Plasma synthetic jet dual-mode modulation system and method

Abstract

The application provides a plasma synthetic jet dual-mode regulation system and method, and relates to the technical field of active flow control and ice prevention of aerospace vehicles. The system comprises a plasma synthetic jet exciter array, a dual-mode high-voltage pulse regulation controller, a multi-target optimization collaborative regulation unit and a sensor sensing layer. The exciter array is arranged in the span direction by dual-mode compatible exciters and is configured to generate shock waves for ice removal under nanosecond pulse driving and generate pulsed jets for flow control under microsecond pulse driving. The controller integrates nanosecond pulse generation modules and microsecond pulse generation modules and realizes switching through a mode switching switch circuit. The regulation unit selects the working mode and determines the excitation parameters according to the real-time working condition information fed back by the ice layer parameter sensing module and the boundary layer flow parameter sensing module of the sensor sensing layer. The application can be compatible with both ice removal and flow control functions on the same hardware platform.

Description

Technical Field

[0001] This application relates to the field of active flow control and de-icing technology for aerospace vehicles, and in particular to a dual-mode control system and method for plasma synthetic jet. Background Technology

[0002] A typical mission for a medium-to-high altitude long-endurance UAV (climb through clouds → high-altitude cruise → return) usually lasts more than 20 hours, with a large flight profile. During the climb phase, the UAV traverses cold water droplet clouds, posing a risk of icing to the wing leading edge and air intake lip. During the high-altitude cruise phase, the UAV is affected by low-density atmosphere and low Reynolds number effects, requiring a large angle of attack to maintain lift, and boundary layer separation is likely to occur. These two types of aerodynamic threats may occur sequentially or concurrently in the same mission, and since the UAV platform does not have a pilot to make human judgments and operate, the aerodynamic protection system must have autonomous perception, decision-making, and execution capabilities.

[0003] Plasma synthetic jet (PSJ) actuators instantaneously heat the gas within a sealed cavity through high-voltage pulsed discharge (typically 2-25kV), generating a high-speed pulsed jet with a peak velocity exceeding 500m / s. They feature fast response, wide bandwidth, compact structure (approximately 12g per unit), and no moving mechanical parts, allowing for direct conformal embedding into the skin and providing excellent adaptability to unmanned aerial platforms with strict weight and space constraints. PSJ technology has made progress in both de-icing and flow control.

[0004] In the field of PSJ de-icing, patent CN115230970B discloses a wing de-icing device based on PSJ, which uses an actuator to push a push rod to break the ice layer through mechanical impact. By adjusting the actuator spacing and discharge energy, complete de-icing or ice shape control can be achieved. A dielectric barrier discharge (DBD) actuator can also be added to enhance the de-icing effect. Patent application CN106314800A discloses a de-icing method based on a three-electrode plasma impact jet, which utilizes an ignition electrode to reduce breakdown voltage and improve energy conversion efficiency, achieving a jet velocity exceeding 500 m / s. The above work verifies the feasibility of using PSJ for aircraft de-icing. Their common feature is that the actuator structure and discharge parameters are designed and optimized around the single function of de-icing.

[0005] In the field of PSJ flow control, patent CN120417205B discloses an intelligent plasma flow control device for incident shock wave boundary layer interference. This device arranges wall-mounted hot-wire sensors upstream and downstream of the PSJ actuator, and the controller dynamically adjusts the high-voltage pulse power supply output based on the sensor signals, forming an intelligent closed-loop control system. Patent CN113955069B discloses a PSJ-based high-speed aircraft drag reduction closed-loop control method. This method establishes a control capability database of flight parameters and actuator parameters, and dynamically adjusts the actuator parameters based on real-time flight parameters. These works establish a closed-loop control framework for PSJ in flow control, and their technical solutions target single-mode control for specific flow phenomena.

[0006] In the area of ​​plasma dual-function integration, patent application CN114340123A proposes a flow control and anti-icing dual-function plasma generator based on the surface dielectric barrier discharge (DBD) technology platform. These schemes verify the feasibility of integrating de-icing and flow control functions onto the same plasma excitation device, both based on DBD plasma excitation. The induced airflow velocity of a DBD exciter is generally below 10 m / s, and the de-icing mechanism mainly relies on the plasma Joule heating effect, which is fundamentally different from the technical characteristics of PSJ, which relies on shock wave stress wave to peel off the ice layer and has a jet velocity exceeding 500 m / s.

[0007] In summary, PSJ actuators offer significant advantages over DBD technology in terms of jet velocity and impact strength. However, existing PSJ solutions are all single-function designs. De-icing schemes are optimized based on the impact wave characteristics of nanosecond pulses, while flow control schemes are designed based on the momentum injection characteristics of microsecond pulses. No solution has yet systematically addressed the compatibility issue between nanosecond de-icing and microsecond flow control modes on the same PSJ hardware platform. The essence of this problem lies in the different preferences of the two modes for discharge cavity volume and jet outlet structure. A dual-objective compatibility design method is needed to determine the geometric parameter window that simultaneously satisfies the performance requirements of both modes. Summary of the Invention

[0008] This application provides a dual-mode control system and method for plasma synthetic jets. Addressing the real-world scenario where icing and flow separation occur sequentially during the same flight mission, this system establishes a quantitative design method for exciter dual-mode compatibility, a nanosecond-microsecond dual-pulse power supply collaborative switching mechanism, and an autonomous decision-making closed-loop architecture based on dual-source sensor fusion. This achieves an integrated aerodynamic protection solution encompassing hardware, dual-mode collaboration, and autonomous decision-making, providing a technical approach that balances de-icing and flow control for unmanned aerial platforms with strict weight and space constraints and requiring autonomous operation.

[0009] In a first aspect, this application provides a dual-mode control system for plasma synthesis jets, comprising: A plasma-synthetic jet actuator array consists of N dual-mode compatible actuators arranged spanwise along the aerodynamic surface of an aircraft. Each dual-mode compatible actuator has a sealed discharge cavity and a jet outlet communicating with the discharge cavity. The dual-mode compatible actuator is configured to generate a shock wave for de-icing under nanosecond pulse drive and a pulsed jet for flow control under microsecond pulse drive. A dual-mode high-voltage pulse control controller is electrically connected to the plasma synthesis jet exciter array. The dual-mode high-voltage pulse control controller integrates a nanosecond pulse generation module and a microsecond pulse generation module, and switches between the nanosecond pulse generation module and the microsecond pulse generation module through a mode switching switch circuit. A multi-objective optimization and collaborative control unit is communicatively connected to the dual-mode high-voltage pulse control controller. The multi-objective optimization and collaborative control unit is configured to select the working mode and determine the excitation parameters based on real-time operating condition information. The sensor perception layer is communicatively connected to the multi-objective optimization and collaborative control unit. The sensor perception layer includes an ice layer parameter perception module and a boundary layer flow parameter perception module.

[0010] Furthermore, the dual-mode compatible exciter includes: The actuator body is made of alumina ceramic and has a block structure. The bottom of the actuator body is provided with a mounting flange for flush installation with the aircraft skin. After installation, the skin sinking amount is ≤0.2mm. The actuator body has a sealed discharge cavity inside, with electrodes inserted at both ends and a jet outlet on the upper surface. The discharge cavity is a cylindrical through-hole structure opened inside the actuator body, with a diameter of 2mm~6mm and a volume of 10mm². 3 ~80mm 3 ; The cathode discharge electrode and the anode discharge electrode are arranged in pairs facing each other, and are inserted laterally into the discharge cavity from the left and right ends of the actuator body, respectively, with their tips facing each other and a distance of 2mm to 8mm between them. The cathode discharge electrode is made of tungsten-copper alloy and plated with gold or platinum, with a frustoconical end and a cathode lead connected to its outer end. The anode discharge electrode is also made of tungsten-copper alloy and plated with gold or platinum, with a hemispherical end and an axial threaded adjustment structure, and anode lead connected to its outer end. Both the cathode lead and the anode lead are led out to the outside of the dual-mode compatible actuator for connection to a high-voltage pulse power supply. A multi-slit jet outlet is located on the upper surface of the actuator body, with its lower end connected to the interior of the discharge cavity. It consists of multiple parallel slits, with 3 to 12 slits, each slit being 0.4 mm to 0.8 mm wide, the center-to-center distance between adjacent slits being 1.5 mm to 3 mm, and the ratio of slit length to slit width being ≥10:1. The high-pressure gas inside the discharge cavity is ejected outward through the multi-slit jet outlet. A miniature thin-film heating element is embedded inside the exciter body, near the outer wall of the discharge cavity, for preheating the discharge cavity in extreme low-temperature environments.

[0011] Furthermore, the nanosecond pulse generation module outputs nanosecond-level high-voltage pulses with a pulse width of 20ns~200ns, a voltage of 8kV~25kV, and a single-shot energy of 15mJ~60mJ; the microsecond pulse generation module outputs microsecond-level pulse sequences with a pulse width of 1μs~20μs, a voltage of 2kV~10kV, a repetition frequency of 500Hz~5kHz, and a single-shot energy of 0.5mJ~5mJ.

[0012] Furthermore, the ice layer parameter sensing module is selected from fiber optic grating icing sensors or magnetostrictive icing sensors, and outputs ice layer thickness, spanwise coverage and icing rate in real time; the boundary layer flow parameter sensing module includes a high-frequency dynamic pressure sensor array and a wall shear stress sensor array, which are arranged 20mm~100mm downstream of the dual-mode compatible exciter on the wall surface. The multi-objective optimization and collaborative control unit includes a DSP processor and an FPGA. The FPGA receives the acquired signals from the ice layer parameter sensing module and the boundary layer flow parameter sensing module through multi-channel synchronous sampling. It performs real-time FFT spectrum analysis on the signals of each channel of the boundary layer flow parameter sensing module and extracts the root mean square value of pressure pulsation, characteristic frequency offset, and wall shear stress attenuation ratio as flow instability criteria. The DSP processor polls the sensor data preprocessed by the FPGA at a fixed period, determines the operating condition based on the ice layer parameters and the flow instability criteria, and sends the determination result and optimized excitation parameters to the dual-mode high-pressure pulse control controller.

[0013] Furthermore, it also includes a spanwise phase modulation module, which is communicatively connected to the multi-objective optimization and collaborative control unit. In the microsecond pulse flow control mode, the multi-objective optimization and collaborative control unit calculates the target phase velocity based on the incoming Mach number and the spanwise characteristics of the boundary layer separation region. The spanwise phase modulation module then calculates the target phase velocity according to the formula... For different spanning positions A dual-mode compatible exciter applies a programmable phase delay. ,in For the excitation frequency, The perturbation phase velocity is used to form a traveling wave excitation mode or a standing wave excitation mode in the spanwise direction; the traveling wave excitation mode is coupled with the natural perturbation mode in the boundary layer, and the phase resolution of the spanwise phase modulation module is ≤10ns.

[0014] Furthermore, the multi-objective optimization collaborative control unit incorporates an offline-trained Gaussian process regression surrogate model. This surrogate model uses the current operating condition vector as input to predict the control effect under different combinations of excitation parameters. The multi-objective optimization collaborative control unit converges to the globally optimal excitation parameters within 20-50 iterations through constrained Bayesian optimization. The multi-objective optimization collaborative control unit also possesses online self-evolution capability. After each control event, it stores the triplet of operating condition parameters, actual excitation parameters, and measured control effect into the running database. When the accumulated number of newly added valid data reaches a set number, it triggers an incremental update of the surrogate model. After the update, the prediction error is gradually reduced to below 5%, achieving intelligent control that adaptively evolves with the expansion of the flight envelope.

[0015] Furthermore, the multi-objective optimization and collaborative control unit is configured with preemptive priority management logic, where the icing threat has a higher priority than flow instability: When the system is in microsecond pulse flow control mode, if the ice layer parameter sensing module detects that the ice layer thickness exceeds a set threshold, the multi-objective optimization and collaborative control unit responds with an interruption delay of ≤0.5ms and executes the following sequence in sequence: save a snapshot of the current flow control state, cut off the output of the microsecond pulse generation module, the mode switching switch circuit switches to the nanosecond pulse generation module, calls the de-icing parameter optimization, performs nanosecond pulse de-icing until the ice layer thickness drops below the residual thickness for confirmation, quickly re-optimizes using the saved snapshot as the initial guess, switches back to the microsecond pulse generation module and restores microsecond pulse flow control; the entire process takes ≤100ms, and the lift coefficient fluctuation is <2%.

[0016] Secondly, this application provides a method for dual-mode control of plasma synthesis jets using the system described above, the method comprising: Ice layer parameters are collected through the ice layer parameter sensing module, and boundary layer flow parameters are collected through the boundary layer flow parameter sensing module. The multi-objective optimization and collaborative control unit determines the operating conditions based on the ice layer parameters and boundary layer flow parameters: When the ice layer thickness At that time, it enters the nanosecond pulse de-icing mode, in which To set a thickness threshold; When the ice layer thickness Furthermore, when the flow instability criterion is triggered, it enters the microsecond pulse flow control mode; When the ice layer thickness If the flow instability criterion is not triggered, it enters standby mode; In nanosecond pulse de-icing mode, the mode switching switch circuit switches to the nanosecond pulse generation module, the nanosecond pulse generation module outputs nanosecond-level high-voltage pulses to the plasma synthetic jet exciter array, and the plasma synthetic jet exciter array generates shock waves to peel off the ice layer. In the microsecond pulse flow control mode, the mode switching switch circuit switches to the microsecond pulse generation module, and the microsecond pulse generation module outputs a microsecond-level pulse sequence to the plasma synthetic jet exciter array. The plasma synthetic jet exciter array generates pulsed jets to suppress boundary layer separation.

[0017] Furthermore, in nanosecond pulse de-icing mode, the multi-objective optimization collaborative control unit invokes a Gaussian process regression surrogate model based on the current operating condition vector to perform constrained Bayesian optimization online search and determine the optimal excitation parameters. ,in This is the discharge voltage. For nanosecond pulse width, The repetition frequency, For a single energy input, The matrix is ​​transposed; the nanosecond pulse generation module outputs a high-voltage pulse according to the optimal excitation parameters; during the de-icing process, the ice layer parameter sensing module continuously monitors the change in ice layer thickness, and terminates the pulse output when the ice layer thickness drops below the residual thickness.

[0018] Furthermore, in the microsecond pulse flow control mode, the multi-objective optimization collaborative control unit invokes a Gaussian process regression surrogate model based on the current operating condition vector to perform constrained Bayesian optimization online search and determine the optimal excitation parameters. ,in This is the discharge voltage. For nanosecond pulse width, The repetition frequency, For a single energy input, The matrix is ​​transposed; the microsecond pulse generation module outputs a microsecond-level pulse sequence according to the optimal excitation parameters; at the same time, the spanwise phase modulation module configures the phase delay according to the coordinates of the dual-mode compatible exciter at different spanwise positions and the incoming flow conditions to form a traveling wave excitation mode or a standing wave excitation mode; during the flow control process, the boundary layer flow parameter sensing module continuously monitors the control effect, and when the flow parameters meet the standard and are continuously set for a set time, the repetition frequency is gradually reduced to the lowest effective frequency.

[0019] The dual-mode control system and method for plasma synthesis jet provided in this application have at least the following beneficial effects: 1. This application enables dual-mode physical coordination of nanosecond impact de-icing and microsecond momentum flow control on the same plasma synthetic jet hardware platform. In nanosecond mode, the dual-mode high-voltage pulse control controller switches to the nanosecond pulse generation module via a mode switching switch circuit, outputting nanosecond-level high-voltage pulses with pulse widths of 20ns~200ns, voltages of 8kV~25kV, and peak power of 300kW~750kW to the plasma synthetic jet exciter array. This generates a strong shock wave within the discharge cavity on a timescale of hundreds of nanoseconds, which radiates outwards through the multi-slit jet outlet. The shock wave penetrates the ice layer, generating tensile stress wave reflections at the ice-skin interface, achieving instantaneous layered delamination of the ice layer. In microsecond mode, the controller switches to the microsecond pulse generation module, outputting a microsecond-level pulse sequence with pulse widths of 1μs~20μs, voltages of 2kV~10kV, and repetition frequencies of 500Hz~5kHz. This continuously injects time-averaged momentum into the boundary layer via a high-speed pulse jet on a timescale of thousands of times, suppressing or delaying boundary layer separation. By limiting the discharge cavity volume to 10 mm 3 ~80mm 3 The multi-slit jet outlet slit width is limited to 0.4mm~0.8mm, the number of slits is limited to 5~8, and the electrode spacing is limited to 2mm~8mm. This allows the same dual-mode compatible exciter to simultaneously meet the key geometric parameter requirements of both modes. Combined with the quantitative criterion that both the nanosecond compatibility factor and the microsecond compatibility factor are not less than 0.75, the balanced realization of dual-mode performance can be ensured.

[0020] 2. This application can significantly reduce the weight and complexity of the airborne system. By merging the two separate hardware sets for de-icing and flow control into a single dual-mode plasma synthetic jet actuator array and a matching dual-mode high-voltage pulse control controller, the number of actuators can be reduced by approximately 50%, the weight of the power system by approximately 30%, and the installation space by approximately 40%. Taking a long-endurance UAV with a wingspan of approximately 12m as an example, it is estimated that 8kg to 15kg of airborne weight can be saved, which can correspondingly increase the payload or extend the endurance. The pulsed intermittent operation of the nanosecond de-icing mode is triggered only when the ice thickness exceeds a threshold, and the output stops after de-icing is completed. Its average energy consumption is significantly lower than that of continuous electrothermal de-icing. The sensor perception layer judges the operating condition with a period of approximately 10ms, and only activates the corresponding operating mode when icing or flow instability occurs. During other periods, the system maintains low-power standby, which can further reduce the system's operating energy consumption.

[0021] 3. This application, through dual-mode collaborative switching and safety priority management, enables the system to autonomously handle alternating icing and flow instability conditions without human intervention. The mode switching switch circuit inside the dual-mode high-pressure pulse control controller uses a solid-state switch to achieve high-speed switching between the nanosecond pulse generation module and the microsecond pulse generation module, with a switching time of no more than 5μs. The multi-objective optimization collaborative control unit is configured with a preemptive scheduling strategy that prioritizes icing threats over flow instability. When the system is in the microsecond pulse flow control mode, if the ice layer parameter sensing module detects that the ice layer thickness exceeds the set threshold, the control unit responds with an interruption delay of no more than 0.5ms. It can sequentially execute the following processes: saving the current flow control state snapshot, cutting off the output of the microsecond pulse generation module, switching the mode switching switch circuit to the nanosecond pulse generation module, calling the de-icing parameter optimization, performing nanosecond pulse de-icing until the ice layer thickness drops below the residual thickness, quickly re-optimizing based on the saved snapshot as the initial guess, and switching back to the microsecond pulse generation module and restoring flow control. The entire process takes no more than 100ms, and the lift coefficient fluctuation is less than 2%, which can ensure the aerodynamic stability of the flight platform under sudden icing conditions.

[0022] 4. This application introduces spatial dimension control degrees of freedom through spanwise traveling wave and standing wave phase modulation, which can improve flow control efficiency. The spanwise phase modulation module is communicatively connected to the multi-objective optimization and collaborative control unit. In the microsecond pulse flow control mode, the control unit calculates the target phase velocity based on the incoming Mach number and the spanwise characteristics of the boundary layer separation region. The spanwise phase modulation module applies a programmable phase delay to the dual-mode compatible exciters at different spanwise positions according to the spanwise coordinates and excitation frequency of each exciter, which can form a traveling wave excitation mode or a standing wave excitation mode in the spanwise direction. The traveling wave excitation mode can be coupled with the spatial modes of natural disturbances in the boundary layer to generate spatial modulation enhancement of flow control. Compared with the in-phase excitation mode, it can improve the flow control efficiency by 15% to 30%, and realize the spanwise spatial modulation capability that a single excitation mode does not have.

[0023] 5. This application utilizes a surrogate model-assisted online optimization and self-evolution capability to enable the system to achieve intelligent control that evolves from fixed parameters to adaptive evolution. The multi-objective optimization collaborative control unit incorporates an offline-trained Gaussian process regression surrogate model. This surrogate model takes the current operating condition vector as input and can predict the control effect under different combinations of excitation parameters. Through constrained Bayesian optimization, it converges to the globally optimal excitation parameters within 20-50 iterations, solving the problem of rapid online optimization in the four-dimensional excitation parameter space. After each control event, the control unit stores the triplet of operating condition parameters, actual excitation parameters, and measured control effect into the operating database. When the accumulated new effective data reaches a set number of sets, the incremental update of the surrogate model is triggered. As operating data accumulates, the model prediction error can gradually decrease from the initial 12% to below 5%, enabling intelligent control that evolves adaptively with the expansion of the flight envelope. Attached Figure Description

[0024] 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.

[0025] Figure 1 A structural diagram of a dual-mode control system for plasma synthesis jet provided in an embodiment of this application; Figure 2 An axial cross-sectional view of a dual-mode compatible exciter in a plasma synthesis jet dual-mode control system provided in this application embodiment; Figure 3 Jet schlieren images of a multi-slit PSJ actuator provided in an embodiment of this application; wherein, (a), t=0μs; (b), t=10μs; (c), t=20μs; (d), t=30μs; (e), t=40μs; (f), t=50μs; Figure 4 A comparison diagram of discharge waveform characteristics between nanosecond de-icing mode and microsecond flow control mode provided in the embodiments of this application; wherein, (a) nanosecond de-icing mode; (b) microsecond flow control mode; (c) comparison results; Figure 5 A flowchart illustrating the process of a dual-mode control system for plasma synthesis jets provided in this application embodiment; Figure 6 A flowchart of a dual-mode control method for plasma synthesis jet provided in this application embodiment.

[0026] Explanation of reference numerals in the attached figures: 100. Plasma synthetic jet actuator array; 110. Dual-mode compatible actuator; 111. Actuator body; 112. Discharge cavity; 113. Cathode discharge electrode; 114. Anode discharge electrode; 115. Multi-slit jet outlet; 116. Cathode lead; 117. Anode lead; 118. Micro-thin film heating element; 119. Mounting flange; 200. Dual-mode high-voltage pulse control controller; 210. Nanosecond pulse generation module; 220. Microsecond pulse generation module; 230. Mode switching switch circuit; 300. Multi-objective optimization and collaborative control unit; 400. Sensor sensing layer; 410. Ice layer parameter sensing module; 420. Boundary layer flow parameter sensing module.

[0027] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation

[0028] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0029] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0030] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0031] Example 1: This application provides a dual-mode control system for plasma synthesis jets, such as... Figure 1 As shown, the system includes a plasma synthetic jet exciter array 100, a dual-mode high-voltage pulse control controller 200, a multi-objective optimization and collaborative control unit 300, and a sensor sensing layer 400.

[0032] The plasma synthetic jet actuator array 100 consists of N dual-mode compatible actuators 110 arranged longitudinally along the aerodynamic surface of the aircraft. Each dual-mode compatible actuator 110 has a sealed discharge cavity and a jet outlet communicating with the discharge cavity. The dual-mode compatible actuator 110 is configured to generate a shock wave for de-icing under nanosecond pulse drive and to generate a pulsed jet for flow control under microsecond pulse drive.

[0033] The dual-mode high-voltage pulse control controller 200 is electrically connected to the plasma synthetic jet exciter array 100. The dual-mode high-voltage pulse control controller 200 integrates a nanosecond pulse generation module 210 and a microsecond pulse generation module 220, and switches between the nanosecond pulse generation module 210 and the microsecond pulse generation module 220 through a mode switching switch circuit 230.

[0034] The mode switching circuit 230 employs solid-state switches based on SiC MOSFETs and IGBTs, achieving rapid switching between the nanosecond pulse generation module 210 and the microsecond pulse generation module 220 through a break-before-make safety logic. Specifically, when a switching command arrives, the MOSFET or IGBT gate drive signal of the currently operating module is pulled low within 100ns. After a dead time of at least 1μs to confirm complete shutdown, the gate drive signal of the target module is pulled high within 100ns and pre-charged to the target bus voltage before outputting enable. The total switching time does not exceed 5μs. The break-before-make logic ensures no arc crosstalk and output overlap during mode transition, preventing short-circuit faults caused by simultaneous conduction of both modules. Through this switching mechanism, the system can seamlessly switch between nanosecond de-icing mode and microsecond flow control mode, meeting the rapid response requirements of the aircraft under alternating icing and flow instability conditions.

[0035] The multi-objective optimization and coordinated control unit 300 is communicatively connected to the dual-mode high-voltage pulse control controller 200. The multi-objective optimization and coordinated control unit 300 is configured to select the working mode and determine the excitation parameters based on real-time operating condition information.

[0036] The multi-objective optimization and collaborative control unit 300, as the core decision-making node of the system, is responsible for running the proxy model, executing online parameter optimization, managing dual-mode switching logic, controlling spanwise phase modulation, and coordinating the collaborative work of various modules. It adopts a heterogeneous architecture of DSP and FPGA, where the FPGA is responsible for high-speed data acquisition and real-time signal preprocessing, and the DSP is responsible for complex algorithm calculations and decision output. Through this division of labor and collaboration, it achieves millisecond-level response to the input of the sensor sensing layer 400.

[0037] The sensor sensing layer 400 is communicatively connected to the multi-objective optimization and collaborative control unit 300. The sensor sensing layer 400 includes an ice layer parameter sensing module 410 and a boundary layer flow parameter sensing module 420.

[0038] The ice layer parameter sensing module 410 is used to sense the icing state of the aircraft surface in real time, and the boundary layer flow parameter sensing module 420 is used to monitor the boundary layer flow separation state in real time. These two types of sensor information constitute the input basis for the system to make dual-mode decisions.

[0039] In some embodiments, such as Figure 2 As shown, the dual-mode compatible actuator 110 includes an actuator body 111, a discharge cavity 112, a cathode discharge electrode 113, an anode discharge electrode 114, a multi-slit jet outlet 115, and a micro-thin film heating element 118.

[0040] The actuator body 111 is made of alumina ceramic and has a block structure. The bottom of the actuator body 111 is provided with a mounting flange 119, which is used to install flush with the aircraft skin. After installation, the skin sinks by ≤0.2mm. The actuator body 111 has a sealed discharge cavity 112 inside, with electrodes inserted at both ends and a jet outlet on the upper surface.

[0041] The alumina ceramic material has a long-term operating temperature of not less than 200℃ and a breakdown field strength of not less than 20kV / mm, and can withstand the high temperature and high pressure impact generated by high voltage pulse discharge. The exciter body 111 is embedded into the aircraft skin for conformal installation via mounting flange 119, which has minimal impact on the aerodynamic shape of the aircraft and is suitable for unmanned aerial platforms with strict weight and space constraints.

[0042] The discharge cavity 112 is a cylindrical through-hole structure opened inside the actuator body 111, with a diameter of 2mm~6mm and a volume of 10mm². 3 ~80mm 3 .

[0043] The cathode discharge electrode 113 and the anode discharge electrode 114 are arranged in pairs facing each other, and are inserted laterally into the discharge cavity 112 from the left and right ends of the exciter body 111, respectively. The tips of the two electrodes face each other, and the electrode spacing is 2mm to 8mm. The cathode discharge electrode 113 is made of tungsten copper alloy and is plated with gold or platinum. The end is frustoconical and the outer end is connected to the cathode lead 116. The anode discharge electrode 114 is made of tungsten copper alloy and is plated with gold or platinum. The end is hemispherical and has an axial thread adjustment structure. The outer end is connected to the anode lead 117. Both the cathode lead 116 and the anode lead 117 are led out to the outside of the dual-mode compatible exciter 110 for connection to a high-voltage pulse power supply.

[0044] The frustoconical end design of the cathode discharge electrode 113 helps to concentrate the electric field and reduce the breakdown voltage. The hemispherical end design of the anode discharge electrode 114 can reduce tip erosion and extend electrode life. Its axial thread adjustment structure can finely adjust the electrode spacing g during assembly to adapt to the breakdown voltage requirements under different air pressure conditions, ensuring reliable breakdown even at an altitude of 20km and an ambient air pressure of about 5.5kPa.

[0045] The multi-slit jet outlet 115 is located on the upper surface of the exciter body 111 and its lower end is connected to the interior of the discharge cavity 112. It consists of multiple parallel slits, with 3 to 12 slits. The width of a single slit is 0.4 mm to 0.8 mm, the center-to-center distance between adjacent slits is 1.5 mm to 3 mm, and the ratio of slit length to slit width is ≥10:1. The high-pressure gas inside the discharge cavity 112 is ejected outward through the multi-slit jet outlet 115.

[0046] The multi-slit jet outlet 115, while maintaining a total outlet area comparable to traditional single-hole or single-slit structures, distributes the jet uniformly along the spanwise direction through multiple parallel narrow slits. This increases the spanwise contact area between the jet and the boundary layer. Simultaneously, the multi-slit flow distribution reduces the mass flow burden of a single slit, alleviating the gas backfill bottleneck in microsecond mode. Taking a typical configuration with 6 slits, a slit width of 0.6 mm, and a slit length of 22 mm as an example, the total outlet area is approximately 7.92 mm². 2 It can form a uniform jet coverage zone with a width of about 12 mm in the spanwise direction.

[0047] The micro-thin film heating element 118 is embedded inside the exciter body 111, near the outer wall of the discharge cavity 112, and is used to preheat the discharge cavity 112 in extreme low temperature environments.

[0048] When the ambient temperature is not higher than -30℃, the micro-thin film heating element 118 starts preheating, raising the temperature of the gas inside the discharge cavity 112 to not lower than -10℃. This ensures that the gas inside the cavity has sufficient initial internal energy to respond to the high-voltage pulse discharge and avoids excessively high gas density or breakdown voltage drift due to low temperature, which could affect discharge stability. The preheating power consumption is approximately 2W to 5W, and the preheating time does not exceed 30s.

[0049] It should be noted that the dual-mode compatible actuator 110 is the physical basis for enabling nanosecond de-icing and microsecond flow control to work collaboratively on a unified hardware platform. Its core design challenge lies in the fact that nanosecond and microsecond pulses impose directional differences on the actuator's key geometric parameters, necessitating the achievement of dual-mode compatibility within an integrated structure. This embodiment achieves the integrated design of the dual-mode compatible actuator 110 through the coordinated optimization of three key parameters: discharge cavity volume, jet outlet structure, and electrode spacing.

[0050] Specifically, regarding the discharge cavity volume Compatibility design. Nanosecond de-icing prefers small cavities (<10mm). 3 (Facilitates rapid pressure boosting and the establishment of a strong shock wave), microsecond flow control prefers large cavities (>80mm) 3 This is beneficial for multi-cycle stable operation and sufficient gas backfilling. Through coupled simulation of PSJ cavity plasma dynamics and compressible Navier-Stokes equations, in... ∈[5,120]mm 3 Within the range, 1mm 3 Step size is used for dual-mode parameter scanning, with the peak pressure of the shock wave measured in nanosecond mode. and microsecond mode average mass flow rate As a dual-objective evaluation metric, the dual-mode compatibility window was determined to be 10-80mm. 3Within this range, the key performance indicators of the two modes are no less than 75% of their respective optimal values. Further optimization is preferred for 30-50mm. 3 It can achieve more than 85% of the optimal performance in both modes.

[0051] For the compatibility design of multi-slit jet outlet structures, the design of the jet outlet slit width w faces similar directional differences: nanosecond mode prefers narrow slits (0.4-0.55mm) to enhance the throttling effect and increase the peak pressure of the shock wave inside the cavity; microsecond mode prefers wide slits (0.6-0.8mm) to reduce exhaust resistance and increase the jet mass flow rate. Based on isentropic jet theory and CFD simulation of the throttling effect, the dual-mode compatible slit width window is determined to be 0.4-0.8mm. The choice of the number of slits also reflects the dual-mode trade-off: fewer slits (3-6 slits) result in a larger single-slit flow rate, with the microsecond mode being superior; more slits (8-12 slits) result in a higher jet velocity, with the nanosecond mode being superior. Taking 5-8 slits as the compatibility benchmark, with the aspect ratio of each slit ≥12 (slit length 10-25mm), the adaptive balance of dual-mode performance is achieved by adjusting the discharge voltage U and pulse width τ in conjunction with the outlet geometry. Compared to the traditional single-hole / single-slit structure, the multi-slit outlet increases the spanwise contact area between the jet and the boundary layer while maintaining a comparable total outlet area. At the same time, the multi-slit diversion reduces the mass flow burden of the single slit and alleviates the gas backfilling bottleneck in microsecond mode.

[0052] To verify the practical effect of the multi-slit jet outlet design, high-speed schlieren imaging technology was used to conduct a visual experimental observation of the exciter jet field. Schlieren images ( Figure 3 The study clearly demonstrated the jet wave system structure and flow field evolution characteristics generated by the multi-slit outlet, confirming the effectiveness of the multi-slit design in forming a uniform jet array in the spanwise direction and increasing the contact area between the jet and the boundary layer. It also provided direct experimental evidence for the engineering feasibility of dual-mode compatible outlet structures.

[0053] For the compatibility design of the electrode spacing g, the electrode spacing is determined to be 2-8mm based on Paschen's breakdown law and corrections for high-altitude low-pressure conditions. A lower limit of 2mm is used to avoid excessively high arc energy density leading to rapid electrode ablation (expected lifespan ≥10). 6 The upper limit of 8mm ensures reliable breakdown at an altitude of 20km (ambient air pressure ≈ 5.5kPa) (breakdown voltage ≤ 25kV design margin). Within this range, both pulse modes can obtain stable and controllable discharge characteristics.

[0054] The actuator housing is made of alumina ceramic (long-term operating temperature ≥200℃, breakdown field strength ≥20kV / mm). The cathode is a tungsten-copper alloy with a frustoconical end, and the anode is a tungsten-copper alloy with a hemispherical end and an axial threaded adjustment mechanism. A mounting flange is provided at the bottom for flush installation with the aircraft skin (sinking ≤0.2mm). Miniature thin-film platinum resistance heating elements are embedded in the outer wall of the cavity for use in extreme low-temperature environments. The preheating chamber is set at -30℃ to ensure reliable start-up of both modes under cold high-altitude conditions.

[0055] The above discharge cavity volume The optimized combination of four geometric parameters—jet outlet slit width w, slit number n, and electrode spacing g—forms the integrated design basis for the dual-mode compatible exciter 110. 30-50mm 3 With an optimal parameter combination of w = 0.5-0.7mm, n = 5-8 stripes, and g = 3-6mm, CFD simulation verified the peak shock wave pressure in nanosecond de-icing mode. P peak The time-averaged mass flow rate can reach 1.5 MPa to 2.5 MPa in microsecond flow control mode. The speed can reach 0.5g / s to 1.2g / s, and the performance indicators of both modes can reach more than 85% of the optimal design of their respective single modes.

[0056] In some embodiments, two fundamentally different operating modes—nanosecond pulse instantaneous de-icing and microsecond pulse continuous flow control—are implemented on the same PSJ actuator hardware platform. Their physical mechanisms and energy characteristics complement each other. A comparison of the dual-mode discharge waveform characteristics is shown below. Figure 4 As shown, the nanosecond pulse generation module 210 outputs nanosecond-level high-voltage pulses with a pulse width of 20ns~200ns, a voltage of 8kV~25kV, and a single-shot energy of 15mJ~60mJ; the microsecond pulse generation module 220 outputs microsecond-level pulse sequences with a pulse width of 1μs~20μs, a voltage of 2kV~10kV, a repetition frequency of 500Hz~5kHz, and a single-shot energy of 0.5mJ~5mJ.

[0057] Specifically, for the nanosecond pulse instantaneous de-icing mode, the high-voltage pulse power supply outputs a nanosecond-level high-voltage pulse with a pulse width τ = 20-200ns and a voltage U = 8-25kV to the exciter discharge cavity. The single discharge energy is E = 15-60mJ, and the peak power reaches 300-750kW. Due to the energy injection rate ( >10 11The pressure (W / s) far exceeds the gas expansion rate, causing the internal pressure to surge to the MPa level, radiating a strong shock wave outward through the multi-slit jet outlet. The shock wave penetrates the skin, generating tensile stress wave reflections at the ice-skin interface (acoustic impedance mismatch) and internal ice defects. When the tensile stress exceeds the ice's fracture strength (-1-2 MPa), penetrating microcracks form within the ice layer, achieving instantaneous layered delamination. This inside-out stress wave delamination mechanism allows energy to act directly on the interface, with a response speed on the order of hundreds of milliseconds. Typical performance: ice layer delamination time for d=1mm. =30-100ms, single de-icing energy consumption ≤3J / cm³ 2 Ice surface. The intensity and range of the shock wave can be controlled by adjusting the discharge voltage U and pulse width τ, adapting to ice layers of different thicknesses and types.

[0058] It should be noted that the operating frequency f of the nanosecond de-icing mode is typically 1Hz~100Hz. A single de-icing process consists of multiple nanosecond pulses, and the ice layer parameter sensing module 410 continuously detects the residual ice thickness during the intervals between adjacent pulses. The ice thickness is used as the basis for... d Taking 1mm as an example, a typical de-icing pulse sequence is: initial discharge voltage U 12kV, pulse width τ 100ns, frequency f At a frequency of 10Hz, approximately 5-10 pulses are sufficient to completely peel off the ice layer. If the ice is thick... d If the voltage is greater than 2mm, the system will automatically adjust the discharge voltage. U Increase the voltage to 18kV~20kV and appropriately extend the pulse width. τ To 150ns~200ns, the shock wave intensity is increased to strip away thicker ice layers.

[0059] Microsecond pulse continuous flow control mode. The pulse power supply switches to the microsecond module, and the output pulse width... τ =1-20μs, voltage U =2-10kV, repetition frequency f =500Hz-5kHz microsecond-level pulse sequence, single energy E =0.5-5mJ. Each pulse cycle undergoes a complete cycle of "discharge heating → gas expansion and injection → intracavitary depressurization → external gas backfilling", forming a high-speed pulse jet (peak velocity >300m / s) at the jet outlet, continuously injecting a time-averaged momentum of -10 into the boundary layer. -2 On the order of N·s, this method overcomes the adverse pressure gradient by enhancing momentum exchange between the near-wall region and the mainstream, thus suppressing or delaying boundary layer separation. This is achieved by adjusting... f , U and τ It can precisely control the time-averaged intensity and spectral characteristics of the injected jet flow to match the separation and suppression requirements under different incoming flow conditions.

[0060] In microsecond pulse flow control mode, the operating frequency of the actuator f The degree of matching with the natural disturbance frequency within the aircraft's boundary layer directly affects the control performance. Taking a UAV wing with a cruising Mach number of 0.5 and a chord length of 1.5m as an example, the characteristic frequency of the natural disturbance within the boundary layer is approximately 1kHz to 3kHz. At this point, the output frequency of the microsecond pulse generator module 220 is set... f With a frequency of 1.5kHz to 2.5kHz, the synthetic jet can resonate with the natural disturbance of the boundary layer, significantly improving the momentum injection efficiency.

[0061] The essential complementarity of the energy characteristics of the two modes is explained as follows: The core characteristics of the nanosecond mode are extremely high instantaneous power and low time-averaged momentum. The peak power of a single pulse is 300-750kW, but the pulse width is only on the order of hundreds of nanoseconds, and the time-averaged momentum is only -10. -4 The N·s level; the core characteristics of the microsecond mode are moderate peak power and high time-averaged momentum; the peak power of a single pulse is <2.5kW, but the time-averaged momentum accumulated through kHz-level repetition frequency reaches -10. -2 The complementarity of the energy-timescales of the two modes is the physical basis for the dual-mode fusion design. The nanosecond mode specializes in the "fast, strong, and instantaneous" de-icing task, while the microsecond mode focuses on the "stable, sustained, and uniform" flow control task. By exploring the differentiated response characteristics of the same physical platform at different timescales, a complete functional spectrum is achieved on the same hardware.

[0062] This embodiment also provides verification criteria for dual-mode compatibility. To ensure that the actuator design truly meets the dual-mode compatibility requirements, this embodiment establishes a quantitative dual-mode compatibility evaluation index and defines a nanosecond compatibility factor. and microsecond compatibility factor ,in and These are the performance indicators for the optimal single-mode design under the corresponding geometric parameter combinations. The dual-mode compatibility condition is... ≥0.75 and ≥0.75. In =30-50mm 3 With the design parameters w=0.5-0.7mm, g=3-6mm, and n=5-8, and It can achieve a value of 0.85 or higher, enabling efficient and compatible operation of two pulse modes on the same hardware platform.

[0063] The aforementioned compatibility evaluation indicators are not only used for parameter verification during the exciter design phase, but can also be embedded in the multi-objective optimization collaborative control unit 300 as performance constraints during mode switching. When the system needs to adjust discharge parameters due to environmental changes, such as changes in gas density caused by high altitude and low temperature, the control unit 300 calculates the current operating conditions in real time. and The value ensures that the compatibility baseline for both modes is no lower than 0.75 under any flight conditions.

[0064] In some embodiments, the ice layer parameter sensing module 410 is selected from a fiber optic grating icing sensor or a magnetostrictive icing sensor, and outputs the ice layer thickness, spanwise coverage and icing rate in real time; the boundary layer flow parameter sensing module 420 includes a high-frequency dynamic pressure sensor array and a wall shear stress sensor array, which are arranged at a wall surface 20mm~100mm downstream of the dual-mode compatible exciter 110.

[0065] In this embodiment, the ice layer parameter sensing module consists of a fiber optic grating (FBG) icing sensor or a magnetostrictive icing sensor, arranged in the spanwise gap between the exciters, and outputs the ice layer thickness in real time. d (Accuracy 0.02-0.05mm), spanwise coverage θ , freezing rate The boundary layer flow parameter sensing module consists of a high-frequency dynamic pressure sensor array (natural frequency ≥ 50 kHz) and a wall shear stress sensor array, which are arranged 20-100 mm downstream of the exciter on the wall surface.

[0066] In some embodiments, the multi-objective optimization and collaborative control unit 300 includes a DSP processor and an FPGA. The FPGA receives the acquired signals from the ice layer parameter sensing module 410 and the boundary layer flow parameter sensing module 420 through multi-channel synchronous sampling. It performs real-time FFT spectrum analysis on the signals of each channel of the boundary layer flow parameter sensing module 420 and extracts the root mean square value of pressure pulsation, characteristic frequency offset, and wall shear stress attenuation ratio as flow instability criteria. The DSP processor polls the sensor data preprocessed by the FPGA at a fixed period, determines the operating condition based on the ice layer parameters and flow instability criteria, and sends the determination result and optimized excitation parameters to the dual-mode high-pressure pulse control controller 200.

[0067] The FPGA performs a 512-point real-time FFT analysis on the pressure signals of each channel, with a time window of approximately 1ms to 5ms, depending on the sampling rate of 1kSPS to 10kSPS, and can extract the root mean square value of pressure pulsations in real time. Characteristic frequencies and its offset Average value of wall shear stress sensor output , and no separation reference value The attenuation ratio was obtained by comparison. .when Greater than 1.2 times the stable benchmark value and Less than 0.8 times When this is detected, it is determined to be a precursor to flow instability, triggering the microsecond flow control mode.

[0068] In this embodiment, the FPGA performs real-time FFT spectrum analysis on the acquired signal to extract flow instability criteria such as the root mean square value of pressure pulsation, characteristic frequency offset, and wall shear stress attenuation ratio.

[0069] The operating condition judgment logic of dual-source sensor fusion is as follows: ice layer sensor data takes precedence over flow sensor data (safety first principle). When the ice layer thickness data is valid and exceeds the threshold, the system unconditionally enters the de-icing mode regardless of the flow state; only after the ice layer threat is eliminated will the system determine whether to activate the flow control mode based on the flow sensor data. This hierarchical fusion strategy ensures the safety-first dual-mode scheduling logic at the sensor level.

[0070] It should be noted that the sampling period of both the ice layer parameter sensing module 410 and the boundary layer flow parameter sensing module 420 is approximately 10ms, which is synchronized with the operating condition judgment period of the multi-objective optimization and collaborative control unit 300. When the ice layer parameter sensing module 410 measures the ice thickness... d When the ice layer thickness is not less than 0.5 mm, the system can trigger the nanosecond de-icing mode within the next 10 ms cycle. The overall response delay is no more than 20 ms, which is much faster than the growth rate of the ice layer under the typical freezing rate, and can achieve rapid suppression in the early stage of freezing.

[0071] In some embodiments, the plasma synthesis jet dual-mode control system further includes a spanwise phase modulation module, which is communicatively connected to a multi-objective optimization and collaborative control unit 300. In microsecond pulse flow control mode, the multi-objective optimization and collaborative control unit 300 calculates the target phase velocity based on the incoming Mach number and the spanwise characteristics of the boundary layer separation region. The spanwise phase modulation module then calculates the target phase velocity according to the formula... For different spanning positions The dual-mode compatible exciter 110 applies a programmable phase delay. ,in For the excitation frequency, The perturbation phase velocity is used to form a traveling wave excitation mode or a standing wave excitation mode in the spanwise direction; the traveling wave excitation mode is coupled with the natural perturbation mode in the boundary layer, and the phase resolution of the spanwise phase modulation module is ≤10ns.

[0072] With 24 exciters arranged along the span, the excitation frequency is... 2kHz, perturbation phase velocity Taking 50 m / s as an example, the spanwise spacing between adjacent exciters When the phase delay is 35mm, Approximately 0.0088 rad, corresponding to about 0.5 degrees. The phase resolution of the spanwise phase modulation module does not exceed 10 ns, corresponding to a phase resolution of about 0.0072 degrees for a 2 kHz signal, far exceeding the required accuracy, and can accurately synthesize traveling wave excitation. When the traveling wave direction is consistent with the incoming flow direction, it is the same as the propagation direction of the Tollmien-Schlichting wave within the boundary layer, which can form resonant coupling, and the control efficiency can be improved by about 15%~30% compared with in-phase excitation. The standing wave excitation mode is suitable for the case of spanwise symmetrical distribution of the separation region. By forming fixed antinode and node positions in the spanwise direction, the maximum momentum injection is applied at the center of the separation region.

[0073] In some embodiments, the multi-objective optimization collaborative control unit 300 has a built-in Gaussian process regression surrogate model trained offline. The Gaussian process regression surrogate model takes the current operating condition vector as input and predicts the control effect under different combinations of excitation parameters. The multi-objective optimization collaborative control unit 300 converges to the globally optimal excitation parameters within 20 to 50 iterations through constrained Bayesian optimization. The multi-objective optimization collaborative control unit 300 also has online self-evolution capability. After each control event is completed, the triplet of operating condition parameters, actual excitation parameters and measured control effect is stored in the running database. When the accumulated new effective data reaches a set number of groups, the incremental update of the surrogate model is triggered. After the update is completed, the prediction error is gradually reduced to below 5%, realizing intelligent control that adaptively evolves with the expansion of the flight envelope.

[0074] The offline training process of the Gaussian process regression surrogate model is as follows: Latin hypercube sampling is used to sample no fewer than 200 operating points within the flight envelope (altitude H = 5–20 km, Mach number Ma = 0.2–1.5, angle of attack α = -2–18 degrees) and the excitation parameter space. A training dataset is constructed through CFD coupled simulation. After training, the model parameters are deployed to the on-chip Flash memory of the DSP processor. During constrained Bayesian optimization, the combination of excitation parameters in each iteration must satisfy the dual-mode compatibility constraint, i.e. Not less than 0.75 and A value of at least 0.75 is used to exclude physically infeasible parameter regions, reducing the search space by approximately 40% and accelerating convergence to the global optimum.

[0075] In some embodiments, the multi-objective optimization and collaborative control unit 300 is configured with preemptive priority management logic, where the icing threat has a higher priority than flow instability: When the system is in microsecond pulse flow control mode, if the ice layer parameter sensing module 410 detects that the ice layer thickness exceeds a set threshold, the multi-objective optimization and collaborative control unit 300 responds with an interruption delay of ≤0.5ms and executes the following sequence in sequence: saves a snapshot of the current flow control state, cuts off the output of the microsecond pulse generation module 220, the mode switching switch circuit 230 switches to the nanosecond pulse generation module 210, calls the de-icing parameter optimization, performs nanosecond pulse de-icing until the ice layer thickness drops below the residual thickness for confirmation, quickly re-optimizes using the saved snapshot as the initial guess, switches back to the microsecond pulse generation module 220 and restores microsecond pulse flow control; the entire process takes ≤100ms and the lift coefficient fluctuation is <2%.

[0076] The typical time allocation for preemptive handover is as follows: interrupt response and state snapshot saving no more than 0.5ms; microsecond module shutdown and nanosecond module switching including pre-charging approximately 3 to 5ms; de-icing parameter optimization limited to 15 CBO iterations approximately 2 to 3ms; nanosecond de-icing execution approximately 30 to 50ms corresponding to ice thickness of 0.5 to 1.5mm; de-icing confirmation sampling approximately 20ms; rapid flow control re-optimization with the snapshot as the initial value for 5 to 10 iterations approximately 2 to 5ms; and switching back to the microsecond module approximately 3 to 4ms, totaling approximately 60 to 90ms, meeting the design target of no more than 100ms. The mechanism to ensure lift coefficient fluctuations of less than 2% is as follows: during handover and de-icing, short-term compensation from the aircraft's inertia and flight control system keeps aerodynamic parameters stable, and flow control is immediately restored after de-icing to prevent boundary layer separation deterioration from causing further lift reduction.

[0077] In some embodiments, such as Figure 5 The diagram shown is a flowchart of a dual-mode control system for plasma synthesis jets provided in this application embodiment. When the dual-mode control system for plasma synthesis jets is in operation, it includes the following steps S1-S6.

[0078] After the system is powered on and initialized, it first completes global parameter initialization, module self-test and communication calibration, and then enters the normal closed-loop control working state.

[0079] S1: Status monitoring.

[0080] The ice layer parameter sensing module 410 and the boundary layer flow sensing module 420 collect the icing parameters on the aircraft surface and the airflow boundary layer flow parameters in real time and continuously, and transmit all the collected data to the multi-objective optimization and collaborative control unit 300 in real time to provide data support for subsequent operating condition judgment and mode control.

[0081] The parameters collected by the ice layer parameter sensing module 410 include ice thickness. d, icing angle θ Ice thickness change rate d d / d t Ambient temperature T Key parameters include Mach number (Ma) and angle of attack. The boundary layer flow sensing module 420 collects parameters such as Mach number (Ma) and angle of attack. α Wall shear stress, wall dynamic shear stress τ Wall pressure pulsation p' Key flow parameters, etc.

[0082] It should be noted that the angle of icing θ This refers to the percentage of ice cover in the spanwise direction of the actuator, that is, the ratio of the spanwise length of the ice cover to the total spanwise length of the actuator array, used to characterize the spanwise non-uniformity of ice distribution. Ice thickness variation rate d d / d t The sign and magnitude of d reflect the tendency of icing to intensify or ablate. d / d t A speed greater than 0.3 mm / min indicates a rapid icing condition. In subsequent de-icing parameter optimization, the system will prioritize the speed target, with icing speed weighting. α Set it to 0.8.

[0083] S2: Operating condition judgment.

[0084] The multi-objective optimization and collaborative control unit 300 performs real-time calculation and analysis on the received monitoring data, and completes the working condition classification judgment based on the preset threshold. The specific judgment logic is as follows: 1. When the detected ice thickness meets the requirements Among them, ice thickness threshold The value range is 0.3mm~0.8mm, and the system will switch to the S3 short pulse de-icing mode. 2. When the detected ice thickness meets the requirements When the sensor detects a precursor signal of flow instability in the airflow boundary layer, the system switches to the S4 long pulse flow control mode. 3. When both the icing state and the airflow state are in a normal and stable state, the system will switch to the S5 standby monitoring mode.

[0085] The default value is 0.5mm, and this threshold can be adjusted online by the flight control computer according to the flight phase: 0.3mm is recommended for the climb and cloud penetration phase, a more conservative strategy to advance de-icing; 0.5mm~0.8mm is recommended for the high-altitude cruise phase, allowing for a small amount of ice accumulation to reduce de-icing energy consumption. The criterion for the precursor signal of flow instability is the root mean square value of wall pressure pulsations. Greater than 1.2 times the stable benchmark value or wall shear stress Less than 0.8 times the non-separation reference value or characteristic frequency offset Greater than 10% of the reference frequency If any one of the above three criteria is met, it is considered a precursor to flow instability.

[0086] S3: Short pulse de-icing mode (nanosecond level) When the system determines that there is excessive icing, it activates the nanosecond-level short-pulse de-icing mode. The specific steps are as follows: Step ①: Retrieve the ice thickness d, ice covering angle θ, ice thickness change rate dd / dt, and ambient temperature T parameters collected by the ice layer parameter sensing module 410 as control inputs; Step 2: Optimize the combination of control parameters for de-icing based on real-time icing parameters. The optimal parameter vector satisfies: ,in This is the discharge voltage. τ The duration of the pulse. f The pulse frequency, E It is a single pulse energy; The optimization solution in step ② uses the constrained Bayesian optimization method, with the surrogate model based on the current ice thickness. d Ice thickness change rate Ambient temperature T Using the input, predict different combinations of excitation parameters. U , τ , f , E De-icing time below t ice and energy consumption W ice The objective function is optimized as follows: The weight , Depend on Decision: Rapid freezing When greater than 0.3 mm / min 0.8 The value is 0.2, emphasizing de-icing speed; slow icing means... When not exceeding 0.3 mm / min 0.5 The value is 0.5, emphasizing energy saving.

[0087] Step 3: The system switches to the nanosecond pulse working module, with a pulse rise time ≤ 5μs, and adopts the BBM discharge mechanism; Step 4: Control the synchronous discharge of the plasma actuators corresponding to the icing areas on the surface of the aircraft to generate instantaneous high-energy plasma shock waves; Step 5: Relying on the instantaneous impact of the high-energy shock wave, the ice layer on the surface of the aircraft is rapidly peeled off and detached; Step 6: After the de-icing operation is completed, monitor the thickness of residual ice on the surface in real time. If the thickness of residual ice still exceeds the preset threshold, the system will automatically increase the discharge voltage. U Perform secondary de-icing operations; Step 7: When the monitored residual ice thickness meets the requirements When the de-icing operation is completed, the system will switch to either S4 flow control mode or S5 standby monitoring mode based on the real-time flow status. Step 8: Synchronously enter all operating parameters, monitoring data, and operation results of this de-icing operation into the system's operating database.

[0088] S3 internal closed-loop feedback logic: The residual ice thickness is monitored in real time throughout the de-icing process. If a single de-icing operation fails to meet the target, the system automatically iteratively optimizes the discharge voltage. U Pulse duration τ Pulse frequency f The parameters are adjusted, and the de-icing operation is re-executed. All operational data is fed back to the S1 status monitoring module and the system operation database in real time, and parameter optimization is continuously performed based on historical operational data; when the cumulative number of valid operational samples of the system meets the requirements... In this way, the control model is automatically retrained and the parameters are iteratively updated to improve the de-icing adaptability under complex icing conditions.

[0089] The voltage increment strategy for secondary de-icing is: initial discharge voltage U 0. Based on the optimization results of the proxy model, if the residual ice thickness after the first de-icing is... d res If it is greater than 0.1mm, then in U Increase the voltage by 10%~20% from the base voltage as the secondary de-icing voltage, while adjusting the pulse width. τ Extend the voltage by 10% to 20% to enhance the shock wave intensity. If the result is still unsatisfactory after the second de-icing, continue increasing the voltage in 20% increments until... d res Less than 0.1mm or U It reached the 25kV upper limit. U For 25kV, τ At the limiting parameter of 200ns, ice layers with a thickness of more than 3mm can be peeled off.

[0090] S4: Long pulse flow control mode (microsecond level).

[0091] When the system determines that there is no excessive icing but there is a risk of flow instability in the airflow boundary layer, it activates the microsecond-level long pulse flow control mode. The specific execution steps are as follows: Step ①: Retrieve Mach number (Ma) and angle of attack from the boundary layer flow sensing module 420. α The wall shear stress τ and the dynamic wall shear stress parameter are used as control inputs. Step 2: Optimize the flow control optimal parameter combination based on real-time airflow parameters. The optimal parameter vector satisfies: ; The optimization objective function in step ② is: ,in The ratio of the separated regions. For shear stress recovery ratio, Frequency efficiency ratio. Weights are preset according to flight phase: takeoff and climb. The value is 0.6, prioritizing separation suppression; during high-altitude cruise. The value is 0.6, prioritizing shear stress recovery; during landing and Both are 0.4, emphasizing overall balance. Constrained Bayesian optimization converges to the global optimum within 20-50 iterations.

[0092] Step 3: The system switches to the microsecond pulse working module, with a pulse response time ≤ 5μs, matching the BBM discharge mechanism; Step 4: Based on the distribution of gas flow instability zones, drive all or a designated area of ​​plasma actuators to continuously generate synthetic jets; Step 5: By continuously and controllably injecting momentum into the low-energy region of the airflow boundary layer through a synthetic jet, boundary layer separation is suppressed and the airflow state is improved; Step 6: Monitor the recovery state of wall shear stress τ and wall pressure pulsation p' in real time to determine the flow control effect; Step 7: Set an effect monitoring window of 1s to 5s. If the flow parameters do not recover to a stable and compliant state within the window, the system will re-complete the parameter optimization and update the control parameters, and then execute the flow control again. The stable compliance status is defined as: No more than 1.2 times and Not less than 0.85 times The monitoring window is set to 3 seconds by default. If the above two conditions are met and maintained simultaneously within 3 seconds, the flow control effect is considered to be satisfactory.

[0093] Step 8: Simultaneously enter the operating parameters, monitoring data, and control results of this flow control operation into the system's operating database.

[0094] S4 internal closed-loop feedback logic: If the discharge voltage is not met within the 1s~5s monitoring window during a single flow control operation, the system automatically iteratively updates the discharge voltage. U Pulse duration τ Pulse frequency f Parameters, restart the S4 flow control process.

[0095] S5: Standby / Lowest Frequency Monitoring Mode.

[0096] When the system monitors icing and airflow conditions and finds them to be stable and normal, it enters a low-power standby monitoring mode. The system maintains a microsecond pulse micro-excitation state to continuously perform online monitoring of the entire domain status with minimal power consumption, collects operating data in real time, and can switch to the corresponding operating mode at any time in response to sudden changes in operating conditions.

[0097] In standby mode, the power consumption is approximately 45W. During this mode, the FPGA and DSP remain operational, the sensor sensing layer 400 operates at a reduced frequency, with its sampling rate decreasing to 20%–30% of normal operating conditions, and the dual-mode high-voltage pulse control controller 200 is in a standby state. When the sensor sensing layer 400 detects any parameter exceeding a preset threshold, the system fully restores the sampling rate and enters the corresponding operating mode within the next 10ms cycle.

[0098] For the safety-priority control logic (highest priority), this system sets up a safety-priority emergency control logic, which has a higher priority than all regular operating modes: During the execution of the S4 flow control mode, if the ice layer parameter sensing module monitors the ice thickness in real time... d If the temperature suddenly increases and exceeds the preset threshold, the system immediately interrupts the S4 process and switches to the S3 short-pulse de-icing mode within 0.5ms. After the de-icing operation is completed (the standard de-icing time is about 30ms), the system automatically resumes the S4 flow control mode and updates the control parameters based on the latest operating conditions.

[0099] Among them, the total time for emergency switching under all operating conditions is ≤100ms, the fluctuation of the aircraft's lift coefficient during the switching and control process is <2%, and the core control logic of the system follows the safety principle of prioritizing icing suppression over flow instability suppression, so as to maximize the safety of the aircraft's flight.

[0100] S6: Multi-exciter coordinated modulation.

[0101] While executing the S4 flow control mode, the system simultaneously initiates multi-exciter cooperative modulation logic to apply differentiated phase delays to the plasma exciters at different spanwise positions of the aircraft. By using phase-coordinated modulation to form controllable traveling wave or standing wave excitations on the wall, precise control of the entire airflow boundary layer can be achieved.

[0102] Differential phase delay The allocation strategy is as follows: For the traveling wave excitation mode, the wave propagation direction is delayed sequentially according to the direction of the incoming flow, forming a wave propagation direction consistent with the mainstream direction, and propagating in the same direction as the natural disturbance of the boundary layer; For the standing wave excitation mode, the exciters on the left and right sides are applied with symmetrical phases with the spanwise center as the axis of symmetry, forming antinodes in the central region, i.e. the maximum momentum injection region, and nodes on both sides, i.e. the minimum momentum injection region, which is suitable for the working condition where the separation region is spanwise symmetrical.

[0103] All control data is fed back to the S1 status monitoring module and system operation database in real time, and dynamic optimization of control parameters is completed based on the continuously accumulated operation data; when the cumulative number of effective operation samples of the system meets the requirements... In this way, the control model is automatically retrained and parameters are updated, continuously improving the accuracy of flow regulation under multiple operating conditions.

[0104] Example 2: This application provides a method for dual-mode control of plasma synthesis jets using a system as described in any of the embodiments in Example 1, such as... Figure 6 As shown, the method includes: S10: Ice layer parameters are collected through ice layer parameter sensing module 410, and boundary layer flow parameters are collected through boundary layer flow parameter sensing module 420.

[0105] S20: Multi-objective optimization and collaborative control unit 300 judges operating conditions based on ice layer parameters and boundary layer flow parameters. S201: When the ice layer thickness At that time, it enters the nanosecond pulse de-icing mode, in which To set a thickness threshold; S202: When the ice layer thickness Furthermore, when the flow instability criterion is triggered, it enters the microsecond pulse flow control mode; S203: When the ice layer thickness If the flow instability criterion is not triggered, it enters standby mode.

[0106] S30: In nanosecond pulse de-icing mode, mode switching switch circuit 230 switches to nanosecond pulse generation module 210. Nanosecond pulse generation module 210 outputs nanosecond-level high-voltage pulse to plasma synthetic jet exciter array 100. Plasma synthetic jet exciter array 100 generates shock wave to peel off the ice layer.

[0107] In this mode, the nanosecond-level high-voltage pulse output by the nanosecond pulse generation module 210 causes the gas in the discharge cavity 112 to rapidly heat and pressurize within a timescale of hundreds of nanoseconds, forming a high-temperature, high-pressure plasma. When the pressure inside the cavity reaches a critical value, the high-pressure gas is ejected outward at high speed through the multi-slit jet outlet 115, forming a strong shock wave in front of the jet outlet 115. This shock wave propagates to the bottom of the ice layer at a speed higher than the speed of sound, generating a reflected stretching wave at the ice-skin interface, causing the ice layer to peel off in layers at the interface, rather than melting the ice layer by heat conduction. Therefore, the energy consumption is significantly lower than that of the electrothermal de-icing method.

[0108] In some embodiments, under nanosecond pulse de-icing mode, the multi-objective optimization collaborative control unit 300 invokes a Gaussian process regression surrogate model based on the current operating condition vector to perform constrained Bayesian optimization online search and determine the optimal excitation parameters. ,in This is the discharge voltage. For nanosecond pulse width, The repetition frequency, For a single energy input, The matrix is ​​transposed; the nanosecond pulse generation module 210 outputs a high-voltage pulse according to the optimal excitation parameters; during the de-icing process, the ice layer parameter sensing module 410 continuously monitors the change in ice layer thickness, and terminates the pulse output when the ice layer thickness drops below the residual thickness.

[0109] The residual thickness threshold is set to 0.08mm~0.1mm by default. The impact of this residual ice layer on aerodynamic performance is negligible, and its lift coefficient loss is less than 0.5%. At the same time, excessive de-icing is avoided, which leads to energy waste.

[0110] S40: In the microsecond pulse flow control mode, the mode switching switch circuit 230 switches to the microsecond pulse generation module 220. The microsecond pulse generation module 220 outputs a microsecond-level pulse sequence to the plasma synthetic jet exciter array 100, and the plasma synthetic jet exciter array 100 generates a pulse jet to suppress boundary layer separation.

[0111] In this mode, each microsecond pulse cycle comprises four stages: discharge heating, gas expansion and injection, intracavitary depressurization, and external gas backfilling, forming a periodic pulsed jet. The jet is injected into the low-energy region of the boundary layer at a normal or oblique angle, drawing the mainstream high-momentum fluid into the near-wall region through entrainment, thus enhancing the boundary layer's resistance to adverse pressure gradients. Since the repetition frequency of the microsecond pulse is on the same order of magnitude as the natural boundary layer perturbation frequency, typically ranging from hundreds to thousands of hertz, the synthesized jet can form favorable interference with the natural boundary layer perturbations, achieving flow separation suppression over a wide frequency range.

[0112] In some embodiments, under the microsecond pulse flow control mode, the multi-objective optimization collaborative control unit 300 invokes a Gaussian process regression surrogate model based on the current operating condition vector to perform constrained Bayesian optimization online search and determine the optimal excitation parameters. ,in This is the discharge voltage. For nanosecond pulse width, The repetition frequency, For a single energy input, The matrix is ​​transposed; the microsecond pulse generation module 220 outputs a microsecond-level pulse sequence according to the optimal excitation parameters; at the same time, the spanwise phase modulation module configures the phase delay according to the coordinates of the dual-mode compatible exciter 110 at different spanwise positions and the incoming flow conditions to form a traveling wave excitation mode or a standing wave excitation mode; during the flow control process, the boundary layer flow parameter sensing module 420 continuously monitors the control effect, and when the flow parameters meet the standard and are continuously set for a time, the repetition frequency is gradually reduced to the lowest effective frequency.

[0113] The minimum effective frequency refers to the lowest repetition frequency at which the flow control effect can still maintain a satisfactory state. The system gradually decreases f in 0.1 kHz steps while continuously monitoring. and If any parameter re-triggers the instability criterion, the system immediately reverts to the previous effective frequency and fixes it at the optimal frequency for that operating condition. This frequency reduction and energy-saving strategy can reduce the average power consumption of flow control by 30% to 40%. (Based on the initial frequency) f Taking 2.5kHz as an example, after gradual frequency reduction, the lowest effective frequency is 1.6kHz~1.8kHz, which is about 64%~72% of the initial frequency.

[0114] The timing relationship of each step in the above method is as follows: S10 is a continuously executed status monitoring step, running continuously with a period of approximately 10ms; S20 is a condition judgment step, triggered in real time based on the monitoring results of S10; S30 and S40 are mutually exclusive execution steps, switched between the two by the mode switching switch circuit 230, and the system can only be in one of the two modes at any given time; after S30 is completed and the ice threat is eliminated, the system can decide whether to switch to S40 based on the working condition judgment result of S20; after S40 is completed, the system returns to S10 to continue monitoring, forming a complete closed-loop control loop. Standby mode S50, i.e. Figure 5 S5, as the third judgment result of S20, is executed when neither freezing nor flow instability is triggered, and is the default operating state of the system.

[0115] Example 3: This embodiment uses a certain type of medium-high altitude long-endurance UAV as an example to illustrate the engineering deployment and complete workflow of the system proposed in this application. The UAV has a wingspan of 12m, a maximum takeoff weight of 800kg, a cruising Mach number of 0.4 to 0.6, and a service ceiling of 12km. After the upgrade and replacement scheme proposed in this application, the total system weight does not exceed 12kg, and the additional power consumption during cruise does not exceed 50W.

[0116] The engineering deployment and complete workflow of the plasma synthesis jet dual-mode control system provided in this embodiment specifically include the following steps S100 to S600.

[0117] Step S100, system deployment and startup initialization, includes the following steps S101 to S102.

[0118] Step S101: Install the exciter array and sensor.

[0119] Twenty-four dual-mode compatible actuators were installed along the spanwise direction of the wing leading edge. The section from the wing root to 50.5% of the wing span was a high-incidence area for icing, and the actuator spacing in this section was 35mm. The section from 50.5% of the wing span to the wingtip was the main flow control area, and the actuator spacing in this section was 60mm, covering a total spanwise length of 1.15m.

[0120] Each dual-mode compatible actuator has an alumina ceramic housing, which can withstand long-term operating temperatures up to 250℃. The discharge chamber volume is 45mm². 3 The upper surface has six parallel laser-cut straight slits, each 0.6 mm wide and 22 mm long, with an aspect ratio of approximately 37:1. The electrode spacing is 4.5 mm, and the electrodes are made of tungsten-copper alloy. After installation, the skin subsidence will not exceed 0.2 mm. A Pt100 thin-film heating element is embedded in the outer wall of the cavity for preheating the discharge cavity in extreme low-temperature environments.

[0121] Twenty-three fiber Bragg grating icing sensors are arranged in the spanwise gap of the actuator. These sensors have a measurement accuracy of 0.02 mm and a response time of 5 ms. Twelve high-frequency dynamic pressure sensors and six wall shear stress sensors are arranged at 50 mm and 80 mm downstream of the actuator, respectively. The natural frequency of the high-frequency dynamic pressure sensors is no less than 500 kHz, and the resolution of the wall shear stress sensors is 0.05 Pa. All sensor signals are synchronously acquired by a 24-bit Σ-Δ analog-to-digital converter and then sent to the FPGA.

[0122] Step S102: Power supply and control unit power-on self-test.

[0123] The nanosecond pulse module is a 6-stage solid-state Marx generator using SiC MOSFET switches. Its output voltage is adjustable from 0 to 24 kV, pulse width from 20 ns to 200 ns, and single-shot energy from 15 mJ to 60 mJ. The microsecond pulse module uses an IGBT half-bridge RLC resonant charging topology. Its output voltage is adjustable from 0 to 10 kV, pulse width from 1 μs to 20 μs, and repetition frequency from 0.5 kHz to 5 kHz. After both modules are powered on, the FPGA hardware state machine verifies that the BBM interlock circuit is working normally and the dead time configuration is correct (default dead time is 2 μs). The DSP processor is a TMS320F28388D, a dual-core processor running at 200 MHz. It loads an offline-trained Gaussian process regression surrogate model from the on-chip Flash, performs CRC verification, and then enters standby monitoring mode. The standby power consumption is approximately 45W. The system sends a heartbeat signal to the flight control computer at a frequency of 1 Hz via the CAN FD bus.

[0124] Step S200: Dual-source sensor monitoring and condition judgment under normal cruise conditions.

[0125] The system performs a cycle of sensor data acquisition and operational condition assessment approximately every 10ms. The fiber optic demodulator reads 23 channels of wavelength data at a sampling rate of 200Hz, which is then converted into ice thickness at each measuring point after moving average filtering. The maximum value among all measuring points is taken as the current ice thickness. Simultaneous calculation of spanwise coverage θ and freezing rate The FPGA acquires pressure and shear stress signals at a sampling rate of 1kSPS, performs real-time 512-point FFT spectrum analysis on each channel, extracts the root mean square value of pressure pulsation and the dominant frequency, and compares them with the offline calibrated benchmark values. The DSP processor polls the preprocessed sensor data every 10ms and determines the operating condition according to priority. First, if d ≥ d threshold This triggers the nanosecond pulse de-icing mode. d threshold The default value is 0.5mm, and this threshold can be adjusted online by the flight control computer during flight.

[0126] Second, if the current ice thickness d < d threshold However, if the flow instability criterion is triggered, the microsecond pulse flow control mode is activated.

[0127] Third, if neither of the above two conditions is triggered, the system will remain in standby monitoring mode.

[0128] It should be noted that the ice layer sensor data has unconditional priority. When the ice layer thickness exceeds the threshold, the system will force the system to enter the de-icing mode regardless of the flow state.

[0129] Step S300, the nanosecond pulse de-icing mode workflow, includes the following steps S301 to S303.

[0130] Step S301: Triggering and parameter optimization.

[0131] The FPGA generates a hardware interrupt within 5μs after the ice layer data is updated. The DSP processor saves a snapshot of the current system state and reads the operating condition vector. The working condition vector includes the current ice thickness. d , Expansion coverage , freezing rate Ambient temperature ,Mach number Angle of attack Wall pressure pulsation and wall shear stress The DSP invokes a Gaussian process regression surrogate model to perform constrained Bayesian optimization online search.

[0132] The objective function for optimization is ,in This refers to the de-icing time. To reduce de-icing energy consumption, and These are the baseline values ​​and weights. and Automatic selection based on icing rate: Under rapid icing conditions, i.e., when the icing rate is greater than 0.3 mm / min, The value is 0.8. A value of 0.2 is used to emphasize de-icing speed; under slow icing conditions, i.e., when the icing rate is no greater than 0.3 mm / min, The value is 0.5. A value of 0.5 is used to prioritize energy efficiency. It typically converges after 20-50 iterations, taking approximately 2ms-10ms, and outputs the optimal excitation parameter vector. .

[0133] Step S302: Mode switching and pulse execution.

[0134] The FPGA executes BBM to switch to the nanosecond pulse module. The switching timing is as follows: first, the current module is turned off within no more than 100ns; then, after a 2μs dead time, complete turn-off is confirmed; subsequently, it is pre-charged to the target bus voltage. V bus Equal to the optimal discharge voltage U optFinally, the output is enabled, with a total time of approximately 3μs~4μs. The FPGA operates according to the optimal excitation parameters. Generate a trigger pulse sequence to drive the nanosecond pulse module.

[0135] During the de-icing process, the ice layer sensor continuously monitors the ice thickness at a sampling rate of 200Hz. d Value change. The DSP processor evaluates the de-icing progress every 5ms: if the ice thickness... d If the ice thickness continues to decrease, the current parameters will remain unchanged; if the rate of decrease in ice thickness is lower than expected, a fast re-optimization will be triggered within 5ms, and the Bayesian optimization will be limited to 10 iterations; if the ice thickness d drops below 0.08mm, the pulse output will be terminated immediately.

[0136] Step S303: Complete confirmation and data recording.

[0137] After de-icing is completed, three consecutive confirmation samples are taken, with each sample interval being 10ms. De-icing is considered successful if the ice thickness d value obtained from all three samples is less than 0.08mm. The working condition vector and optimal excitation parameters are then used to confirm the process. Actual de-icing time t ice Actual measured de-icing energy consumption Wice Data such as residual ice thickness are stored in the operating database. The system decides whether to return to standby mode or resume flow control mode based on the saved state snapshot.

[0138] Step S400, the workflow of the microsecond pulse flow control mode, includes the following steps S401 to S403.

[0139] Step S401: Instability confirmation and parameter optimization.

[0140] The DSP processor confirms that the instability criterion has been triggered within three consecutive sampling cycles to avoid false triggering; this confirmation process is completed within 30ms. Subsequently, the operating condition vector is read. ,in Mach number, For the angle of attack, For wall pressure, For wall shear stress, This represents the root mean square value of the pressure pulsation. This is the root mean square reference value for pressure pulsation under steady-state conditions. This is the average value when the wall shear stress is present. This is the reference value for wall shear stress under non-separation conditions. The frequency is a characteristic frequency of the pressure pulsation spectrum. The DSP calls a Gaussian process regression surrogate model to perform constrained Bayesian optimization.

[0141] The objective function for optimization is ,in The area of ​​the separation region, The reference value for the area of ​​the separation zone. For shear stress recovery ratio, For current flow control efficiency, This is the baseline value for flow control efficiency. Used to characterize the degree of deviation of the current efficiency from the benchmark value. Weights are preset according to flight phase: takeoff and climb phase. A value of 0.6 is used, prioritizing separation suppression; during the high-altitude cruise phase. A value of 0.6 is used, prioritizing shear stress recovery; during the landing phase. and All values ​​are set to 0.4, emphasizing overall balance. The optimal excitation parameter vector is output. ,in The optimal discharge voltage is... To achieve the optimal microsecond pulse width, To achieve the optimal repetition frequency, This represents the optimal single-pass energy.

[0142] Step S402: Pulse output and phase modulation.

[0143] The FPGA executes BBM switching to the microsecond pulse module, according to the optimal excitation parameters. A microsecond pulse sequence is generated. Simultaneously, the DSP processor configures the phase delay of each exciter based on the incoming Mach number Ma and the spanwise characteristics of the separation region. ,in For the first i spanwise coordinates of an exciter For the target excitation frequency, The perturbation phase velocity forms a traveling wave excitation mode or a standing wave excitation mode in the spanwise direction, with a phase resolution of 10 ns.

[0144] Step S403: Effect monitoring and adaptive adjustment.

[0145] The FPGA continuously acquires sensor data at a sampling rate of 1kSPS, evaluating the control effect every second. When the control effect meets the target, i.e. and After 5 seconds, the system initiates a frequency reduction and energy-saving strategy: gradually reducing the repetition frequency f to the lowest effective frequency in 0.1kHz steps, typically 60%–70% of the initial frequency. If the control effect is insufficient, a fast re-optimization is triggered, limited to 15 iterations. Once the flow returns to normal, the system automatically exits the flow control mode and returns to standby. After each flow control event is completed, the operation record is stored in the database.

[0146] Step S500: Sudden freezing dual-mode preemptive switching workflow.

[0147] When the system encounters icing during microsecond pulse flow control, i.e. In such cases, the emergency response will be automatically completed according to the following preemptive procedure: Step S501: Preemption triggered, record the current time as T = 0.

[0148] The fiber optic demodulator detected the ice thickness within 5ms. d If the set threshold is exceeded, the FPGA generates a highest-priority interrupt with a latency of less than 0.5ms. The DSP processor terminates all current tasks and saves a snapshot of the flow control state to a dedicated snapshot buffer. This snapshot contains the current optimal flow control parameters. X flow Information such as FFT spectrum data, shear stress mean value, and phase configuration table are displayed, and an emergency de-icing flag is set.

[0149] Step S502: BBM switching, at this stage T It takes approximately 0.5ms to 5ms.

[0150] The FPGA executes a microsecond pulse module shutdown, with a shutdown time not exceeding 100 ns. After a 2 μs dead zone confirmation, the nanosecond pulse module precharges for 1 ms to 2 ms, and then enables the output, with a total time of approximately 3 ms to 5 ms. Simultaneously, the DSP processor utilizes the precharge time window, using the historical best parameter average as the initial guess, and performs 15 constrained Bayesian optimization iterations to obtain the optimal de-icing parameters within approximately 3 ms. X opt_ice .

[0151] Step S503: De-icing is performed. The time T during this stage is approximately 5ms to 50ms.

[0152] Nanosecond pulse module according to optimal de-icing parameters X opt_ice Outputs a high-voltage pulse. The ice sensor sampling rate is increased to 500Hz to accelerate response. Pulse output terminates when the ice thickness d drops below 0.08mm. Typical de-icing time, corresponding to an ice thickness of 1mm, is approximately 30ms to 50ms.

[0153] Step S504: Restore flow control. The time T during this stage is approximately 50ms to 100ms.

[0154] The DSP processor reads the flow control state snapshot from the snapshot. X flow As an initial guess, constrained Bayesian optimization only requires 5 to 10 iterations, taking approximately 2 to 5 ms, to converge to the optimal flow control parameters. X opt_flowThe FPGA executes the BBM switchback to the microsecond pulse module, with a switching time of approximately 3ms~4ms, restoring pulse output and phase modulation configuration. The aircraft's lift coefficient is measured throughout the entire process. C L The fluctuation is less than 2%. When the lift coefficient fluctuation exceeds the warning threshold, the system sends a warning to the flight control computer via the CAN bus.

[0155] The total time for the above preemptive full process is no more than 100ms, of which mode switching is no more than 5ms, de-icing is 30ms~50ms, and flow control recovery is no more than 20ms, ensuring the unmanned platform's ability to autonomously cope with sudden icing.

[0156] Step S600: Online learning and parameter updating of the agent model.

[0157] After each control event is completed, the DSP processor appends the execution log to the execution database. The execution log includes flight condition vectors, sensor state vectors, and actual excitation parameters. X actual The database contains measured control performance indicators and timestamps. It utilizes a 16MB SPIFlash W25Q128 memory, stored in a circular overlay configuration, capable of storing approximately 800 records.

[0158] The number of new valid records added since the last update N new When the value is ≥50, the DSP background task calls the sparse Gaussian process regression incremental update algorithm, updating only the position of the Gaussian process regression induced point and the variational parameters. This background task runs on the DSP's CPU2 core, with a lower priority than the CPU1 control task. A single update takes no more than 30 seconds, and the foreground control task is unaffected.

[0159] After the update is complete, calculate the normalized root mean square error (RMSE) of the most recent 20 data sets. If this error is at least 10% lower than the current model, write the new model to the on-chip Flash memory and enable it in the next control event; otherwise, retain the current model and continue accumulating until the current model is reached. N new When the value reaches 100, an update is attempted again. The running data shows that after 3-5 incremental updates, the model's predicted normalized root mean square error gradually decreased from the initial 12% to below 5%, and the number of convergence iterations for constrained Bayesian optimization decreased from approximately 32 to approximately 22.

[0160] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A dual-mode control system for plasma synthesis jets, characterized in that, include: A plasma-synthetic jet actuator array (100) consists of N dual-mode compatible actuators (110) arranged spanwise along the aerodynamic surface of the aircraft. Each dual-mode compatible actuator (110) has a sealed discharge cavity and a jet outlet communicating with the discharge cavity. The dual-mode compatible actuator (110) is configured to generate a shock wave for de-icing under nanosecond pulse drive and a pulsed jet for flow control under microsecond pulse drive. A dual-mode high-voltage pulse control controller (200) is electrically connected to the plasma synthesis jet exciter array (100). The dual-mode high-voltage pulse control controller (200) integrates a nanosecond pulse generation module (210) and a microsecond pulse generation module (220), and switches between the nanosecond pulse generation module (210) and the microsecond pulse generation module (220) through a mode switching switch circuit (230). A multi-objective optimization and collaborative control unit (300) is communicatively connected to the dual-mode high-voltage pulse control controller (200). The multi-objective optimization and collaborative control unit (300) is configured to select the working mode and determine the excitation parameters based on real-time operating condition information. The sensor sensing layer (400) is communicatively connected to the multi-objective optimization and collaborative control unit (300). The sensor sensing layer (400) includes an ice layer parameter sensing module (410) and a boundary layer flow parameter sensing module (420).

2. The dual-mode control system for plasma synthesis jets according to claim 1, characterized in that, The dual-mode compatible exciter (110) includes: The actuator body (111) is made of alumina ceramic and has a block structure. The bottom of the actuator body (111) is provided with a mounting flange (119). The mounting flange (119) is used to install flush with the aircraft skin. After installation, the skin sinks by ≤0.2mm. The actuator body (111) has a sealed discharge cavity (112) inside, with electrodes inserted at both ends and a jet outlet on the upper surface. The discharge cavity (112) is a cylindrical through-hole structure opened inside the actuator body (111), with a diameter of 2mm~6mm and a volume of 10mm². 3 ~80mm 3 ; The cathode discharge electrode (113) and the anode discharge electrode (114) are arranged in pairs facing each other, and are inserted laterally into the discharge cavity (112) from the left and right ends of the exciter body (111), with their tips facing each other and the electrode spacing being 2mm~8mm; the cathode discharge electrode (113) is made of tungsten copper alloy and is plated with gold or platinum, with a frustum-shaped end and the outer end connected to the cathode lead (116); the anode discharge electrode (114) is made of tungsten copper alloy and is plated with gold or platinum, with a hemispherical end and an axial thread adjustment structure, and the outer end connected to the anode lead (117); both the cathode lead (116) and the anode lead (117) are led out to the outside of the dual-mode compatible exciter (110) for connection to a high-voltage pulse power supply; A multi-slit jet outlet (115) is located on the upper surface of the actuator body (111) and its lower end is connected to the interior of the discharge cavity (112). It consists of multiple parallel slits, with 3 to 12 slits. The width of a single slit is 0.4 mm to 0.8 mm, the center-to-center distance between adjacent slits is 1.5 mm to 3 mm, and the ratio of slit length to slit width is ≥10:

1. The high-pressure gas inside the discharge cavity (112) is ejected outward through the multi-slit jet outlet (115). A micro-thin film heating element (118) is embedded inside the exciter body (111) near the outer wall of the discharge cavity (112) and is used to preheat the discharge cavity (112) in an extreme low temperature environment.

3. The dual-mode control system for plasma synthesis jets according to claim 1, characterized in that, The nanosecond pulse generation module (210) outputs a nanosecond-level high-voltage pulse with a pulse width of 20ns~200ns, a voltage of 8kV~25kV, and a single-shot energy of 15mJ~60mJ; the microsecond pulse generation module (220) outputs a microsecond-level pulse sequence with a pulse width of 1μs~20μs, a voltage of 2kV~10kV, a repetition frequency of 500Hz~5kHz, and a single-shot energy of 0.5mJ~5mJ.

4. The dual-mode control system for plasma synthesis jets according to claim 1, characterized in that, The ice layer parameter sensing module (410) is selected from fiber optic grating icing sensor or magnetostrictive icing sensor, and outputs ice layer thickness, spanwise coverage and icing rate in real time; the boundary layer flow parameter sensing module (420) includes a high-frequency dynamic pressure sensor array and a wall shear stress sensor array, which are arranged 20mm~100mm downstream of the dual-mode compatible exciter (110) on the wall surface. The multi-objective optimization and collaborative control unit (300) includes a DSP processor and an FPGA. The FPGA receives the collected signals from the ice layer parameter sensing module (410) and the boundary layer flow parameter sensing module (420) through multi-channel synchronous sampling. It performs real-time FFT spectrum analysis on the signals of each channel of the boundary layer flow parameter sensing module (420) and extracts the root mean square value of pressure pulsation, characteristic frequency offset and wall shear stress attenuation ratio as flow instability criteria. The DSP processor polls the sensor data preprocessed by the FPGA at a fixed period, performs working condition determination according to the ice layer parameters and the flow instability criteria, and sends the determination result and optimized excitation parameters to the dual-mode high-pressure pulse control controller (200).

5. The dual-mode control system for plasma synthesis jets according to claim 1, characterized in that, It also includes a spanwise phase modulation module, which is communicatively connected to the multi-objective optimization and collaborative control unit (300). In the microsecond pulse flow control mode, the multi-objective optimization and collaborative control unit (300) calculates the target phase velocity based on the incoming Mach number and the spanwise characteristics of the boundary layer separation region. The spanwise phase modulation module then calculates the target phase velocity according to the formula... For different spanning positions The dual-mode compatible exciter (110) applies a programmable phase delay. ,in For the excitation frequency, The perturbation phase velocity is used to form a traveling wave excitation mode or a standing wave excitation mode in the spanwise direction; the traveling wave excitation mode is coupled with the natural perturbation mode in the boundary layer, and the phase resolution of the spanwise phase modulation module is ≤10ns.

6. The dual-mode control system for plasma synthesis jets according to claim 1, characterized in that, The multi-objective optimization collaborative control unit (300) has a built-in Gaussian process regression surrogate model trained offline. The Gaussian process regression surrogate model takes the current operating condition vector as input and predicts the control effect under different combinations of excitation parameters. The multi-objective optimization collaborative control unit (300) converges to the globally optimal excitation parameters within 20 to 50 iterations through constrained Bayesian optimization. The multi-objective optimization collaborative control unit (300) also has online self-evolution capability. After each control event is completed, the triplet of operating condition parameters, actual excitation parameters and measured control effect is stored in the running database. When the cumulative number of newly added effective data reaches the set number, the incremental update of the surrogate model is triggered. After the update is completed, the prediction error is gradually reduced to below 5%, realizing intelligent control that adaptively evolves with the expansion of the flight envelope.

7. The dual-mode control system for plasma synthesis jets according to claim 1, characterized in that, The multi-objective optimization and collaborative control unit (300) is configured with preemptive priority management logic, with icing threat having a higher priority than flow instability: when the system is in microsecond pulse flow control mode, if the ice layer parameter sensing module (410) detects that the ice layer thickness exceeds the set threshold, the multi-objective optimization and collaborative control unit (300) responds with an interruption delay of ≤0.5ms and executes the following sequence in sequence: saves a snapshot of the current flow control state, cuts off the output of the microsecond pulse generation module (220), the mode switching switch circuit (230) switches to the nanosecond pulse generation module (210), calls the de-icing parameter optimization, performs nanosecond pulse de-icing until the ice layer thickness drops below the residual thickness for confirmation, quickly re-optimizes using the saved snapshot as the initial guess, switches back to the microsecond pulse generation module (220) and restores microsecond pulse flow control; the entire process takes ≤100ms and the lift coefficient fluctuation is <2%.

8. A method for dual-mode control of plasma synthesis jets using the system described in any one of claims 1 to 7, characterized in that, The method includes: Ice layer parameters are collected through the ice layer parameter sensing module (410), and boundary layer flow parameters are collected through the boundary layer flow parameter sensing module (420). The multi-objective optimization and collaborative control unit (300) makes a condition judgment based on the ice layer parameters and the boundary layer flow parameters: When the ice layer thickness At that time, it enters the nanosecond pulse de-icing mode, in which To set a thickness threshold; When the ice layer thickness Furthermore, when the flow instability criterion is triggered, it enters the microsecond pulse flow control mode; When the ice layer thickness If the flow instability criterion is not triggered, it enters standby mode; In the nanosecond pulse de-icing mode, the mode switching switch circuit (230) switches to the nanosecond pulse generation module (210), the nanosecond pulse generation module (210) outputs nanosecond-level high-voltage pulses to the plasma synthetic jet exciter array (100), and the plasma synthetic jet exciter array (100) generates shock waves to peel off the ice layer. In the microsecond pulse flow control mode, the mode switching switch circuit (230) switches to the microsecond pulse generation module (220), the microsecond pulse generation module (220) outputs a microsecond-level pulse sequence to the plasma synthetic jet exciter array (100), and the plasma synthetic jet exciter array (100) generates pulse jets to suppress boundary layer separation.

9. The dual-mode control method for plasma synthesis jet according to claim 8, characterized in that, In nanosecond pulse de-icing mode, the multi-objective optimization collaborative control unit (300) calls the Gaussian process regression surrogate model to perform constrained Bayesian optimization online search based on the current operating condition vector to determine the optimal excitation parameters. ,in This is the discharge voltage. For nanosecond pulse width, The repetition frequency, For a single energy input, The matrix is ​​transposed; the nanosecond pulse generation module (210) outputs a high-voltage pulse according to the optimal excitation parameters; during the de-icing process, the ice layer parameter sensing module (410) continuously monitors the change in ice layer thickness, and terminates the pulse output when the ice layer thickness drops below the residual thickness.

10. The dual-mode control method for plasma synthesis jet according to claim 8, characterized in that, In the microsecond pulse flow control mode, the multi-objective optimization collaborative control unit (300) calls the Gaussian process regression surrogate model to perform constrained Bayesian optimization online search based on the current operating condition vector to determine the optimal excitation parameters. ,in This is the discharge voltage. For nanosecond pulse width, The repetition frequency, For a single energy input, The matrix is ​​transposed; the microsecond pulse generation module (220) outputs a microsecond-level pulse sequence according to the optimal excitation parameters; at the same time, the spanwise phase modulation module configures the phase delay according to the coordinates of the dual-mode compatible exciter (110) at different spanwise positions and the incoming flow conditions to form a traveling wave excitation mode or a standing wave excitation mode; during the flow control process, the boundary layer flow parameter sensing module (420) continuously monitors the control effect, and when the flow parameters meet the standard and the set time is continuously set, the repetition frequency is gradually reduced to the lowest effective frequency.

Images