 205 (e.g., between the ADSR envelope 182 and the operational event indicator 203) in Fig. 9, which in some embodiments corresponds to the midpoint on the vector of the drop 184 of the ADSR envelope 182. Accordingly, the ECU 25 can superimpose the ADSR envelope 182 of the fingerprint 230 on the noise signal plotted against the position of the crankshaft 54 ​​and determine that the operating event occurred at the x-coordinate (e.g., the time or position coordinate of the crankshaft 54) of the intersection point 205 on the ADSR envelope 182.

[0073] If the fingerprint 230 does not match any part of the noise signal during decision 242, process 234 can either return to block 236 (e.g., capturing an engine noise) or to block 238 (accessing the fingerprint(s) in the database). For example, in some embodiments, process 234 can be used to monitor multiple operational events. Accordingly, process 234 can involve accessing multiple fingerprints 230 for correlation with the noise signal. The multiple fingerprints 230 can be accessed in one step, or each fingerprint 230 can be accessed and then independently correlated with the noise signal to determine and verify operational events.

[0074] According to the present disclosure, it should be noted that operating event(s) and conditions can be an operating event or condition of the engine 10. For example, the operating event can be the opening of the exhaust valve 64, the closing of the exhaust valve 64, the opening of the intake valve 62, the closing of the intake valve 62, a peak firing pressure, or any other operating event of the engine 10. The event can additionally include valve blockage, valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation, or valve leakage. Furthermore, it should be noted that the crank angle at which the operating event occurs can be determined by the same or similar process steps described above.For example, in some embodiments, the operational event may occur at a different point along the decay vector 184 or along one of the other vectors of the ADSR envelope 182. The operational event indicator 203, which is in . Fig. As shown in Figure 9, the ECU 25 can be provided by a limit switch or by another mechanism configured to capture the operating event during the basic data acquisition process (e.g., process 188), which may not be present in the motor 10 during normal operation. Furthermore, it should be noted that the fingerprints 200, 226, 230 associated with each operating event may vary for each operating event and may vary for each model, make, or series of motors 10. Thus, the basic data acquisition process (e.g., process 188) to determine fingerprints 200, 226, 230 for different operating events can be performed for each specific motor 10, and each motor 10 may have different fingerprints 200, 226, 230 for the same operating event.

[0075] With reference to Fig. Figures 12-16 represent statistical techniques useful in detecting events of the valve train 70 and the drift or deviation of the valve clearance 80. For example, shows Fig. Twelve embodiments of two diagrams 300 and 302 sharing the same x-axis 304. Diagram 300 comprises a dynamic valve lift in a thermodynamic cycle versus the crank angle on the x-axis 304, with corresponding measured valve lift data on a y-axis 306. Diagram 302 is correlated with diagram 300 and has the same crank angle degree x-axis 304. The uppermost diagram 300 includes signals or curves 308 and 310, where curve 308 is measured valve lift, while curve 310 is a valve velocity (e.g., a derivative of valve lift 308 over time). Diagram 302 includes a signal 312 (e.g., a vibration or noise signal) representing the dynamic response of the cylinder head 14 as measured by the knock sensor(s) 23. While signal 312 is typically used to detect knocking (e.g.To detect engine “pinging”, it was found that signal 312 has components that are representative, for example, of valves 62, 64, which are in contact with valve seats and / or other components of the valve train 70. A section of the decay 312 of signal 312, such as the decay noise 314 of cylinder head 14, can be used to extract the dynamics of the valve train 70.

[0076] In the illustrated embodiment, a valve closing event 316 is identified, and the dynamic noise response signal 312, detected by the knock sensor, can contain data representing, for example, a valve timing that can be correlated with the actual valve clearance 80 during operation. Monitoring the valve timing drift can be used to optimize the maintenance adjustment interval, detect early valve and seat ring wear, and / or identify a valve train structural failure (such as valve stem disengagement 72). To derive the valve lift event 316, the noise signal 312 can be analyzed to search for a pattern representative of a start when the dynamic section 314 is shown. To derive a model suitable for identifying specific states of the valve train 70 (e.g.,Valve clearance drift 80, valve stem separation 70), a test bed can be used to detect vibration and acceleration of the engine 10. One or more vibrometers (e.g., differential laser vibrometers) and one or more accelerometers can be used, for example, to acquire valve seat data 308, 310 and valve noise data 312; and the crank angle sensor 66 can provide crank angle data to define the x-axis 304. Measurements can be performed in steps, such as a first step that involves identifying the natural frequencies of the signals 308, 310, and / or 312. Certain angles 318, 320, 322, for example, can each have a respective natural frequency F and associated phase Φ(F).

[0077] A second full-load operation step can also be used to observe the engine 10 during full load. The full-load operation can be further subdivided into a first phase, observed during the operation of two or more exhaust (or intake) valves 62, 64 per cylinder, and a second phase, observed during the operation of a single exhaust (or intake) valve 62, 64 per cylinder. The observations 308, 310, 312 can then be used to create certain graphs or models, such as the graphs shown in the Fig. 13 and Fig. 14 are shown, to derive. In one example, a quantity (e.g. |F|, where F is a frequency range) and a phase (e.g. Φ(F)) of a dynamic response of the cylinder head 14, measured by one or more sensors 23 in a crank angle range (e.g. between -720° to 720°), comprise information regarding seat excitation force and timing of the valves 62, 64, respectively, and can thus be used to derive the graphs or models of the Fig. 13 and Fig. to generate 14.

[0078] More precisely, Fig. Figure 13 shows an embodiment of a graph 400, which displays a valve seat velocity change (e.g., mm / s on a y-axis 402) and an exhaust clearance (e.g., mm on an x-axis 404). A second graph 406, which is shown in Fig. Graph 13 shows a valve noise (measured as the maximum of the cylinder head accelerometer signal around the valve closing time) on a y-axis 408 and an exhaust clearance (e.g., mm) on an x-axis 410. Graph 400 can be derived from vibrometer data, while graph 406 can be derived from accelerometer data. Legend 412 shows the maximum, minimum, and quartiles for boxes 414, 416, 418, and 420 of graphs 400 and 406. In the examples shown, graph 400 is a box diagram, with boxes 414 representing a first analysis phase (e.g., an analysis phase using two exhaust valves 64 or two intake valves 62) and boxes 416 representing a second analysis phase (e.g., an analysis phase using a single exhaust valve 64 or a single intake valve 62).Similarly, graph 406 is a box diagram with boxes 418 that were analyzed during the first analysis phase and boxes 420 that were analyzed during the second analysis phase.

[0079] As can be observed in graph 400, when the clearance (x-axis 404) increases (i.e., drifts), the valve seat excitation (y-axis 402) also increases. However, using only vibrometer data cannot correlate with (or predict) the valve clearance 80 and / or the valve clearance drift as accurately as desired. Similarly, graph 406 shows that the valve noise (x-axis 410) increases due to the higher valve seat excitation. However, noise data also cannot correlate with (or predict) the valve clearance 80 and / or the valve clearance drift as accurately as desired.

[0080] Advantageously, it was observed that adding crank angle data (e.g., data derived from sensor 66) to the data in graphs 400 and 406 can improve prediction accuracy. Accordingly, this shows Fig. 14 embodiments of graphs 430, 432, which include crank angle measurements. In particular, graph 430 includes the degree of crank angle closure (e.g., measured by a laser vibrometer) on a Y-axis 434, while graph 432 also includes the degree of crank closing (e.g., measured by a trigger knock sensor signal) on a Y-axis 436. A legend 438 is also shown, which presents the maximum, minimum, and quartiles (e.g., over a population of 100 thermodynamic cycles) for boxes 438, 440, 442, 444 of graphs 430 and 432 (statistically representative of steady-state engine operation).

[0081] Crankshaft angle measurements can provide improved accuracy, including the predictive accuracy of the exhaust clearance and / or clearance drift shown on the x-axes 446, 448. For example, long-term drift tendencies 450 and / or a short-term drift 452 (e.g., shaft 72 removal) can be more easily derived. In certain embodiments, the data used to derive graphs 300, 302, 400, 406, 430, and / or 432 can be used to construct certain models or graphs suitable for acquiring crankshaft angle data using the crankshaft angle sensor 66 and engine noise data using the knock sensor(s) 23, and for deriving certain engine conditions that are useful, for example, in condition-based maintenance, based on the acquired data, such as graphs or models related to the Fig. 15 and Fig. 16 are shown.

[0082] With reference to Fig. Figure 15 shows an embodiment of a statistical valve train model 460 (e.g., a threshold valve clearance model) that may be suitable for detecting certain valve train states. The model 460 can be provided as computer instructions or code stored in memory 74 and executable by processor 72. The model 460 can also be stored and executed by external systems, such as external computer systems. In the embodiment shown, the model 460 can be generated by various techniques, such as mathematical techniques suitable for analyzing the data of graphs 300, 302, 400, 406, 430, and / or 432. For example, curve-fitting techniques (e.g., polynomial curve fitting, least squares regression analysis, linear interpolation, nonlinear interpolation), data mining techniques (e.g.,Data cluster analysis, k-mean analysis), regression analysis and the like, are used to transform the data from sensors 23, 66 into model 260.

[0083] As shown, model 460 comprises a graph or curve 462 with statistical quartiles 464 and 466 suitable for expressing median values ​​468 and variations from the median value 468. For example, a detail section 470 of the model is shown, which includes the first quartile 464, the third quartile 466, and the median 468. In fact, model 462 can provide improved analysis by allowing a data point to be identified as a median or quartile data point, or as a point outside the curve 462. Model 460 includes an x-axis 472, which is representative of a hot valve clearance or a current valve clearance 80 (e.g., in mm) present during engine operation. Model 460 additionally features a y-axis 474, which represents the degrees of rotation of the released crankshaft. In use, the data from the knock sensor 23 can be used to determine that certain dynamics of the valve train 70 have occurred.For example, decay section 314 of . Fig. 12, that certain dynamics of the valve train 70 have occurred with respect to the valve train 70. The crankshaft sensor 66 can then be used to determine the crank angle at which the dynamic occurred. Given a crank angle (e.g., determined statistically by triggering the accelerometer sensors), curve 462 can be used to derive the hot valve clearance or valve clearance 80 present during the operation of the machine 10. For example, a horizontal line can be drawn from the y-axis at the measured crank angle to intersect curve 462, and the intersection point can then correspond to the currently existing valve clearance 80. Accordingly, noise can be detected and analyzed to determine the current exhaust valve clearance 80 (e.g., point on the x-axis 472).Additionally, the current exhaust valve clearance 80 can be determined as a valve clearance in the middle quartile, valve clearance in the first quartile or valve clearance in the second quartile, with the point falling into the first quartile 464, the third quartile 466 and the median 468.

[0084] When the valve clearance settings are adjusted, a user can log or otherwise save the valve clearance setting. The valve drift can then be derived using the equation: Valve drift = Clearance setting - Hot clearance. If, for example, valves 62 and 64 wear during use, successive adjustments can be made and recorded. Since each valve clearance setting is successively recorded, the model can be... Fig. Device 15 can be used to obtain valve drift over time, for example, due to a reduction in clearance 80. If, for example, clearance 80 is reduced, the angle of a triggered crankshaft may also drift lower in the combustion cycle, as shown by trend 276. Accordingly, valve timing drift can be detected, and certain control measures can be implemented, for example, via devices 27 and 31. Additionally or alternatively, condition-based maintenance can be performed, for example, to trigger a maintenance interval and compensate for valve timing drift during the engine's lifetime. Instead of performing maintenance according to a fixed schedule, the techniques described herein can monitor engine noise, detect valve timing drift, and then alert or inform an interested party or perform other actions (e.g., automatically adjust the timing of valves 62 and 64).

[0085] Similarly, the drift of the valve clearance 80 during operation can be detected when related to the wear acceleration of the cylinder head 14, the overstretching of the components of the valve train 70, the loosening of the adjusting screw, and so on. In fact, by triggering engine noise, model 460 can be used to derive the hot valve clearance 80 (e.g., x-axis) and then to deduce how the valve clearance 80 may change over time due to valve clearance drift. Such movements can then be applied to plan the adjustment of the valve clearance 80, detect clearance consumption rates, determine whether the cylinder head 14 is wearing as expected, and / or determine whether there is overstretching of the components of the valve train 70. Other models can also be created based on the techniques described here.

[0086] For example, it shows Fig. 16 An embodiment of model 480 (e.g., valve timing change model) suitable for deriving a sudden valve timing change due to a fault in a valve train structure, such as a valve stem 72, during operation of the engine 10. Model 480 comprises an x-axis 482, which is representative of the exhaust clearance 80 and is measured in increasing mm. Model 480 also comprises a y-axis 484, which is representative of a valve 62, 64, with the time drift being measured in degrees. The box diagrams 486 are the variation of the valve timing due to valve slippage, derived by means of the knock sensor 23 installed next to the exhaust valve 64; the boxes 488 are representative of the variation from the knock sensor 23 installed near the intake valve 62.The derived data and boxes 490 are representative of data derived using a charge accelerometer bolted to the cylinder head. As shown, the angle of the triggered crank of the cylinder head acceleration signal can be detected and used to determine sudden valve train failure, actual valve clearance (x-coordinate on x-axis 482), and similarly, a valve timing drift (y-coordinate on y-axis 484) can be derived. By correlating the instantaneous clearance with the valve timing, it can be determined whether boxes 486, 488, and 490 have a point with the x,y-coordinate. If this is the case, it is likely that the valve stem 72 has separated. The closer the valve timing drift is to the median of a box (e.g., boxes 486, 488, and 490), the more likely the occurrence of the interruption event.In other words, a rapid change in valve timing can be detected by applying statistical analysis to the cylinder head acceleration signal, for example, curve 462. This can be achieved if, during engine operation, it is found that the hot-clearance valve derived from curve 462 varies by a certain amount (e.g., between 1-2 mm, 0.5-4 mm, or more) at a specific time (e.g., between 0.5 and 10 seconds, 0.05 and 10 minutes, or more).

[0087] Referring to Fig. Figure 17 shows a flowchart of an embodiment of a process 500, which may be suitable for applying certain signal processing techniques of the knock sensor 23 to detect states of the valve train 70 and to derive control measures based on the detected states. The process 500 can be implemented as computer code or instructions that are stored in memory 74 and executable by processors 72. In the embodiment shown, the process 500 can first receive a signal 504 from one or more knock sensors 23 (block 502). The signal 504 can then be processed using ADSR techniques (block 506) and / or statistical techniques (block 508). The ADSR techniques used were previously described with respect to the Fig. 4-11 described, while the statistical techniques previously referred to the Fig. 12-16 were described.

[0088] Applying (Block 506) ADSR signal processing can lead to the inference (Block 510) that an event of the valve train 70 has occurred, such as a drift of the valve clearance 80 out of range, a valve jam, a valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation, or a valve leak. Likewise, applying (Block 508) statistical techniques can lead to the inference (Block 510) of events of the valve train 70, including a drift of the valve clearance 80 out of range, a valve jam, a valve interruption, excessive valve timing drift, excessive valve seat velocity, complete valve clearance consumption during operation, a valve leak, and so on.

[0089] Certain detected valve events can be corrected or mitigated by specific control measures. For example, valve clearance or valve timing drift outside a certain range can be corrected via the valve adjusting device 27 and / or by applying variable valve timing via the camless valve actuator 31 or similar systems. Variable valve timing can include delayed intake valve closing techniques, in which the intake valve 62 is held open longer than normal, causing the piston 20 to force air through the cylinder 26 and back into an intake manifold during the compression stroke. Variable valve timing can also include early intake valve closing techniques, in which the intake valve 62 closes earlier than normal, for example, midway through the intake stroke, resulting in a reduction of pumping losses.The variable valve timing can additionally feature an early / late exhaust valve closure 64, which allows manipulation of an exhaust gas quantity remaining in the cylinder 26.

[0090] To provide control measures, Process 500 (Block 512) can derive features related to the event derived in Block 510. For example, if the event is a valve clearance drift event, Process 500 (Block 512) can derive a valve clearance 80 value and / or a drift value over time. If the event is a valve timing event, a measure of timing deviation can be derived (Block 512). Similarly, excessive valve seat velocity features, such as seat velocity, can be derived (Block 512). Other derivations include clearance 80 consumption measures, valve leakage measures, and the like. Based on the derived (Block 510) event(s) and the derived (Block 512) features for the event(s), Process 500 can adjust the valve lift and / or valve timing (Block 514).For example, the ECU 25 can control the valve adjusting device 27 and / or a camless valve actuator 31 to change the valve lift and / or the valve timing. Accordingly, the process 500 can detect and respond to a variety of valve states via control actions (block 514), which can improve engine life, engine efficiency, and maintenance schedules.

[0091] Technical effects of the invention include the application of noise sensor data, such as knock sensor data, to generate one or more models suitable for deriving valve drift. The models can be ADSR models and statistical models. In one embodiment, the one or more models can then be used to provide condition-based maintenance (CBM) for an engine. For example, the one or more models can monitor engine noise, detect valve timing drift, and then alert or inform an interested party or perform other actions (e.g., automatically adjust the valve timing). Similarly, clearance consumption can be detected if it is related to the cylinder head wear acceleration and / or the overextension of the components of the valve train 70.

[0092] This written description uses examples to disclose the invention, including the best mode of use, and to enable any person skilled in the art to put the invention into practice, including making and using any devices or systems and carrying out the methods contained therein. The patentable scope of the invention is defined by the claims and may include other examples that might occur to a person skilled in the art. Such other examples shall be within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they contain equivalent structural elements with insignificant differences from the literal language of the claims.

[0093] In one embodiment, a system comprises an engine control system configured to control an engine. The engine control system includes a processor configured to receive a vibration signal detected by a knock sensor located in the engine. The processor is further configured to correlate the vibration signal with a fingerprint that includes at least one ADSR envelope indicating the operating event of a valve train of the internal combustion engine, to analyze the vibration signal with a statistical valve train model, or a combination thereof. The processor is also configured to detect whether the operating event has occurred based on the correlation of the noise signal with the fingerprint, based on the analysis of the vibration signal with a statistical valve train model, or a combination thereof, and to control the valve train based on the operating event.

Claims

[1] System, encompassing: an engine control system configured to control an internal combustion engine (10), the engine control system comprising a processor (72) configured to: to receive a vibration signal that is detected by a knock sensor (23) arranged in the internal combustion engine (10); to correlate the vibration signal with a fingerprint (200) that has at least one ADSR envelope (182) indicating the operating event of a valve train (70) of the internal combustion engine (10), to analyze the vibration signal with a statistical valve train model (460) or a combination thereof; based on the correlation of the vibration signal with the fingerprint (200), based on the analysis of the vibration signal with a statistical valve train model (460) or a combination thereof, to detect whether the operational event has occurred; and to control the valve train (70) based on the operational event; wherein the processor (72) is configured to derive the ADSR envelope (182) from a basic data acquisition noise signal indicating the operational event and to plot the ADSR envelope and operational event indicator data over time to derive a location of the ADSR envelope (182) where the operational event occurs; wherein the location of the ADSR envelope (182) where the operational event occurs lies within a mid-range of a decay vector (184) of the ADSR envelope (182). [2] System according to claim 1, wherein the processor (72) is configured to control the valve train (70) by controlling a valve adjustment device (27), a variable valve timing or a combination thereof. [3] System according to claim 2, wherein the valve adjusting device (27) comprises a variable valve tappet and wherein controlling the variable valve timing comprises controlling a valve actuating device (31), the variable valve tappet or a combination thereof. [4] System according to one of the preceding claims, wherein the processor (72) is configured to detect operating event characteristics for the operating event, wherein the engine control system is configured to control the valve train (70) based on the operating event and the operating event characteristics. [5] System according to claim 4, wherein the operating event features include a valve clearance distance, a valve clearance consumption, a valve timing drift measurement, a valve seat velocity or a combination thereof. [6] System according to any of the preceding claims, wherein the processor (72) is configured to: to receive a crankshaft signal detected by a crank angle sensor (66) arranged in the internal combustion engine (10), wherein the crank angle signal is representative of an engine crank angle; and to monitor a valve timing by deriving a cylinder head acceleration measurement using the vibration signal received from the knock sensor (23), wherein the control system to monitor the valve timing is to derive a valve clearance based on the vibration signal, the engine crank angle and a threshold valve clearance model contained in the statistical valve train model (460). [7] Procedures, comprehensive: Receiving a vibration signal detected by a knock sensor (23) arranged in an internal combustion engine (10); Correlating the vibration signal with a fingerprint (200) that has at least one ADSR envelope (182) indicating the operating event of a valve train (70) of the internal combustion engine (10), analyzing the vibration signal with a statistical valve train model (460) or a combination thereof; Detecting whether the operational event has occurred, based on correlating the vibration signal with the fingerprint (200), based on analyzing the vibration signal with a statistical valve train model (460), or a combination thereof; and Control of the valve train (70) based on the operational event; wherein the ADSR envelope (182) is derived from a basic data acquisition noise signal indicating the operational event and the ADSR envelope and operational event indicator data are plotted over time to derive a location of the ADSR envelope (182) where the operational event occurs; wherein the location of the ADSR envelope (182) where the operational event occurs lies within a mid-range of a decay vector (184) of the ADSR envelope (182). [8] Tangible, non-transitory, computer-readable medium that stores code set up to induce a processor (72): to receive a vibration signal that is detected by a knock sensor (23) arranged in an internal combustion engine (10); to correlate the vibration signal with a fingerprint (200) that has at least one ADSR envelope (182) indicating the operating event of a valve train (70) of the internal combustion engine (10), to analyze the vibration signal with a statistical valve train model (460) or a combination thereof; based on correlating the vibration signal with the fingerprint (200), based on analyzing the vibration signal with a statistical valve train model (460), or a combination thereof, to detect whether the operational event has occurred; and to control the valve train (70) based on the operational event; wherein the code is set up to cause the processor (72) to derive the ADSR envelope (182) from a basic data acquisition noise signal indicating the operational event and to plot the ADSR envelope and operational event indicator data over time to derive a location of the ADSR envelope (182) where the operational event occurs; wherein the location of the ADSR envelope (182) where the operational event occurs lies within a mid-range of a decay vector (184) of the ADSR envelope (184).

Images