CN-121977846-A - Multi-Mach number adaptive separator flow field online reconstruction device
Abstract
The invention relates to the technical field of hypersonic flow field reconstruction and control, and discloses a multi-Mach number adaptive isolated segment flow field online reconstruction device which comprises a test segment, wherein a spray pipe is arranged at the upstream of the test segment, the air inlet end of the spray pipe is connected with a stabilizing segment, the front end of the stabilizing segment is connected with an air storage tank through a pneumatic control valve, and the downstream of the test segment is connected with an exhaust pipeline; the inner wall surface of the test section is embedded with a dynamic pressure sensor, and the outlet of the test section is provided with a back pressure sensor and a flap driven by a servo motor. According to the invention, pressure data is processed through working condition sensing weighting and dynamic sliding window filtering, a flow field image is reconstructed by utilizing a parallel double-branch neural network, and the topological form of a shock wave string is identified, so that a high-level or pulse width modulation signal is output to perform differential control on the electromagnetic valve, high-fidelity flow field reconstruction of sparse pressure data under multiple working conditions is realized, and the problem of signal hysteresis is solved.
Inventors
- KONG CHEN
- CHANG JUNTAO
- WANG ZIAO
- LV CHENGKUN
Assignees
- 哈尔滨工业大学
Dates
- Publication Date
- 20260505
- Application Date
- 20260126
Claims (10)
- 1. The multi-Mach number adaptive separator flow field online reconstruction device comprises a test section (5) and is characterized in that the upstream of the test section (5) is sequentially communicated with a spray pipe (4), a stabilizing section (3) and an air storage tank (1) connected through a pneumatic control valve (2), the downstream of the test section (5) is connected with an exhaust pipeline (8), a total pressure sensor (19) is arranged on the stabilizing section (3), static pressure sensors (7) are embedded on the wall surface of the spray pipe (4), dynamic pressure sensor arrays (20) are symmetrically arranged on the inner wall surface of the test section (5), and a back pressure sensor (11) and a flap (9) driven by a servo motor (10) are arranged at the outlet of the test section (5); The test section is characterized in that the total pressure sensor (19), the static pressure sensor (7), the dynamic pressure sensor array (20) and the back pressure sensor (11) are electrically connected with the synchronous data acquisition card (12), the synchronous data acquisition card (12) is in communication connection with the edge computing platform (14), the wall surface of the test section (5) is provided with a boundary layer suction groove (15) and is connected with a vacuum suction source (18) through a pipeline, an electromagnetic valve (16) is arranged on the pipeline, and the electromagnetic valve (16) is connected with the edge computing platform (14) through a relay (17).
- 2. A multiple mach number adapted separator segment flow field online reconstruction apparatus as defined in claim 1, wherein said edge computation platform (14) comprises: The self-adaptive signal preprocessing module is used for carrying out characteristic weighting and dynamic filtering on the pressure data uploaded by the synchronous data acquisition card (12); the flow field reconstruction reasoning module is used for mapping the processed pressure data into a flow field gray level image; the intelligent state sensing module is used for extracting shock wave string characteristics from the flow field gray level image; and the differentiation control decision module is used for generating a control instruction according to the characteristics of the shock wave strings and sending the control instruction to the relay (17).
- 3. The multi-mach number adapted separator flow field online reconstruction device of claim 2, wherein the adaptive signal preprocessing module comprises: the working condition sensing unit is used for calculating the ratio of the pressure value of the static pressure sensor (7) positioned at the spray pipe (4) to the total pressure value of incoming flow acquired by the total pressure sensor (19); And the attention weighting unit is used for generating attention weight vectors for all nodes in the dynamic pressure sensor array (20) according to the ratio and carrying out element-by-element weighting operation on the original pressure data by utilizing the attention weight vectors.
- 4. The multiple mach number adapted separator segment flow field online reconstruction device of claim 2, wherein the adaptive signal preprocessing module further comprises: The dynamic sliding window filter is used for receiving the shock wave string moving speed value fed back by the intelligent state sensing module; when the absolute value of the shock wave string moving speed value is larger than a set speed threshold value, automatically reducing the length of a filtering window; And when the absolute value of the shock wave string moving speed value is smaller than or equal to a set speed threshold value, automatically increasing the length of the filtering window.
- 5. The multi-mach number adaptive separator flow field online reconstruction device according to claim 2, wherein the flow field reconstruction reasoning module adopts a fused neural network architecture of encoding and decoding, and the device comprises: the encoder comprises a one-dimensional convolution feature extraction branch and a full-connection feature extraction branch which are arranged in parallel, wherein the one-dimensional convolution feature extraction branch is configured to directly carry out convolution operation on a one-dimensional pressure data sequence to extract local spatial gradient features along the flow direction; And the decoder is configured to splice and fuse the feature vectors output by the two branches, and then map and up-sample the fusion features step by step through a multi-layer transposition convolution operation to generate a two-dimensional flow field reconstruction image.
- 6. The multi-mach number adapted separator flow field online reconstruction device of claim 2, wherein the intelligent state sensing module comprises: The target detection unit is used for executing a target detection algorithm on the flow field reconstruction image and outputting flow direction position coordinates of the front edge of the shock wave string and the topological form category of the shock wave string, wherein the topological form category at least comprises a normal shock wave string category and an oblique shock wave string category.
- 7. The multi-mach number adapted separator flow field online reconstruction device of claim 4, wherein said intelligent state awareness module further comprises: And the speed resolving unit is used for calculating the difference value of the flow direction position coordinates of the front edge of the shock wave string detected at the current moment and the last moment, dividing the difference value by the time interval to obtain the moving speed value of the shock wave string, and feeding back the speed value to the dynamic sliding window filter in real time.
- 8. The device for online reconstruction of a multi-mach number adapted separator segment flow field of claim 6, wherein the differential control decision module is configured to: when the position of the front edge of the shock wave string is detected to exceed a set safety threshold value and the topology type is the forward shock wave string type, outputting a continuous high-level signal to control the electromagnetic valve (16) to be in a full-open state; When the position of the front edge of the shock wave string is detected to exceed a set safety threshold value and the topology type is the oblique shock wave string type, a pulse width modulation signal is output to control the electromagnetic valve (16) to be in an intermittent opening state.
- 9. The multiple mach number adapted separator segment flow field online reconstruction device according to claim 1, wherein the solenoid valve (16) is a high frequency response solenoid valve and the vacuum suction source (18) is a vacuum pump.
- 10. The multi-Mach number adaptive separator flow field online reconstruction device according to claim 1, wherein fused quartz observation windows (6) are installed on two side wall surfaces of the test section (5), and the fused quartz observation windows (6) are opposite to a high-speed camera (13) and are used for collecting real flow field images in the test section (5).
Description
Multi-Mach number adaptive separator flow field online reconstruction device Technical Field The invention relates to the technical field of hypersonic flow field reconstruction and control, in particular to an on-line reconstruction device for a multi-Mach number adaptive isolation section flow field. Background The scramjet engine is used as a core power device for realizing hypersonic flight, and the isolation section of the scramjet engine bears the key function of matching aerodynamic parameters of an air inlet channel and a combustion chamber. The shock wave string structure formed inside the isolation section can balance the back pressure generated by the combustion chamber through wave system interaction. However, in the event of a motorized flight of the aircraft or a severe change in the operating conditions of the combustion chamber, an excessively strong counter-pressure gradient may cause the shock wave train to move upstream. If the front edge of the shock wave string overflows the throat of the air inlet channel, the phenomenon that the air inlet channel is not started is caused, namely, the supersonic air flow channel is congested, so that the air inflow suddenly drops, the total pressure recovery coefficient is deteriorated, the thrust of the engine is instantaneously lost, and the flameout of the combustion chamber or the structural damage of the aircraft can be caused when the shock wave string is serious. Therefore, accurate sensing and real-time regulation and control of the flow field structure in the isolation section have decisive effects on maintaining the starting state of the engine and guaranteeing the flight safety. The existing flow field observation technology is mainly divided into non-contact optical measurement and contact discrete measurement. Although optical means such as schlieren imaging and Particle Image Velocimetry (PIV) can provide visual two-dimensional flow field images, an optical path system is complex, equipment is huge and sensitive to vibration environments, and the optical means are difficult to deploy in a narrow space inside an aircraft or run on line for a long time in engineering application. The contact measurement based on the wall dynamic pressure sensor has the advantages of simple structure, high response speed and easy integration, but the sensor can only acquire one-dimensional discrete pressure distribution of the flow field boundary, and cannot directly represent the complex wave system structure and the time-space evolution rule thereof. To solve the above problems, related studies have attempted to establish a mapping relationship between wall pressure and flow field structure using an interpolation algorithm or a basic neural network. However, existing approaches have limitations in facing wide-speed-range flight conditions. In the climbing or maneuvering process of the hypersonic aircraft, the incoming flow Mach number and the boundary condition are in dynamic change, and the reconstruction model trained based on a single working condition is insufficient in generalization capability under the working condition of non-design points. Meanwhile, the existing pressure signal processing mainly adopts a filtering algorithm with fixed parameters, and the dynamic characteristics of the flow field are not considered. When the shock wave string is in a high-frequency oscillation or fast moving state, the fixed window filtering easily causes the key transient characteristics to be smoothly lost or phase lag to be generated, so that the system cannot capture the precursor characteristics of the starting in time. In terms of flow control, existing boundary layer suction systems typically employ simple on-off control logic based on pressure thresholds. The control mode lacks of fine identification of the topological form of the shock wave string, and cannot distinguish different influences of the normal shock wave string and the oblique shock wave string on the stability margin of the flow field. When the limit working condition is not reached, excessive suction can accelerate the consumption of a vacuum source or an injection air source with limited load, so that the cruising capability of the system is reduced, and when the forward shock wave string moves forward due to the strong reverse pressure gradient, the occurrence of the non-starting of the air inlet channel is difficult to effectively inhibit if the targeted strong intervention is lacking. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a multi-Mach number adaptive on-line flow field reconstruction device for an isolation section, which solves the problems that an internal flow field structure is difficult to reconstruct accurately due to sparse pressure data processing hysteresis under a variable working condition of a hypersonic isolation section, and a traditional single control strategy cannot consider the stability margin of the flow field and the e