Gas compressor unsteady flow control method based on casing wall surface micro-jet

Abstract

The invention provides a gas compressor unsteady flow control method based on casing wall surface micro-jet, and relates to the technical field of unsteady flow control. Micro-jet is introduced into a gas compressor through the wall surface of a casing, unsteady flow is adjusted through closed-loop feedback, and the flow rate of the gas compressor is improved. The disturbance which is generated by the unsteady micro-jet in the flow field and has the same frequency and the opposite phase with the original flow field fluctuation is obtained, then the unsteady flow is counteracted, and the unsteady flow can be efficiently inhibited.

Description

Technical Field

[0001] The present invention provides a method for controlling unsteady flow in a compressor based on micro-jet on the casing wall, which relates to the technical field of unsteady flow control, specifically to the technical field of unsteady flow control in a compressor with micro-jet. Background Art

[0002] The internal flow of a compressor has significant unsteady characteristics, and typical phenomena include rotating stall, unsteady tip leakage flow, and dynamic evolution of flow separation, etc. Existing unsteady flow control technologies are mainly divided into two categories: passive control and active control. Both traditional passive control technologies and active control technologies have disadvantages such as low control accuracy, weak performance, and instability, and it is difficult to accurately and efficiently suppress unsteady flow. Summary of the Invention

[0003] The present invention provides a method for controlling unsteady flow in a compressor based on micro-jet on the casing wall to solve the above problems: A method for controlling unsteady flow in a compressor based on micro-jet on the casing wall proposed by the present invention, the method includes: Introduce micro-jet into the compressor through the wall of the casing, and adjust the unsteady flow through closed-loop feedback to obtain perturbations generated by the unsteady micro-jet in the flow field that are in-phase and opposite in frequency to the fluctuations of the original flow field, thereby canceling the unsteady flow; Obtain the unsteady flow region, arrange n micro-jet orifices at the casing in the unsteady flow region, install a dynamic pressure sensor at a preset position corresponding to each micro-jet orifice, and determine the parameters of the micro-jet orifices through 5 variables, then the total number of variables of the parameters of the n micro-jet orifices is 5n; The 5 variables include the flow rate m, the derivative of the flow rate with respect to time dm / dt, the angle between the jet direction and the normal direction of the S1 flow surface , the angle between the projection of the jet direction on the S1 flow surface and the axial direction , and the compressor operating flow coefficient .

[0004] Further, the introducing micro-jet into the compressor through the wall of the casing, and adjusting the unsteady flow through closed-loop feedback to obtain perturbations generated by the unsteady micro-jet in the flow field that are in-phase and opposite in frequency to the fluctuations of the original flow field, thereby canceling the unsteady flow, includes: S1. Obtain the coordinate and number data of the given micro-jet orifices; Judge the opening state and closing state of the micro-jet orifices to obtain the working state judgment information of the micro-jet orifices; S2. Conduct non-working analysis and working analysis according to the working state judgment information of the micro-jet orifices to obtain non-working analysis data and working analysis data; S3. Construct an unsteady multimodal microjet perturbation based on the non - working analysis data and the working analysis data, and perform perturbation cancellation; S4. Judge whether the microjets can sufficiently suppress the rotating instability waves, and complete the establishment of flow control or reconstruct the dynamic mode decomposition according to the judgment information.

[0005] Further, the S1 includes: Obtain the switch states of each microjet orifice; When the switch state information of all microjet orifices is in the open state, determine that the overall microjet orifices are in the working state; When the switch state information of all microjet orifices is in the closed state, determine that the overall microjet orifices are in the non - working state; When the switch state information of all microjet orifices has both open and closed states, determine that the overall microjet orifices are in the working state, and the switch state of the overall microjet orifices is the working state judgment information of the microjet orifices.

[0006] Further, the S2 includes: Obtain the working state judgment information of the microjet orifices; When the working state judgment information of the microjet orifices is that the microjets are not working, obtain the pressure time series, perform dynamic mode decomposition, and obtain the reconstructed signal; When the working state judgment information of the microjet orifices is that the microjets are working, perform online learning through machine learning, and then establish the mapping relationship between the microjet parameters and the generated perturbations.

[0007] Further, the step of, when the working state judgment information of the microjet orifices is that the microjets are not working, obtaining the pressure time series, performing dynamic mode decomposition, and obtaining the reconstructed signal, includes: Construct a mathematical equation for the unsteadiness of the flow field; When the microjets are not working, analyze the structure of the solid - wall casing, and do not perform the correlation analysis between the perturbations measured by the pressure sensors and the jets; Perform dynamic mode decomposition on the time - series snapshots p(t) composed of the signals p1 to pn of n pressure sensors fixed at different positions on the wall of the casing, perform band - pass filtering on the dynamic mode decomposition spectrum, and select the first r modes for reconstruction to obtain the reconstructed signal; The calculation formula of the reconstructed signal is: .

[0008] Further, the step of, when the working state judgment information of the microjet orifices is that the microjets are working, performing online learning through machine learning, and then establishing the mapping relationship between the microjet parameters and the generated perturbations, includes: Establish the mapping relationship between the micro-jet parameters and the generated perturbation, where the perturbation is determined by two variables, the pressure p monitored by the dynamic pressure sensor and its time derivative dp / dt; The mapping relationship is obtained by adjusting the micro-jet parameters, establishing samples for machine learning, and performing online learning using a neural network.

[0009] Further, the S3 includes: Construct an unsteady micro-jet perturbation through the reconstructed signal and the mapping relationship, so that the pressure sensor receives a signal with the same frequency and opposite phase to the reconstructed signal; The calculation formula for the signal with the same frequency and opposite phase is: .

[0010] Further, the S4 includes: Judge whether the micro-jet sufficiently suppresses the rotating instability wave, and introduce an error function e;

[0011] Wherein,

[0012] Judge whether the error function e meets the preset target, obtain the target judgment result, and perform adjustment operations according to the target judgment result.

[0013] Further, the judgment of whether the error function e meets the preset target, obtaining the target judgment result, and performing adjustment operations according to the target judgment result includes: When the error function e meets the preset target, the establishment of flow control is completed; When the error function e does not meet the preset target, the selected number of modes r is increased to correct the dynamic mode analysis.

[0014] Further, the control system includes: A micro-jet orifice state analysis module for obtaining the coordinate and number data of the given micro-jet orifice; Judge the open state and closed state of the micro-jet orifice to obtain the working state judgment information of the micro-jet orifice; A classification analysis module for performing non-working analysis and working analysis according to the working state judgment information of the micro-jet orifice to obtain non-working analysis data and working analysis data; A perturbation cancellation module for constructing an unsteady multi-modal micro-jet perturbation according to the non-working analysis data and the working analysis data to perform perturbation cancellation; A completion judgment module for judging whether the micro-jet sufficiently suppresses the rotating instability wave, and completing the establishment of flow control or reconstructing the dynamic mode decomposition according to the judgment information.

[0015] Advantages of the present invention: The present invention proposes an active control method for unsteady flow of a compressor based on micro-jet on the casing wall and closed-loop feedback. Compared with the prior art, it has the following significant advantages: (1) Precise matching of dynamic response and unsteady flow Closed-loop feedback control: The pressure fluctuation of the flow field is monitored in real time through a dynamic pressure sensor, and the dominant frequency and phase characteristics of the unsteady flow are extracted by combining dynamic mode decomposition (DMD) to ensure that the micro-jet perturbation is in-phase opposition with the original flow fluctuation, so as to actively cancel the unsteady perturbation (such as rotating stall and surge precursor oscillation).

[0016] Adaptive frequency-domain matching: Through DMD spectral band-pass filtering and reconstruction, key unsteady modes (such as low-frequency rotating stall and high-frequency tip vortex breakdown) are automatically screened, avoiding control failure caused by frequency-domain mismatch in traditional technologies (such as synthetic jets).

[0017] (2) Multi-parameter collaborative optimization and efficient energy utilization Design of jet parameters with multiple degrees of freedom: The collaborative control of the control variables of each micro-jet orifice breaks through the limitations of single parameters (such as steady jet) in the prior art.

[0018] Modeling of mapping relationship driven by machine learning: A neural network is used to establish a non-linear mapping relationship between jet parameters and perturbation effects, and the jet strategy is dynamically optimized to avoid energy waste or insufficient control caused by fixed jet parameters in traditional methods.

[0019] (3) High-precision and low-intrusiveness design Non-contact dynamic sensing: Wall-mounted embedded pressure sensors are used to avoid problems such as insufficient spatial resolution and blockage effect in traditional contact measurements (such as LDV). At the same time, the small size design of the micro-jet orifice reduces the additional interference to the flow field.

[0020] Lightweight and low-cost implementation: No complex mechanical structures (such as variable guide vanes) are required. Only by integrating micro-jets on the casing wall and sensors, the system complexity and maintenance cost are reduced, and it is applicable to extreme environments such as high temperature and high pressure. Description of the drawings

[0021] Figure 1 It is a schematic diagram of a method for controlling unsteady flow of a compressor based on micro-jet on the casing wall; Figure 2 It is a schematic diagram of the main idea; Figure 3 It is a schematic diagram of the control process of micro-jet on the casing wall based on online learning. Detailed implementation manners

[0022] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0023] In one embodiment of the present invention, a compressor unsteady flow control method based on micro-jet injection through the casing wall is proposed. The method includes: Introducing micro-jet into the compressor through the casing wall, regulating the unsteady flow through closed-loop feedback, obtaining perturbations generated by the unsteady micro-jet in the flow field that are in-phase and opposite in frequency to the original flow field fluctuations, and then canceling the unsteady flow. Obtain the unsteady flow region. At the casing of the unsteady flow region, arrange n micro-jet orifices, install a dynamic pressure sensor at a preset position corresponding to each micro-jet orifice, and determine the parameters of the micro-jet orifices through 5 variables. Then the total number of variables of the parameters of the n micro-jet orifices is 5n. The corresponding preset position includes a preset range centered on the micro-jet orifice, which does not affect the installation of other micro-jet orifices and dynamic pressure sensors, as Figure 2 shown; The 5 variables include the flow rate m, the derivative of the flow rate with respect to time dm / dt, the angle between the jet direction and the normal of the S1 stream surface , the angle between the projection of the jet direction on the S1 stream surface and the axial direction , and the compressor operating flow coefficient .

[0024] The working principle of the above technical solution is as follows: The existing unsteady flow control technologies are mainly divided into two categories: passive control and active control.

[0025] (1) Passive control technology Casing treatment structures such as slot-type and circumferential grooves suppress the development of leakage vortices by changing the tip clearance flow structure; The serrated design of the blade leading edge delays separation by promoting boundary layer mixing.

[0026] (2) Active control technology The synthetic jet technology injects momentum into the flow field using a high-frequency oscillating jet (frequency up to the kHz level) to interfere with the vortex structure in the separation zone; The plasma actuator generates an induced air flow by ionizing the gas and changes the local flow direction, which is suitable for flow reattachment control in high-temperature environments; The variable guide vane adjustment changes the air flow angle of attack by mechanically adjusting the blade angle to meet the flow requirements under different working conditions.

[0027] In the present invention, microjets are introduced into the compressor through the wall of the casing, and the unsteady flow is regulated through closed-loop feedback to obtain the perturbation generated by the unsteady microjets in the flow field, which is in the same frequency but opposite phase to the fluctuation of the original flow field, so as to cancel the unsteady flow and improve the defects of the prior art.

[0028] The defects of the prior art mainly include: The passive control technology includes casing treatment structures (such as slotted, circumferential grooves, etc.) that suppress the development of leakage vortices by changing the tip clearance flow. However, the structural modification may cause significant loss of aerodynamic efficiency, such as increasing flow resistance or inducing local secondary flow. In addition, the optimization effect of such structures is usually limited to specific operating conditions, and it may even exacerbate flow separation or induce new instability phenomena when deviating from the design point. In engineering applications, the wear problem under high-speed rotation and high-temperature environment may reduce the reliability of the casing structure, and even lead to crack propagation due to stress concentration, threatening the overall life of the engine.

[0029] The serrated design of the blade leading edge delays flow separation by promoting boundary layer mixing. However, the serrated geometry increases the surface friction loss of the blade, especially in high-speed flows, which may reduce the stage efficiency. The complex serrated edge faces accuracy challenges during machining and requires high-cost special processes (such as five-axis milling or 3D printing). Moreover, the mechanical strength of the blade leading edge may be difficult to withstand high-frequency vibration loads due to geometric weakening, and there is a risk of fatigue fracture during long-term operation.

[0030] The active control technology includes synthetic jet technology that relies on high-frequency oscillating jets to interfere with the vortex structure in the separation zone. However, its core defect lies in the high dependence on external energy supply systems and complex control logics. The actuator needs to continuously input energy to maintain the jet, significantly increasing the system energy consumption. At the same time, the precise matching of the kHz-level jet frequency with the flow field dynamics poses strict requirements on sensors and controllers, and any phase deviation or response delay may exacerbate flow instability. In addition, piezoelectric materials or electromagnetic drive components are prone to thermal failure under high-temperature and high-pressure environments, resulting in soaring maintenance costs.

[0031] The plasma actuator generates induced airflows by ionizing gases to change the local flow direction. However, its high-voltage ionization process may cause electromagnetic interference to on-board electronic devices, and the rapid ablation of electrode materials at high temperatures shortens the service life. Although this technology is applicable to high-temperature environments, the ionization energy consumption is high and the efficiency is low, making it difficult to achieve global flow control in large-scale compressors and can only be used as a local auxiliary means.

[0032] Variable guide vane adjustment adapts to different operating conditions by mechanically adjusting the blade angle, but its mechanical drive system (such as connecting rods, servo mechanisms) significantly increases the weight and complexity of the compression components, which runs counter to the lightweight requirements of aeroengines. Frequent adjustment of the guide vane angle may accelerate the wear of bearings and transmission components, and the thermal expansion of metals at high temperatures will also lead to a decrease in adjustment accuracy. The problem of dynamic response delay makes it even more difficult to cope with the rapid switching of transient operating conditions.

[0033] The technical effects of the above technical solution are as follows: Compared with the prior art, this method has the following significant advantages: (1) Precise matching of dynamic response and unsteady flow Closed-loop feedback control: The pressure fluctuations in the flow field are monitored in real time through dynamic pressure sensors, and the dominant frequencies and phase characteristics of unsteady flow are extracted by combining dynamic mode decomposition (DMD) to ensure that the microjet perturbation is in-phase and opposite in frequency to the original flow fluctuation, realizing the active cancellation of unsteady perturbations (such as rotating stall, precursor oscillation of surge).

[0034] Adaptive frequency-domain matching: Through DMD spectral band-pass filtering and reconstruction, key unsteady modes (such as low-frequency rotating stall and high-frequency tip vortex breakdown) are automatically selected to avoid control failure caused by frequency-domain mismatch in traditional technologies (such as synthetic jets).

[0035] (2) Multi-parameter collaborative optimization and efficient energy utilization Design of multi-degree-of-freedom jet parameters: The collaborative control of control variables for each microjet orifice breaks through the limitations of single parameters (such as steady jets) in existing technologies.

[0036] Machine learning-driven mapping relationship modeling: A neural network is used to establish a non-linear mapping relationship between jet parameters and perturbation effects, dynamically optimizing the jet strategy to avoid energy waste or insufficient control caused by fixed jet parameters in traditional methods.

[0037] (3) High-precision and low-intrusiveness design Non-contact dynamic sensing: Wall-embedded pressure sensors are used to avoid problems such as insufficient spatial resolution and blockage effects in traditional contact measurements (such as LDV), and at the same time, the small size design of the microjet orifice reduces the additional interference to the flow field.

[0038] Lightweight and low-cost implementation: Without complex mechanical structures (such as variable guide vanes), only by integrating microjets and sensors on the casing wall, the system complexity and maintenance cost are reduced, and it is applicable to extreme environments such as high temperature and high pressure.

[0039] In an embodiment of the present invention, microjets are introduced into the compressor through the wall of the casing, and the unsteady flow is regulated through closed-loop feedback to obtain perturbations generated by the unsteady microjets in the flow field that are in-phase and opposite in frequency to the original flow field fluctuations, thereby canceling the unsteady flow, including: S1. Obtain the coordinate and number data of the given micro jet orifices; Judge the opening state and closing state of the micro jet orifices to obtain the working state judgment information of the micro jet orifices; S2. Conduct non - working analysis and working analysis according to the working state judgment information of the micro jet orifices to obtain non - working analysis data and working analysis data; S3. Construct an unsteady multi - modal micro jet perturbation based on the non - working analysis data and working analysis data to conduct perturbation cancellation; S4. Judge whether the micro jets can sufficiently suppress the rotating instability waves, and complete the establishment of flow control or the reconstruction of dynamic mode decomposition according to the judgment information, as Figure 1 shown.

[0040] The working principle of the above technical solution is: The main idea of the above method is as Figure 2 shown. Micro jets are introduced on the casing wall. For unsteady flows, through closed - loop feedback regulation, the unsteady micro jets generate perturbations in the flow field that are in the same frequency and opposite phase as the fluctuations of the original flow field, thereby canceling the unsteady flow.

[0041] As Figure 2 shown, at the casing near the unsteady flow region, n micro jet orifices are arranged, and a dynamic pressure sensor is installed near each micro jet orifice. For the parameters of each micro jet orifice, it is planned to be determined by a total of 5 variables including the flow rate m, the derivative of the flow rate with respect to time dm / dt, the angle between the jet direction and the normal of the S1 flow surface , the angle between the projection of the jet direction on the S1 flow surface and the axial direction , and the compressor working flow coefficient . Then the total number of variables of the parameters of n micro jet orifices is 5n.

[0042] Step 1: Establish a mathematical equation describing the unsteadiness of the flow field. When the micro jets are not working, the research object is the solid - wall casing structure. At this time, the perturbations measured by the pressure sensors have nothing to do with the jets. Conduct dynamic mode decomposition on the time - series snapshots p(t) composed of the signals p1, p2,..., pn of n pressure sensors fixed at different positions on the wall, and perform band - pass filtering on the dynamic mode decomposition (Dynamic mode decomposition, DMD) spectrum, and select the first r modes for reconstruction to obtain the reconstructed signal .

[0043] Step 2: Establish the mapping relationship between the micro-jet parameters and the generated perturbation, where the perturbation is determined by two variables, the pressure p monitored by the pressure sensor and its time derivative dp / dt. By varying the jet parameters, samples for machine learning are established, and online learning is carried out using a neural network (such as the Radial basis function neural network, RBFNN) to establish the mapping relationship.

[0044] Step 3: For the reconstructed in Step 1, according to the mapping relationship established in Step 2, construct an unsteady micro-jet perturbation so that the pressure sensor receives a signal with the same frequency and opposite phase : Step 4: Determine whether the micro-jet sufficiently suppresses the rotating instability wave. Introduce an error function e: where

[0045] Judge whether the error e meets the established goal. If it does not meet the goal, return to Step 1 and increase the selected number of modes r. In addition, the influence of the coordinates and number of micro-jet orifices on rotating instability is to be studied. Through multiple rounds of iteration, an active control scheme for unsteady flow suitable for the current research object and working conditions is finally determined, as shown in Figure 3 Figure.

[0046] The technical effects of the above technical solutions are as follows: The pressure fluctuations in the flow field are monitored in real time by a dynamic pressure sensor, and the dominant frequency and phase characteristics of the unsteady flow are extracted by combining the Dynamic Mode Decomposition (DMD) to ensure that the micro-jet perturbation is in the same frequency and opposite phase as the original flow fluctuation, so as to actively cancel the unsteady perturbation (such as rotating stall and surge precursor oscillation).

[0047] By DMD spectral band-pass filtering and reconstruction, the key unsteady modes (such as low-frequency rotating stall and high-frequency tip vortex breakdown) are automatically screened, avoiding control failure caused by frequency domain mismatch in traditional technologies (such as synthetic jets).

[0048] The coordinated control of the control variables of each micro-jet orifice breaks through the limitations of single parameters (such as steady jets) in existing technologies.

[0049] Machine learning-driven mapping relationship modeling: A neural network is used to establish a non-linear mapping relationship between jet parameters and perturbation effects, dynamically optimizing the jet strategy to avoid energy waste or insufficient control caused by fixed jet parameters in traditional methods.

[0050] Wall-embedded pressure sensors are used to avoid the problems of insufficient spatial resolution and blockage effects in traditional contact measurements (such as LDV), and at the same time, the small size design of the micro-jet orifices reduces the additional interference to the flow field.

[0051] Without a complex mechanical structure (such as variable guide vanes), by integrating micro jetting on the casing wall surface with sensors, the system complexity and maintenance cost are reduced, and it is applicable to extreme environments such as high temperature and high pressure.

[0052] In one embodiment of the present invention, the S1 includes: Obtain the switch state of each micro jetting port; When the switch state information of all micro jetting ports is in the open state, it is determined that the overall micro jetting ports are in the working state; When the switch state information of all micro jetting ports is in the closed state, it is determined that the overall micro jetting ports are in the non - working state; When the switch state information of all micro jetting ports has both open and closed states, it is determined that the overall micro jetting ports are in the working state, and the switch state of the overall micro jetting ports is the working state judgment information of the micro jetting ports.

[0053] The working principle of the above - mentioned technical solution is: Obtain the switch state information of each micro jetting port. The switch state information is detected and obtained through sensors or control signals, etc., and can reflect the current state of each micro jetting port in real - time and accurately.

[0054] According to the obtained switch state information, logical judgment is carried out: If the switch states of all micro jetting ports are in the open state, it is determined that the overall micro jetting ports are in the working state. If the switch states of all micro jetting ports are in the closed state, it is determined that the overall micro jetting ports are in the non - working state. If the switch states of the micro jetting ports have both open and closed states, it is also determined that the overall micro jetting ports are in the working state. The efficient judgment of the overall micro jetting ports is realized. The judgment result is output as the working state judgment information of the micro jetting ports.

[0055] The technical effect of the above - mentioned technical solution is: By monitoring the switch state of each micro jetting port in real - time, the state changes of the micro jetting ports can be discovered and responded to in a timely manner, thereby improving the reliability and stability of the entire system.

[0056] The judgment logic is clear and concise, easy to implement and program. Just make a simple logical judgment according to the obtained switch state information to obtain the working state of the overall micro jetting ports, reducing the complexity of the control logic.

[0057] It allows different switch state combinations between micro jetting ports. As long as the judgment conditions are met, it can be considered that the overall micro jetting ports are in the working state.

[0058] In one embodiment of the present invention, the S2 includes: Obtain the working state judgment information of the micro jetting ports; When the working state judgment information of the micro jet orifice indicates that the micro jet is not working, obtain the pressure time series, perform dynamic mode decomposition, and obtain the reconstructed signal. When the working state judgment information of the micro jet orifice indicates that the micro jet is working, perform online learning through machine learning, and then establish the mapping relationship between the micro jet parameters and the generated disturbances.

[0059] The working principle of the above technical solution is as follows: obtain the working state judgment information of the micro jet orifice by real-time monitoring or receiving signals from sensors. Specifically, based on parameters such as the switch state, flow rate, and pressure of the micro jet orifice, comprehensively determine whether the micro jet orifice is currently in a working state.

[0060] When the system determines that the working state of the micro jet orifice is "not working", further obtain the pressure time series data and perform the corresponding analysis for non-working; perform dynamic mode decomposition (Dynamic Mode Decomposition, DMD) on the obtained pressure time series. DMD is a data-driven analysis method that can extract the dynamic characteristics of the system from time series data, such as modes, frequencies, and growth rates.

[0061] Through DMD analysis, a reconstructed signal can be obtained.

[0062] When the system determines that the working state of the micro jet orifice is "working", a machine learning algorithm will be started for online learning. During the learning process, parameter data of the micro jet orifice (such as jet volume, jet frequency, jet angle, etc.) and corresponding disturbance data (such as airflow disturbance, pressure change, etc.) will be collected. These data are used to train the machine learning model so that it can learn the mapping relationship between the micro jet parameters and the generated disturbances.

[0063] Through online learning, continuously update and optimize the mapping relationship model.

[0064] The technical effect of the above technical solution is as follows: by obtaining the working state judgment information of the micro jet orifice, the working state of the micro jet orifice is monitored in real time.

[0065] When the micro jet orifice is not working, a reconstructed signal is obtained through dynamic mode decomposition, which can deeply understand the dynamic characteristics of the system.

[0066] When the micro jet orifice is working, by establishing the mapping relationship between the micro jet parameters and the generated disturbances through machine learning, the disturbances generated by the micro jet orifice can be predicted and controlled more accurately. By adjusting the working parameters of the micro jet orifice, improve the mapping accuracy.

[0067] The online learning mechanism enables the system to continuously adapt to various data, automatically update and optimize the model parameters. Improve the accuracy of the mapping relationship.

[0068] In one embodiment of the present invention, when the working state determination information of the micro-jet orifice indicates that the micro-jet is not working, a pressure time series is acquired and dynamic mode decomposition is performed to obtain a reconstructed signal, including: Construct a mathematical equation for the unsteadiness of the flow field; When the micro-jet is not working, analyze the structure of the solid wall casing and do not analyze the correlation between the perturbations measured by the pressure sensors and the jetting; Perform dynamic mode decomposition on the time series snapshot p(t) composed of the signals p1 to pn of n pressure sensors fixed at different positions on the wall of the casing, perform band-pass filtering on the dynamic mode decomposition spectrum, and select the first r modes for reconstruction to obtain a reconstructed signal; The calculation formula of the reconstructed signal is: .

[0069] The working principle of the above technical solution is as follows: Before processing the pressure time series when the micro-jet orifice is not working, it is first necessary to construct a mathematical equation describing the unsteadiness of the flow field. When the micro-jet orifice is not working, analyze the structure of the solid wall casing. Obtain the inherent dynamic characteristics of the casing structure in the absence of jetting. At this time, do not analyze the correlation between the perturbations measured by the pressure sensors and the jetting, because the jet orifice is in a non-working state, so the jetting has no influence on the perturbations of the pressure sensors.

[0070] Perform dynamic mode decomposition (DMD) on the time series snapshot p(t) composed of the signals p1 to pn of n pressure sensors fixed at different positions on the wall of the casing. DMD is a data-driven analysis method that can extract the dynamic modes of the system from time series data, including the frequencies, growth rates, and mode shapes of the modes.

[0071] Perform band-pass filtering on the DMD spectrum to remove noise and unwanted frequency components and retain the frequency components related to the dynamic characteristics of the flow field.

[0072] Select the first r modes for reconstruction to obtain a reconstructed signal.

[0073] The technical effects of the above technical solution are as follows: By constructing a mathematical equation for the unsteadiness of the flow field and performing dynamic mode decomposition, it is possible to deeply understand the dynamic characteristics of the flow field when the micro-jet orifice is not working. The natural frequencies, mode shapes, and growth rates of the flow field can be accurately identified.

[0074] The band-pass filtering and mode selection steps can remove noise and unwanted frequency components and retain the signals related to the dynamic characteristics of the flow field. Improve the accuracy of signal processing, so that the reconstructed signal can better approximate the original pressure time series.

[0075] Through the reconstructed signal, the reconstruction and prediction of the flow field state can be realized.

[0076] The analysis results of the reconstructed signal can identify abnormal behaviors or fault modes in the flow field. Meanwhile, based on the analysis results of the reconstructed signal, the casing structure or operating parameters can be adjusted to optimize the flow field performance.

[0077] In one embodiment of the present invention, when the working state judgment information of the micro jet orifice is that the micro jet is working, online learning is carried out through machine learning, and then a mapping relationship between the micro jet parameters and the generated disturbance is established, including: Establish a mapping relationship between the micro jet parameters and the generated disturbance, where the disturbance is determined by two variables, the pressure p monitored by the dynamic pressure sensor and its time derivative dp / dt; The mapping relationship is obtained by adjusting the micro jet parameters to establish samples for machine learning and using a neural network for online learning.

[0078] The working principle of the above technical solution is as follows: The disturbance is determined by two variables, the pressure p monitored by the dynamic pressure sensor and its time derivative dp / dt. The pressure p reflects the current pressure state of the flow field, while the time derivative dp / dt reflects the rate of change of the pressure with time, that is, the dynamic characteristics of the pressure. These two variables together constitute the feature vector describing the disturbance.

[0079] In order to establish a mapping relationship between the micro jet parameters and the generated disturbance, it is necessary to adjust the micro jet parameters. These parameters may include the jet volume, jet frequency, jet angle, etc. By changing these parameters, the changes in the flow field disturbance under different parameter settings can be observed.

[0080] During the process of adjusting the micro jet parameters, the dynamic pressure sensor is used to record the pressure p and its time derivative dp / dt, forming a series of sample data. Each sample data contains a set of micro jet parameter settings and the corresponding disturbance feature vector.

[0081] Machine learning is carried out using a neural network, and the above sample data is used as the training set. The input layer of the neural network receives the micro jet parameters, and the output layer predicts the disturbance feature vector (i.e., the pressure p and the time derivative dp / dt).

[0082] By training the neural network, it learns the complex mapping relationship between the micro jet parameters and the disturbance. During the training process, the neural network continuously adjusts its internal weights to minimize the prediction error.

[0083] The online learning mechanism allows the neural network to continuously receive new sample data during the actual operation process and update its internal weights in real time to adapt to the changes in the flow field conditions and the adjustment of the micro jet parameters.

[0084] The technical effects of the above technical solution are as follows: By establishing the mapping relationship between the microjet parameters and the generated perturbations, the perturbation situation of the flow field under different microjet parameter settings can be predicted more accurately.

[0085] The online learning mechanism enables the neural network to adapt to the changes in the flow field conditions and the adjustment of the microjet parameters in real time. It enhances the adaptive ability of the system, enabling it to maintain stable performance under different working conditions.

[0086] By predicting the perturbation situation, the setting of the microjet parameters can be optimized to minimize the adverse effects on the flow field or maximize the desired perturbation effect, which can improve the overall performance and efficiency of the system.

[0087] In one embodiment of the present invention, S3 includes: By reconstructing the signal and the mapping relationship, an unsteady microjet perturbation is constructed, so that the pressure sensor receives the same-frequency anti-phase signal of the reconstructed signal; The calculation formula of the same-frequency anti-phase signal is: .

[0088] The working principle of the above technical solution is as follows: The reconstructed signal is extracted and reconstructed from the pressure time series through methods such as dynamic mode decomposition (DMD), reflecting the dynamic characteristics of the flow field under specific conditions.

[0089] The mapping relationship is the association between the microjet parameters and the generated perturbations established through machine learning (such as neural networks), and is used to predict the perturbation situation under different microjet parameter settings.

[0090] Based on the reconstructed signal and the mapping relationship, an unsteady microjet perturbation can be constructed.

[0091] In order to make the pressure sensor receive the same-frequency anti-phase signal of the reconstructed signal, the reconstructed signal needs to be subjected to phase inversion processing.

[0092] The generated same-frequency anti-phase signal is used in the control of the microjet system as a feedback signal to adjust the microjet parameters to achieve precise control of the flow field perturbation.

[0093] The technical effects of the above technical solution are as follows: By constructing an unsteady microjet perturbation and generating a same-frequency anti-phase signal, precise control of the flow field perturbation can be achieved. It can reduce the undesired perturbations and improve the stability and performance of the system.

[0094] By actively generating and applying the same-frequency anti-phase signal, active suppression or enhancement of the flow field perturbation can be achieved.

[0095] In experiments or simulations, using the reconstructed signal and the same-frequency anti-phase signal can evaluate the effects of different control strategies faster, reducing the trial-and-error cost and time.

[0096] In one embodiment of the present invention, S4 includes: Judge whether the microjet can sufficiently suppress the rotating instability wave, and introduce an error function e;

[0097] Wherein,

[0098] Judge whether the error function e meets the preset target, obtain the target judgment result, and perform adjustment operations according to the target judgment result.

[0099] The step of judging whether the error function e meets the preset target, obtaining the target judgment result, and performing adjustment operations according to the target judgment result includes: When the error function e meets the preset target, the establishment of flow control is completed; When the error function e does not meet the preset target, increase the selected modal number r and correct the dynamic modal analysis.

[0100] The working principle of the above technical solution is as follows: In the process of judging whether the microjet can sufficiently suppress the rotating instability wave, an error function e is introduced as a quantization index. The error function e is used to measure the difference between the actual flow state and the desired flow state.

[0101] Compare the error function e with the preset target. The preset target is a threshold or range set in advance, which is used to judge whether the flow control reaches the desired effect. If the error function e meets the preset target, it is considered that the microjet has sufficiently suppressed the rotating instability wave.

[0102] Obtain the target judgment result according to the comparison result between the error function e and the preset target.

[0103] When the target judgment result is "yes", it means that the flow control has been successfully established, and at this time, the establishment process of the flow control can be completed.

[0104] When the target judgment result is "no", it means that the flow control does not reach the desired effect. At this time, it is necessary to increase the selected modal number r and correct the dynamic mode decomposition (DMD). Increasing the modal number r means including more dynamic characteristics when reconstructing the signal, so as to possibly more accurately reflect the actual state of the flow field, and then more effectively suppress the rotating instability wave.

[0105] The technical effect of the above technical solution is as follows: By introducing the error function e and performing quantization judgment, the effect of flow control can be more accurately evaluated, so as to timely adjust the control strategy and improve the accuracy of flow control.

[0106] When the error function e does not meet the preset target, the system can automatically increase the number of modes r and correct the DMD to adapt to the changes in the flow field conditions and the complexity of the rotating instability waves. This enhances the adaptive ability of the system and enables it to maintain stable performance under different working conditions.

[0107] By continuously adjusting the number of modes r and correcting the DMD, the control strategy can be optimized to more effectively suppress the rotating instability waves. This helps to reduce energy losses, improve system efficiency, and may extend the service life of the equipment.

[0108] Through steps such as error function judgment, target judgment result acquisition, and adjustment operations, a closed-loop control system is formed, reducing manual intervention and improving control efficiency.

[0109] In one embodiment of the present invention, the control system includes: A micro jet orifice state analysis module for obtaining the coordinate and number data of the given micro jet orifices; Judging the opening state and closing state of the micro jet orifices to obtain the working state judgment information of the micro jet orifices; A classification analysis module for performing non-working analysis and working analysis according to the working state judgment information of the micro jet orifices to obtain non-working analysis data and working analysis data; A perturbation cancellation module for constructing an unsteady multi-modal micro jet perturbation according to the non-working analysis data and the working analysis data to perform perturbation cancellation; A completion judgment module for judging whether the micro jets can sufficiently suppress the rotating instability waves and completing the establishment of flow control or reconstructing the dynamic mode decomposition according to the judgment information.

[0110] The working principle of the above technical solution is as follows: Micro jets are introduced on the casing wall. For unsteady flow, through closed-loop feedback regulation, the unsteady micro jets generate perturbations in the flow field that are in the same frequency and opposite phase as the fluctuations of the original flow field, thereby canceling the unsteady flow.

[0111] At the casing near the unsteady flow region, n micro jet orifices are arranged, and a dynamic pressure sensor is installed near each micro jet orifice. For the parameters of each micro jet orifice, it is planned to be determined by a total of 5 variables including the flow rate m, the derivative of the flow rate with respect to time dm / dt, the angle between the jet direction and the normal of the S1 flow surface 、the projection of the jet direction on the S1 flow surface and the axial angle 、the compressor working flow coefficient . Then the total number of variables of the parameters of the n micro jet orifices is 5n.

[0112] The technical effects of the above technical solution are as follows: By using a dynamic pressure sensor to monitor the pressure fluctuations in the flow field in real time, and combining with dynamic mode decomposition (DMD) to extract the dominant frequencies and phase characteristics of unsteady flows, it is ensured that the microjet perturbations are in-phase opposition with the original flow fluctuations, achieving the active cancellation of unsteady perturbations (such as rotating stall and surge precursor oscillations).

[0113] Through DMD spectral band-pass filtering and reconstruction, key unsteady modes (such as low-frequency rotating stall and high-frequency tip vortex breakdown) are automatically screened, avoiding control failures caused by frequency-domain mismatches in traditional technologies (such as synthetic jets).

[0114] The coordinated control of the control variables for each microjet orifice breaks through the limitations of single parameters (such as steady jets) in existing technologies.

[0115] Machine learning-driven mapping relationship modeling: A neural network is used to establish a non-linear mapping relationship between jet parameters and perturbation effects, dynamically optimizing the jetting strategy and avoiding energy waste or insufficient control caused by fixed jet parameters in traditional methods.

[0116] Wall-embedded pressure sensors are used to avoid problems such as insufficient spatial resolution and blockage effects in traditional contact measurements (such as LDV). At the same time, the small size design of the microjet orifices reduces the additional interference to the flow field.

[0117] There is no need for complex mechanical structures (such as variable guide vanes). By integrating only the microjets and sensors on the casing wall, the system complexity and maintenance costs are reduced, and it is applicable to extreme environments such as high temperature and high pressure.

[0118] Obviously, those skilled in the art can make various modifications and deformations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and deformations of the present invention fall within the scope of the claims of the present invention and its equivalent technologies, the present invention is also intended to include these modifications and deformations.

Claims

1. A compressor unsteady flow control method based on micro-jet on the casing wall surface, characterized in that The method includes: Introducing microjets into the compressor through the wall of the casing, regulating the unsteady flow through closed-loop feedback, obtaining the perturbation generated by the unsteady microjets in the flow field that is in-phase and opposite in frequency to the original flow field fluctuations, and then canceling the unsteady flow; Obtaining the unsteady flow region, arranging n microjet orifices at the casing of the unsteady flow region, installing a dynamic pressure sensor at a preset position corresponding to each microjet orifice, and determining the parameters of the microjet orifices through 5 variables, so the total number of variables of the parameters of the n microjet orifices is 5n; The five variables include the flow rate m, the derivative of the flow rate with respect to time dm / dt, and the angle between the jet direction and the normal of the S1 flow surface , and the angle between the projection of the jet direction on the S1 flow surface and the axial direction , and the compressor operating flow coefficient .

2. The compressor unsteady flow control method based on micro air jet on the casing wall surface according to claim 1, characterized in that The introducing microjets into the compressor through the wall of the casing, regulating the unsteady flow through closed-loop feedback, obtaining the perturbation generated by the unsteady microjets in the flow field that is in-phase and opposite in frequency to the original flow field fluctuations, and then canceling the unsteady flow includes: S1. Obtaining the coordinate and number data of the given microjet orifices; Judging the opening state and closing state of the microjet orifices to obtain the working state judgment information of the microjet orifices; S2. Conducting non-working analysis and working analysis according to the working state judgment information of the microjet orifices to obtain non-working analysis data and working analysis data; S3. Constructing unsteady multimodal microjet perturbations according to the non-working analysis data and working analysis data to conduct perturbation cancellation; S4. Judging whether the microjets can sufficiently suppress the rotating instability wave, and completing the establishment of flow control or reconstructing the dynamic mode decomposition according to the judgment information.

3. The compressor unsteady flow control method based on micro jet on the casing wall surface according to claim 2, wherein The S1 includes: Obtaining the switch state of each microjet orifice; When the switch state information of all microjet orifices is in the open state, determining that the overall microjet orifices are in the working state; When the switch state information of all microjet orifices is in the closed state, determining that the overall microjet orifices are in the non-working state; When the switch state information of all microjet orifices has both open states and closed states, determining that the overall microjet orifices are in the working state, and the switch state of the overall microjet orifices is the working state judgment information of the microjet orifices.

4. The compressor unsteady flow control method based on micro jet on the casing wall surface according to claim 2, characterized in that, The S2 includes: Obtaining the working state judgment information of the microjet orifices; When the working state judgment information of the microjet orifices is that the microjets are not working, obtaining the pressure time series, conducting dynamic mode decomposition, and obtaining the reconstructed signal; When the working state judgment information of the microjet orifices is that the microjets are working, conducting online learning through machine learning, and then establishing the mapping relationship between the microjet parameters and the generated perturbations.

5. The compressor unsteady flow control method based on micro air jets on the casing wall surface according to claim 4, characterized in that, The when the working state judgment information of the microjet orifices is that the microjets are not working, obtaining the pressure time series, conducting dynamic mode decomposition, and obtaining the reconstructed signal includes: Constructing the mathematical equation of the unsteadiness of the flow field; When the microjets are not working, analyzing the structure of the solid wall casing, and not conducting the correlation analysis between the perturbation measured by the pressure sensor and the jet; Conducting dynamic mode decomposition on the time series snapshots p(t) composed of the signals p1 to pn of n pressure sensors fixed at different positions on the wall of the casing, performing band-pass filtering on the dynamic mode decomposition spectrum, and selecting the first r modes for reconstruction to obtain the reconstructed signal; The calculation formula of the reconstructed signal is: 。 6. The compressor unsteady flow control method based on micro jet on the casing wall surface according to claim 4, characterized in that, When the working state judgment information of the micro-jet orifice is that the micro-jet is working, online learning is carried out through machine learning, and then a mapping relationship between micro-jet parameters and generated disturbances is established, including: Establish a mapping relationship between micro-jet parameters and generated disturbances, where the disturbances are determined by two variables, the pressure p monitored by the dynamic pressure sensor and its time derivative dp / dt; The mapping relationship is obtained by adjusting the micro-jet parameters to establish samples for machine learning and using a neural network for online learning.

7. The compressor unsteady flow control method based on micro jet on the casing wall surface according to claim 2, characterized in that, The S3 includes: Construct unsteady micro-jet disturbances through the reconstructed signal and the mapping relationship, so that the pressure sensor receives a signal with the same frequency and opposite phase as the reconstructed signal; The calculation formula for the signal with the same frequency and opposite phase is: 。 8. The compressor unsteady flow control method based on micro jet on the casing wall surface according to claim 2, characterized in that The S4 includes: Judge whether the micro-jet sufficiently suppresses the rotating instability wave, and introduce an error function e; Wherein, Judge whether the error function e meets the preset target, obtain a target judgment result, and perform an adjustment operation according to the target judgment result.

9. The compressor unsteady flow control method based on micro jet on the casing wall surface according to claim 8, characterized in that, The judgment of whether the error function e meets the preset target, obtaining a target judgment result, and performing an adjustment operation according to the target judgment result includes: When the error function e meets the preset target, the establishment of flow control is completed; When the error function e does not meet the preset target, the selected number of modes r is increased to correct the dynamic mode analysis.

10. A control system for implementing the compressor unsteady flow control method based on micro jet on the casing wall surface according to claim 2, characterized in that, The control system includes: A micro-jet orifice state analysis module for obtaining coordinate and number data of a given micro-jet orifice; Judge the open state and closed state of the micro-jet orifice to obtain the working state judgment information of the micro-jet orifice; A classification analysis module for performing non-working analysis and working analysis according to the working state judgment information of the micro-jet orifice to obtain non-working analysis data and working analysis data; A disturbance cancellation module for constructing unsteady multi-modal micro-jet disturbances according to the non-working analysis data and the working analysis data to perform disturbance cancellation; A completion judgment module for judging whether the micro-jet sufficiently suppresses the rotating instability wave, and completing the establishment of flow control or reconstructing the dynamic mode decomposition according to the judgment information.

Images