Knock control method and system for gas engine

Abstract

This application discloses a knock control method and system for a gas engine, belonging to the field of knock recognition technology. A cloud server stores a pre-trained knock recognition model, allowing the vehicle's control device to simply collect engine speed signals, engine torque signals, acoustic emission signals, and camshaft signals. After processing the engine speed, torque, and acoustic emission signals based on the camshaft signal, the data is sent to the cloud server, which then uses the pre-trained knock recognition model to perform knock recognition and obtain the result. This improves the accuracy and efficiency of knock recognition. Furthermore, knock recognition can be performed using acoustic emission signals collected by a single acoustic emission sensor located at the target measurement point, reducing costs compared to solutions requiring separate knock sensors on each cylinder.

Description

Technical Field

[0001] This application belongs to the field of knock detection technology, and particularly relates to a knock control method and system for a gas engine. Background Technology

[0002] Engine knock is a phenomenon caused by abnormal combustion in an engine. When engine knock occurs, it results in poor emissions, fuel economy, and drivability. In severe knocking, it can even cause mechanical damage. Therefore, effective knock detection is necessary to implement appropriate knock suppression measures. Knock signal processing is one of the essential technologies for effective knock detection.

[0003] In existing technologies, knock detection is mostly achieved by installing knock sensors on each cylinder of the engine. The detection accuracy largely depends on the design of the engine block and reinforcing ribs, as well as the placement of the knock sensors. This places high demands on the shape, size, and layout of the engine components, increasing R&D and vehicle costs. Especially at low or high speeds, the knock detection effect is poor in some engines or cylinders, leading to false positives and false negatives. Summary of the Invention

[0004] The embodiments of this application provide a knock control method and system for a gas engine, which can improve the accuracy and efficiency of knock identification and reduce costs.

[0005] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.

[0006] According to a first aspect of the embodiments of this application, a knock control method for a gas engine is provided, the method comprising:

[0007] Acquire a first signal, which includes an engine speed signal, an engine torque signal, and an acoustic emission signal. The acoustic emission signal is acquired by an acoustic emission sensor located at a target measuring point, and the target measuring point is the measuring point most sensitive to the acoustic emission source of each cylinder.

[0008] Acquire camshaft signals and process the first signal based on the camshaft signals;

[0009] The processed first signal is sent to the cloud server so that the cloud server can call the trained detonation recognition model to recognize the first signal and obtain the recognition result.

[0010] When the identification result indicates that knocking has occurred, knocking elimination processing is performed according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation.

[0011] In some embodiments of this application, based on the foregoing scheme, the method includes the following steps before acquiring the first signal:

[0012] Analyze the sensitivity of different measuring points to various acoustic emission sources;

[0013] Select the measurement point with the highest sensitivity as the target measurement point;

[0014] An acoustic emission sensor is installed at the target measuring point so that the acoustic emission sensor can collect the acoustic emission signals of each cylinder.

[0015] In some embodiments of this application, based on the foregoing scheme, the analysis of the sensitivity of different measurement points to various acoustic emission sources includes:

[0016] The center of the bottom of each cylinder head is taken as the excitation point, and the first acoustic emission sensor is set at each excitation point;

[0017] The center position from the air intake port to the exhaust port on each cylinder head is taken as the measuring point, and a second acoustic emission sensor is set at each measuring point;

[0018] An acoustic emission source is applied at each of the excitation points, and the first acoustic emission signal collected by each of the first acoustic emission sensors and the second acoustic emission signal collected by each of the second acoustic emission sensors are acquired.

[0019] Based on each of the first acoustic emission signals and each of the second acoustic emission signals, the energy attenuation of each measuring point relative to each acoustic emission source is calculated.

[0020] Based on the energy attenuation of each measuring point relative to each acoustic emission source, the sensitivity of different measuring points to each acoustic emission source is determined. The smaller the energy attenuation, the more sensitive the measuring point is to the acoustic emission source.

[0021] In some embodiments of this application, based on the foregoing scheme, determining the sensitivity of different measuring points to each acoustic emission source according to the energy attenuation of each measuring point relative to each acoustic emission source includes:

[0022] Calculate the total energy attenuation of each measuring point relative to each acoustic emission source;

[0023] The sensitivity of each measuring point to each acoustic emission source is determined based on the total energy attenuation. The smaller the total energy attenuation, the stronger the sensitivity of the measuring point to each acoustic emission source.

[0024] In some embodiments of this application, based on the foregoing scheme, processing the first signal according to the camshaft signal includes:

[0025] Periodic identification is performed based on the camshaft signal;

[0026] Based on the identified period, the first signal is processed into multiple period-divided first signals to obtain multiple period-divided first signals.

[0027] The first signal corresponding to each sub-cycle is subjected to periodic averaging processing so that each sub-cycle corresponds to an engine average speed signal, an engine average torque signal and an average acoustic emission signal.

[0028] In some embodiments of this application, based on the foregoing scheme, when the identification result indicates that knocking has occurred, knock elimination processing is performed according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation, including:

[0029] When the identification result is severe knocking under high load, the throttle is reduced to decrease the load and eliminate knocking.

[0030] When the identification result is high load slight knocking, but it is single cylinder knocking, the ignition advance angle of the cylinder where knocking occurred is reduced to eliminate knocking;

[0031] When the identification result is high load slight knocking, but it is multi-cylinder knocking, the excess air coefficient is reduced to eliminate knocking.

[0032] In some embodiments of this application, based on the foregoing scheme, when the identification result indicates that knocking has occurred, knock elimination processing is performed according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation, including:

[0033] When the identification result is severe knocking under medium to low load, but it is multi-cylinder knocking, control the throttle to reduce the load in order to eliminate knocking;

[0034] When the identification result is severe knocking under medium and low loads, but it is single-cylinder knocking, the ignition advance angle of the cylinder where knocking occurred is reduced to eliminate knocking;

[0035] When the identification result is mild knocking under medium to low load, but it is single-cylinder knocking, the ignition advance angle of the cylinder where knocking occurred is reduced to eliminate knocking;

[0036] When the identification result is mild knocking under medium to low load, but it is multi-cylinder knocking, the excess air coefficient is reduced to eliminate knocking.

[0037] In some embodiments of this application, based on the foregoing scheme, the method further includes training the detonation recognition model, including:

[0038] The first signal and camshaft signal under different working conditions are collected, and the first signal is processed based on the camshaft signal;

[0039] Training and test sets are constructed based on the processed first signal;

[0040] After training the detonation recognition model using the training set, test the recognition accuracy of the detonation recognition model using the test set.

[0041] The training process of the detonation recognition model is controlled based on the test results.

[0042] According to a second aspect of the embodiments of this application, a knock control system for a gas engine is provided, comprising:

[0043] An acoustic emission sensor is set at a target measuring point, which is the measuring point most sensitive to the acoustic emission source of each cylinder.

[0044] The control device includes a data acquisition module, a processing module, a transmission module, and a control module;

[0045] The cloud server is communicatively connected to the control device.

[0046] The acquisition module is connected to the acoustic emission sensor, and the acquisition module is used to acquire the first signal and the camshaft signal;

[0047] The processing module is connected to the acquisition module, and the processing module is used to process the first signal according to the camshaft signal;

[0048] The sending module is connected to the processing module, and the sending module is used to send the processed first signal to the cloud server;

[0049] The cloud server is used to call the trained detonation recognition model to identify the first signal and obtain the recognition result;

[0050] The control module is used to acquire the identification result, and when the identification result indicates that knocking has occurred, it performs knock elimination processing according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation.

[0051] In some embodiments of this application, based on the foregoing scheme, the control device is located next to the gas engine, and the control device further includes a heat-insulating coating and a radiator.

[0052] Based on the technical solution proposed in this application, the cloud server stores a pre-trained knock recognition model. This allows the vehicle's control device to simply collect engine speed signals, engine torque signals, acoustic emission signals, and camshaft signals. After processing the engine speed, torque, and acoustic emission signals based on the camshaft signal, the data is sent to the cloud server. The cloud server then uses the pre-trained knock recognition model to perform knock recognition and obtain the result. This improves the accuracy and efficiency of knock recognition. Furthermore, knock recognition can be performed using acoustic emission signals collected by a single acoustic emission sensor located at the target measurement point, which reduces costs compared to solutions requiring separate knock sensors on each cylinder.

[0053] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0054] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0055] Figure 1 This is a first structural schematic diagram of a gas engine knock control system provided in an embodiment of this application.

[0056] Figure 2 This is a second structural schematic diagram of a gas engine knock control system provided in an embodiment of this application.

[0057] Figure 3 This is a flowchart of a knock control method for a gas engine provided in an embodiment of this application.

[0058] Figure 4 This is a flowchart of the steps performed before acquiring the first signal, provided in an embodiment of this application.

[0059] Figure 5 This is a flowchart of the steps for analyzing the sensitivity of different measurement points to various acoustic emission sources, provided in one embodiment of this application.

[0060] Figure 6 This is a schematic diagram showing the installation positions of the first and second acoustic emission sensors provided in the embodiments of this application.

[0061] Figure 7 This is a schematic diagram showing the kx values ​​between different measurement points and various acoustic emission sources provided in the embodiments of this application.

[0062] Figure 8 This is a schematic diagram showing the sum of the kx values ​​of all acoustic emission sources at various different measurement points provided in the embodiments of this application.

[0063] Figure 9 This is a flowchart illustrating the steps of processing a first signal based on a camshaft signal, as provided in an embodiment of this application.

[0064] Figure 10 This is a flowchart illustrating the steps involved in training a detonation recognition model, as provided in an embodiment of this application.

[0065] Figure 11 This is a flowchart of the first step of knock elimination processing provided in the embodiments of this application when the identification result is knocking, based on the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation.

[0066] Figure 12 This is a flowchart of the second step of knock elimination processing provided in the embodiments of this application when the identification result is knocking, based on the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation.

[0067] Figure 13 This is a schematic diagram of the knock control processing logic provided in the embodiments of this application.

[0068] Figure label:

[0069] Acoustic emission sensor 1, control device 2, cloud server 3, acquisition module 21, processing module 22, transmission module 23, control module 24, heat insulation coating 25, heat sink 26. Detailed Implementation

[0070] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0071] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0072] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0073] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0074] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "multiple" means two or more.

[0075] To enable those skilled in the art to better understand this application, the application scenarios involved in this application will be briefly described first.

[0076] Knocking in a gas turbine engine is an abnormal combustion phenomenon that occurs under specific conditions, primarily manifested as an abnormal knocking sound during engine operation. Knocking occurs when, under certain conditions (such as an excessively high compression ratio), the combustion process in a gas turbine engine becomes abnormal, resulting in high-frequency, large-amplitude fluctuations in the pressure curve. At this time, the flame propagation speed and flame front shape change drastically, leading to strong pressure waves in the combustion chamber, which in turn causes engine vibration, resulting in reduced engine power, increased fuel consumption, and increased noise.

[0077] Currently, most knock detection methods rely on installing knock sensors on each cylinder of the engine. The detection accuracy depends heavily on the design of the engine block and reinforcing ribs, as well as the placement of the knock sensors. This places high demands on the shape, size, and layout of the engine components, increasing research and development costs and overall vehicle costs. Especially at low or high engine speeds, knock detection is often poor in some engines or cylinders, leading to false positives and false negatives.

[0078] Based on this, this application proposes a knock control method for a gas engine. By using a pre-trained knock recognition model for knock recognition, the accuracy and efficiency of knock recognition can be improved, and the cost can be reduced.

[0079] Reference Figure 1 , Figure 1 This is a first structural schematic diagram of a knock control system for a gas engine according to an embodiment of this application. Figure 1 As shown, the knock control system includes an acoustic emission sensor 1, a control device 2, and a cloud server 3. The acoustic emission sensor 1 is positioned at a target measurement point, which is the point most sensitive to acoustic emission sources from each cylinder. For example, in a gas engine with six cylinders, based on sensitivity analysis of acoustic emission signals, the target measurement point is determined to be the center position between the intake and exhaust ports on the cylinder head of the third cylinder, meaning this position is most sensitive to acoustic emission sources. Therefore, the acoustic emission sensor 1 is positioned at this center position on the cylinder head of the third cylinder to collect acoustic emission signals. The control device 2 is communicatively connected to the cloud server 3. The control device 2 includes a data acquisition module 21, a processing module 22, a transmission module 23, and a control module 24. The data acquisition module 21 is electrically connected to the acoustic emission sensor 1, thereby acquiring the acoustic emission signals collected by the sensor. Simultaneously, the data acquisition module 21 can also be used to acquire engine speed signals, engine torque signals, acoustic emission signals, and camshaft signals. The processing module 22 is connected to the acquisition module 21, enabling the processing module 22 to acquire the camshaft signal, engine speed signal, engine torque signal, and acoustic emission signal acquired by the acquisition module 21. It then processes the engine speed signal, engine torque signal, and acoustic emission signal based on the camshaft signal to remove abnormal data. The sending module 23 is connected to the processing module 22, enabling the sending module 23 to receive the first signal (including the engine speed signal, engine torque signal, and acoustic emission signal) processed by the processing module 22 and send the processed first signal to the cloud server 3. After receiving the processed first signal, the cloud server 3 can call a pre-trained knock recognition model to perform knock recognition on the processed first signal and obtain the recognition result. After obtaining the recognition result, the cloud server 3 can send the recognition result to the control module 24 of the control device 2, allowing the control module 23 to control various functional modules of the vehicle to perform corresponding processing based on the recognition result. Specifically, when the recognition result indicates knocking, the control module 23 can perform knock elimination processing according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation to ensure the normal operation of the engine.

[0080] Reference Figure 2 , Figure 2 This is a schematic diagram of the second structure of a knock control system for a gas engine according to an embodiment of this application. Figure 2As shown, the control device 2 also includes a heat-insulating coating 25 and a radiator 26. In this embodiment, the control device 2 is located beside the gas engine, that is, near the gas engine. Therefore, a heat-insulating coating 25 is required to isolate the engine's heat radiation. The heat-insulating coating 25 is located on the outermost layer of the control device 2 to prevent other modules (such as the acquisition module 21, processing module 22, transmission module 23, and control module 24) from being affected by the engine's heat radiation. The control device 2 is also equipped with a radiator 26 to dissipate the heat generated by the operation of each module (such as the acquisition module 21, processing module 22, transmission module 23, and control module 24), thereby ensuring that each module can operate normally.

[0081] Reference Figure 3 , Figure 3 This is a flowchart of a knock control method for a gas engine provided in an embodiment of this application, which is executed by the control device 2 described in the embodiment of this application, including but not limited to steps S310 to S340.

[0082] Step S310: Obtain a first signal, which includes an engine speed signal, an engine torque signal, and an acoustic emission signal. The acoustic emission signal is acquired by an acoustic emission sensor set at a target measuring point, which is the measuring point most sensitive to the acoustic emission source of each cylinder.

[0083] Step S320: Acquire camshaft signals and process the first signal based on the camshaft signals;

[0084] Step S330: The processed first signal is sent to the cloud server so that the cloud server can call the trained detonation recognition model to recognize the first signal and obtain the recognition result.

[0085] Step S340: When the identification result is that knocking has occurred, knocking elimination processing is performed according to the severity of the knocking and whether it is a single cylinder or multiple cylinders.

[0086] In this embodiment, the acquisition module 21 in the control device 2 acquires engine speed signals, engine torque signals, acoustic emission signals, and camshaft signals, and then sends them to the processing module 22. The processing module 22 processes the engine speed signals, engine torque signals, and acoustic emission signals based on the camshaft signals to remove abnormal data and improve the accuracy of subsequent knock identification. After processing, the processing module 22 sends the processed engine speed signals, engine torque signals, and acoustic emission signals to the cloud server 3 via the sending module 23. Upon receiving the processed engine speed signals, engine torque signals, and acoustic emission signals, the cloud server 3 can call a pre-trained knock identification model to identify the processed engine speed signals, engine torque signals, and acoustic emission signals to obtain the identification results. The engine speed signals and engine torque signals can be used to identify the engine load condition, while the acoustic emission signals are used to identify whether knock has occurred, the degree of knock, and whether, if knock occurs, it is single-cylinder knock or multi-cylinder knock. After the cloud server 3 calls the knock recognition model to obtain the recognition result, it can send the recognition result to the control module 24 in the control device 2. This allows the control module 24 to perform knock elimination processing according to the severity of the knock and whether it is a single-cylinder or multi-cylinder knock when the recognition result indicates that knock has occurred, so as to ensure the normal operation of the engine.

[0087] It should be noted that, in this embodiment of the application, considering the significant difference between the signals acquired in real-time and those in a laboratory environment, in order to extract useful signals, chi-square test, information gain ratio, and correlation coefficient test can also be used in practical applications to identify useful signals; then, the signal is simplified through a dimensionality reduction algorithm, and singular values ​​are identified and processed using DBSCAN. That is, in the actual knock detection process, it is also necessary to preprocess the acquired first signal camshaft signal to extract useful signals.

[0088] Reference Figure 4 , Figure 4 This is a flowchart of steps performed before acquiring the first signal, provided in an embodiment of this application, including but not limited to steps S410 to S430.

[0089] Step S410: Analyze the sensitivity of different measuring points to each acoustic emission source;

[0090] Step S420: Select the most sensitive measurement point as the target measurement point;

[0091] Step S430: Set an acoustic emission sensor at the target measuring point so that the acoustic emission sensor can collect the acoustic emission signals of each cylinder.

[0092] In this embodiment, considering that the acoustic emission wave excited by the acoustic emission source inside the engine cylinder undergoes complex reflection, refraction, mode conversion, and energy attenuation as it propagates through the structurally complex cylinder head to the outer surface, the acoustic emission signal acquired by the sensor at the cylinder head differs significantly from the original acoustic emission source signal, resulting in different recognition effects. Therefore, when practically applying acoustic emission technology to knock detection, it is necessary to first consider whether the acoustic emission source signal can be captured by the sensor or whether the signal-to-noise ratio of the captured signal is high. That is, it is necessary to first study the sensitivity of different measurement points (sensor installation positions) to different acoustic emission sources. The smaller the attenuation of the acoustic emission signal energy captured by the sensor relative to the acoustic emission source energy, the higher the sensitivity of the measurement point; the greater the attenuation, the lower the sensitivity of the measurement point.

[0093] To ensure that a single acoustic emission sensor 1 can collect acoustic emission signals from each cylinder of the engine to identify the combustion state of each cylinder, the installation location of the acoustic emission sensor 1 is crucial. Specifically, a target measuring point needs to be identified, which is most sensitive to the acoustic emission source corresponding to each cylinder. Therefore, only the acoustic emission signals collected by the acoustic emission sensor 1 installed at this target measuring point can accurately identify the combustion state of each cylinder. For this purpose, the embodiments of this application pre-analyze the sensitivity of measuring points on each cylinder head to the acoustic emission source of each cylinder, thereby selecting the most sensitive measuring point as the target measuring point. Then, by installing the acoustic emission sensor 1 at the target measuring point, it can be ensured that the acoustic emission sensor 1 can collect the acoustic emission signals of each cylinder, thus improving the accuracy of knock detection.

[0094] Reference Figure 5 , Figure 5 This is a flowchart of the steps for analyzing the sensitivity of different measurement points to various acoustic emission sources provided in an embodiment of this application, including but not limited to steps S510 to S550.

[0095] Step S510: Take the center of the bottom of each cylinder head as the excitation point and set the first acoustic emission sensor at each excitation point;

[0096] Step S520: Take the center position from the air inlet to the exhaust outlet on each cylinder head as the measuring point, and set a second acoustic emission sensor at each measuring point;

[0097] Step S530: Apply acoustic emission sources to each excitation point and acquire the first acoustic emission signals collected by each first acoustic emission sensor and the second acoustic emission signals collected by each second acoustic emission sensor.

[0098] Step S540: Calculate the energy attenuation of each measuring point relative to each acoustic emission source based on each first acoustic emission signal and each second acoustic emission signal.

[0099] Step S550: Determine the sensitivity of each measuring point to each acoustic emission source based on the energy attenuation of each measuring point relative to each acoustic emission source. The smaller the energy attenuation, the more sensitive the measuring point is to the acoustic emission source.

[0100] In this embodiment of the application, to determine the target measurement point, it is necessary to analyze the sensitivity of each different measurement point to each acoustic emission source. Based on the sensitivity of each different measurement point to each acoustic emission source, the target measurement point can be determined. Specifically, the center of the bottom of each cylinder head can be used as the excitation point, and a first acoustic emission sensor can be set at each excitation point; the center position from the intake port to the exhaust port on each cylinder head can be used as the measurement point, and a second acoustic emission sensor can be set at each measurement point. For example, refer to... Figure 6 , Figure 6 This is a schematic diagram showing the installation positions of the first and second acoustic emission sensors provided in this embodiment. The gas engine includes six cylinders, corresponding to six measuring points: P1, P2, P3, P4, P5, and P6. These six measuring points are the center positions from the intake port to the exhaust port on the cylinder head of each cylinder. There are also six excitation points: S1, S2, S3, S4, S5, and S6. These six excitation points are the center of the bottom of the cylinder head of each cylinder. A first acoustic emission sensor is installed at each excitation point, and a second acoustic emission sensor is installed at each measuring point. An acoustic emission source, such as a standard Hsu-Neilson acoustic emission source (broken pencil lead), is applied to each excitation point, simultaneously acquiring the signals collected by the first and second acoustic emission sensors. Since the first acoustic emission sensor is located at the excitation point, the signal collected by the first acoustic emission sensor can be used as the original acoustic emission wave signal. The signal collected by the second acoustic emission sensor is the signal from the original acoustic emission wave propagating to each measuring point. By selecting signals from two acoustic emission sensors within a certain time window, the energy attenuation of each measuring point relative to each acoustic emission source can be calculated using Equations 1 and 2. The sensitivity of each measuring point to the acoustic emission source of each cylinder can then be determined using these energy attenuation values. The final value of the energy attenuation of each measuring point relative to each acoustic emission source can be obtained by averaging five data acquisitions. Equations 1 and 2 are shown below:

[0101]

[0102] In Equation 1, E represents the energy of the acoustic emission wave, V represents the voltage value of the acoustic emission signal, and t0-t1 represents the time window.

[0103] kx = lnE0 - lnE(x) (Equation 2);

[0104] In Equation 2, k represents the attenuation factor, x represents the distance from the measuring point to the acoustic emission source (excitation point), E0 represents the energy of the acoustic emission wave at the excitation point, and E(x) represents the energy of the acoustic emission wave at the measuring point.

[0105] In this embodiment, Equation 2 can be used to represent the energy attenuation between the measuring point and the acoustic emission source. However, the attenuation characteristics of the medium need to be obtained experimentally. Since the attenuation factor of different measuring points relative to the same acoustic emission source differs for complex media such as cylinder heads, the magnitude of energy attenuation at different measuring points is related to both the attenuation factor and the distance. The product of the attenuation factor and the distance (kx) can be used as a parameter to measure the energy attenuation (measuring point sensitivity) between the measuring point and the acoustic emission source. Therefore, kx can characterize the amount of energy attenuation of the acoustic emission wave at the measuring point relative to the acoustic emission source. A smaller kx value indicates smaller energy attenuation and higher measuring point sensitivity; a larger kx value indicates larger energy attenuation and lower measuring point sensitivity. In this embodiment, the first and second acoustic emission sensors are used to collect the acoustic emission signals of each excitation point and each measuring point, respectively. Then, using Equations 1 and 2, the kx values ​​between each different measuring point and each acoustic emission source can be calculated.

[0106] Specifically, in this embodiment, two additional preamplifiers can be provided to further amplify the signals acquired by the sensor. The acoustic emission signals acquired by the sensor can be acquired by a data acquisition card. The data acquisition card can be a PCI type data acquisition card with four analog voltage input channels suitable for high-speed acquisition of signals such as radar, sonar, and ultrasound. In this embodiment, refer to... Figure 7 , Figure 7 This is a schematic diagram illustrating the kx values ​​between different measurement points and various acoustic emission sources provided in the embodiments of this application. For example... Figure 7 As shown, based on the energy attenuation (kx) of each measuring point relative to each acoustic emission source, the sensitivity of different measuring points to each acoustic emission source can be determined. The smaller the energy attenuation (kx), the more sensitive the measuring point is to the acoustic emission source.

[0107] Reference Figure 8 , Figure 8 This is a schematic diagram illustrating the sum of kx values ​​for all acoustic emission sources at various measurement points according to embodiments of this application. Embodiments of this application calculate the sum of energy attenuation at each measurement point relative to each acoustic emission source, thereby determining the sensitivity of each measurement point to each acoustic emission source based on the sum of energy attenuation. The smaller the sum of energy attenuation, the stronger the sensitivity of the measurement point to each acoustic emission source. For example... Figure 8 As shown, through calculation, it can be determined that the total energy attenuation of each acoustic emission source corresponding to measuring point P5 is the smallest, thus it can be determined that measuring point P5 is the most sensitive to each acoustic emission source, and therefore measuring point P5 can be determined as the target measuring point.

[0108] In this embodiment of the application, after the target measuring point is determined in advance, an acoustic emission sensor 1 can be installed at the target measuring point to detect the acoustic emission signals of each cylinder of the gas engine.

[0109] Reference Figure 9 , Figure 9 This is a flowchart of the steps for processing a first signal based on a camshaft signal, provided in an embodiment of this application, including but not limited to steps S910 to S930.

[0110] Step S910: Periodic identification is performed based on the camshaft signal;

[0111] Step S920: Based on the identified period, the first signal is processed by periodic division to obtain multiple periodic corresponding first signals;

[0112] Step S930: Perform periodic averaging on the first signal corresponding to each sub-cycle, so that each sub-cycle corresponds to an engine average speed signal, an engine average torque signal, and an average acoustic emission signal.

[0113] In this embodiment, after acquiring the camshaft signal, engine speed signal, engine torque signal, and acoustic emission signal collected by the acquisition module 21, the processing module 22 in the control device 2 can process the engine speed signal, engine torque signal, and acoustic emission signal based on the camshaft signal. Specifically, considering that the engine runs in a reciprocating cycle, it is necessary to divide the acquired signal into cycles to facilitate subsequent averaging and to identify abnormal points. Therefore, this embodiment also acquires the camshaft signal to perform cycle identification and cycle averaging based on the camshaft signal. Specifically, cycle identification is first performed based on the camshaft signal. The camshaft is designed with 60 teeth missing one tooth. The rising edge can be found through the difference algorithm, the rising edge time interval is calculated, and the missing tooth position is found. This moment can be used as the cycle signal. Based on the identified cycle, the engine speed signal, engine torque signal, and acoustic emission signal are processed in a cycle. That is, the engine speed signal is processed in a cycle to obtain multiple engine speed signals corresponding to multiple cycles, the engine torque signal is processed in a cycle to obtain multiple engine torque signals corresponding to multiple cycles, and the acoustic emission signal is processed in a cycle to obtain multiple acoustic emission signals corresponding to multiple cycles. The engine speed signal corresponding to each cycle is periodically averaged to ensure that each cycle corresponds to an average engine speed signal. Similarly, the engine torque signal corresponding to each cycle is periodically averaged to ensure that each cycle corresponds to an average engine torque signal. The acoustic emission signal corresponding to each cycle is also periodically averaged to ensure that each cycle corresponds to an average acoustic emission signal. Thus, by processing the first signal (including the engine speed signal, engine torque signal, and acoustic emission signal) and then sending it to the cloud server 3 via the sending module 23, the cloud server 3 can use a pre-trained knock recognition model to identify the processed first signal and obtain the recognition result.

[0114] Reference Figure 10 , Figure 10 This is a flowchart of the steps for training a detonation recognition model provided in the embodiments of this application, including but not limited to steps S1010 to S1040.

[0115] Step S1010: Collect the first signal and camshaft signal under different working conditions, and process the first signal based on the camshaft signal;

[0116] Step S1020: Construct a training set and a test set based on the processed first signal;

[0117] Step S1030: After training the detonation recognition model with the training set, obtain the test set to test the recognition accuracy of the detonation recognition model.

[0118] Step S1040: Control the training process of the detonation recognition model based on the test results.

[0119] In this embodiment, cloud server 3 needs to pre-build and train a detonation recognition model. This embodiment builds the detonation recognition model based on the Support Vector Machine (SVM) algorithm. SVM was originally applied to binary classification problems, and its principle is to find the optimal classification hyperplane. Assume the equation of the decision surface is:

[0120] w T x + b = 0; where the dimension of w is the same as the number of feature parameters, and the distance of each sample from the hyperplane is: In the formula: ||w|| is the magnitude of vector w; x is the coordinate of the sample point; w T b and b are both decision surface parameters.

[0121] For each hyperplane, it is necessary to calculate the distance with all samples, and select the minimum distance, which is the minimum distance d for each hyperplane. min Select all d min The hyperplane corresponding to the largest hyperplane in the dataset is the optimal hyperplane, and the corresponding sample data is the support vector. After determining the optimal hyperplane, the test samples are input, and the test samples are classified according to the regions defined by the optimal hyperplane. The hyperplane determined by the maximum hyperplane principle can achieve optimal classification.

[0122] In this embodiment of the application, the engine is subjected to different operating conditions during testing, and engine speed, engine torque, acoustic emission signals, and camshaft signals are collected under different operating conditions. Specifically, each test operating condition of the engine can be considered as a working state, and the processed first signal is extracted as a training set and a test set. The knock recognition model is then trained based on the training set and the test set.

[0123] Specifically, the system can simulate the engine's operating conditions under normal operation, severe knocking fault, and slight knocking fault, and collect and process the corresponding first signals. The processed first signals under different operating conditions are divided into training sets and test sets. The training set is then input into the knocking recognition model for training. After training for a period of time, the test set is input into the knocking recognition model for testing. When the recognition accuracy of the knocking recognition model reaches a preset value (such as 98%), the model is considered to be well trained, and the training can be terminated.

[0124] Referring to Table 1, which is a schematic diagram of engine state division, the engine consists of three cylinders: cylinder #1, cylinder #2, and cylinder #3. The test conditions may include the original engine state (corresponding to the normal engine operation state) and different ignition advance angles for different cylinders (which may result in minor knocking faults and severe knocking faults).

[0125] Table 1. Engine Status Classification Diagram

[0126]

[0127] Correspondingly, the division of the training set and the test set is shown in Table 2, which is a schematic diagram of the division of the training set and the test set. As shown in Table 2, the training set may include two groups, A1 and A2, and the corresponding test set also includes two groups, A1 and A2. Among them, A1 includes the first signal collected under normal engine operation and under slight knock failure. A2 includes the first signal collected under normal engine operation and under severe knock failure.

[0128] Table 2. Schematic diagram of training and test set division

[0129]

[0130] In this embodiment, acoustic emission signal feature values ​​corresponding to each cylinder under different loads can be extracted under data tag F1. These acoustic emission signal feature values ​​can be values ​​collected from multiple experiments. Additionally, acoustic emission signal feature values ​​corresponding to each cylinder (including cylinders 1#, 2#, and 3#) under different loads can be extracted under data tag F2, and acoustic emission signal feature values ​​corresponding to each cylinder (including cylinders 1#, 2#, and 3#) under different loads can be extracted under data tag F3. Model training and testing are then performed based on the extracted acoustic emission signal feature values.

[0131] It's important to note that parameter tuning is crucial during training. SVM uses the Python function `GridSearchCV` to optimize parameters using a grid search method. The main principle is to perform a traversal search of the algorithm's key parameters using a grid format, allowing for a custom search step size. The accuracy of cross-validation on the training set is compared for different parameter combinations, and the combination with the highest accuracy is used in the algorithm model. Once the algorithm parameters are determined, different sample sets with varying groupings can be used to train different algorithms, and the recognition accuracy can be tested using the corresponding test sets.

[0132] Reference Figure 11 , Figure 11 This application provides a flowchart of the first step of knock elimination processing when the identification result is knocking, based on the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation, including but not limited to steps S1110 to S1130.

[0133] Step S1110: When the identification result is severe knocking under high load, control the throttle to reduce the load in order to eliminate knocking;

[0134] Step S1120: When the identification result is high load slight knocking, but it is single cylinder knocking, control to reduce the ignition advance angle of the cylinder where knocking occurred in order to eliminate knocking;

[0135] Step S1130: When the identification result is high load slight knocking, but it is multi-cylinder knocking, control to reduce the excess air coefficient to eliminate knocking.

[0136] In this embodiment, the cloud server 3 calls a pre-trained knock recognition model to identify the processed first signal. After obtaining the recognition result, it can send the result to the control module 24 of the vehicle-side control device 2. This allows the control module 24 to eliminate knock by reducing the load when the recognition result indicates severe knocking under high load. When the recognition result indicates slight knocking under high load, but it is single-cylinder knocking, the control module 24 can eliminate knocking by reducing the ignition advance angle of the cylinder experiencing knocking. When the recognition result indicates slight knocking under high load, but it is multi-cylinder knocking, the control module 24 can eliminate knocking by reducing the excess air coefficient.

[0137] Reference Figure 12 , Figure 12 This application provides a flowchart of the second step of knock elimination processing when the identification result is knocking, based on the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation, including but not limited to steps S1210 to S1240.

[0138] Step S1210: When the identification result is severe knocking under medium and low load, but it is multi-cylinder knocking, control the throttle to reduce the load in order to eliminate knocking;

[0139] Step S1220: When the identification result is severe knocking under medium and low load, but it is single-cylinder knocking, control the reduction of the ignition advance angle of the cylinder where knocking occurred to eliminate knocking;

[0140] Step S1230: When the identification result is mild knocking under medium and low load, but it is single-cylinder knocking, control the reduction of the ignition advance angle of the cylinder where knocking occurred to eliminate knocking;

[0141] Step S1240: When the identification result is mild knocking under medium and low load, but it is multi-cylinder knocking, control the reduction of the excess air coefficient to eliminate knocking.

[0142] In this embodiment, the cloud server 3 calls a pre-trained knock recognition model to identify the processed first signal. After obtaining the recognition result, it can send the result to the control module 24 of the vehicle-side control device 2. This allows the control module 24 to eliminate knock by reducing the throttle when the recognition result indicates severe knocking under low to medium load, but involving multiple cylinders. Similarly, when the recognition result indicates severe knocking under low to medium load, but involving only one cylinder, the control module 24 can eliminate knocking by reducing the ignition advance angle of the cylinder experiencing knocking. Furthermore, when the recognition result indicates slight knocking under low to medium load, but involving only one cylinder, the control module 24 can eliminate knocking by reducing the ignition advance angle of the cylinder experiencing knocking. Finally, when the recognition result indicates slight knocking under low to medium load, but involving multiple cylinders, the control module 24 can eliminate knocking by reducing the excess air coefficient.

[0143] It should be noted that when the identification result is slight knocking under high load but in a single cylinder, or slight knocking under medium-low load but in a single cylinder, the knocking is eliminated by controlling the reduction of the ignition advance angle of the cylinder experiencing knocking, but the reduction amounts differ. The reduction amount for slight knocking under high load is greater than that for slight knocking under medium-low load. Similarly, when the identification result is severe knocking under medium-low load but in a single cylinder, or slight knocking under medium-low load but in a single cylinder, the knocking is eliminated by controlling the reduction of the ignition advance angle of the cylinder experiencing knocking, but the reduction amounts differ again. The reduction amount for severe knocking under medium-low load is greater than that for slight knocking under medium-low load.

[0144] Reference Figure 13 , Figure 13 This is a schematic diagram of the knock control processing logic provided in an embodiment of this application. For example... Figure 13 As shown, after receiving the identification result sent by the cloud server 3, the control module 24 can perform knock elimination processing according to the severity of the knock and whether it is a single cylinder or multiple cylinders when the identification result indicates that knock has occurred.

[0145] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0146] The units described as separate components may or may not be physically separate. Similarly, the components of the control device may or may not be physical units; they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0147] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing computer program instructions, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0148] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for controlling knock in a gas engine, characterized in that, The method includes: Acquire a first signal, which includes an engine speed signal, an engine torque signal, and an acoustic emission signal. The acoustic emission signal is acquired by an acoustic emission sensor located at a target measuring point, and the target measuring point is the measuring point most sensitive to the acoustic emission source of each cylinder. The camshaft signal is acquired, and the period is identified based on the camshaft signal. Based on the identified period, the first signal is divided into multiple periods to obtain multiple first signals corresponding to each period. The first signals corresponding to each period are then averaged to ensure that each period corresponds to an engine average speed signal, an engine average torque signal, and an average acoustic emission signal. The processed first signal is sent to the cloud server so that the cloud server can call the trained detonation recognition model to recognize the first signal and obtain the recognition result. When the identification result indicates that knocking has occurred, knocking elimination processing is performed according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation.

2. The method according to claim 1, characterized in that, Before acquiring the first signal, the method includes: Analyze the sensitivity of different measuring points to various acoustic emission sources; Select the measurement point with the highest sensitivity as the target measurement point; An acoustic emission sensor is installed at the target measuring point so that the acoustic emission sensor can collect the acoustic emission signals of each cylinder.

3. The method according to claim 2, characterized in that, The analysis of the sensitivity of different measuring points to various acoustic emission sources includes: The center of the bottom of each cylinder head is taken as the excitation point, and the first acoustic emission sensor is set at each excitation point; The center position from the air intake port to the exhaust port on each cylinder head is taken as the measuring point, and a second acoustic emission sensor is set at each measuring point; An acoustic emission source is applied at each of the excitation points, and the first acoustic emission signal collected by each of the first acoustic emission sensors and the second acoustic emission signal collected by each of the second acoustic emission sensors are acquired. Based on each of the first acoustic emission signals and each of the second acoustic emission signals, the energy attenuation of each measuring point relative to each acoustic emission source is calculated. Based on the energy attenuation of each measuring point relative to each acoustic emission source, the sensitivity of different measuring points to each acoustic emission source is determined. The smaller the energy attenuation, the more sensitive the measuring point is to the acoustic emission source.

4. The method according to claim 3, characterized in that, The process of determining the sensitivity of different measuring points to different acoustic emission sources based on the energy attenuation of each measuring point relative to each acoustic emission source includes: Calculate the total energy attenuation of each measuring point relative to each acoustic emission source; The sensitivity of each measuring point to each acoustic emission source is determined based on the total energy attenuation. The smaller the total energy attenuation, the stronger the sensitivity of the measuring point to each acoustic emission source.

5. The method according to claim 1, characterized in that, When the identification result indicates that knocking has occurred, knock elimination processing is performed according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation, including: When the identification result is severe knocking under high load, the throttle is reduced to decrease the load and eliminate knocking. When the identification result is high load slight knocking, but it is single cylinder knocking, the ignition advance angle of the cylinder where knocking occurred is reduced to eliminate knocking; When the identification result is high load slight knocking, but it is multi-cylinder knocking, the excess air coefficient is reduced to eliminate knocking.

6. The method according to claim 1 or 5, characterized in that, When the identification result indicates that knocking has occurred, knock elimination processing is performed according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation, including: When the identification result is severe knocking under medium to low load, but it is multi-cylinder knocking, control the throttle to reduce the load in order to eliminate knocking; When the identification result is severe knocking under medium and low loads, but it is single-cylinder knocking, the ignition advance angle of the cylinder where knocking occurred is reduced to eliminate knocking; When the identification result is mild knocking under medium to low load, but it is single-cylinder knocking, the ignition advance angle of the cylinder where knocking occurred is reduced to eliminate knocking; When the identification result is mild knocking under medium to low load, but it is multi-cylinder knocking, the excess air coefficient is reduced to eliminate knocking.

7. The method according to claim 1, characterized in that, The method further includes training the detonation recognition model, including: The first signal and camshaft signal under different working conditions are collected, and the first signal is processed based on the camshaft signal; Training and test sets are constructed based on the processed first signal; After training the detonation recognition model using the training set, test the recognition accuracy of the detonation recognition model using the test set. The training process of the detonation recognition model is controlled based on the test results.

8. A knock control system for a gas engine, characterized in that, include: An acoustic emission sensor is set at a target measuring point, which is the measuring point most sensitive to the acoustic emission source of each cylinder. The control device includes a data acquisition module, a processing module, a transmission module, and a control module; The cloud server is communicatively connected to the control device. The acquisition module is connected to the acoustic emission sensor, and the acquisition module is used to acquire the first signal and the camshaft signal; The processing module is connected to the acquisition module. The processing module is used to identify the period based on the camshaft signal; based on the identified period, the first signal is processed by period division to obtain multiple first signals corresponding to the period division; the first signals corresponding to each period division are processed by period averaging so that each period division corresponds to an engine average speed signal, an engine average torque signal and an average acoustic emission signal. The sending module is connected to the processing module, and the sending module is used to send the processed first signal to the cloud server; The cloud server is used to call the trained detonation recognition model to identify the first signal and obtain the recognition result; The control module is used to acquire the identification result, and when the identification result indicates that knocking has occurred, it performs knock elimination processing according to the severity of the knocking and whether it is a single-cylinder or multi-cylinder situation.

9. The detonation control system according to claim 8, characterized in that, The control device is located next to the gas engine, and the control device also includes a heat-insulating coating and a radiator.

Images