An adaptive shock wave augmentation method and system

Abstract

The application relates to the technical field of knock combustion control, and discloses a self-adaptive knock wave enhancement method and system, which comprises the following steps: acquiring multiple-source physical quantities of a combustion chamber, extracting a knock wave wave front position, a peak pressure, an actual propagation speed and a characteristic temperature parameter; calculating an overdrive parameter based on the actual propagation speed, combining the wave front position to generate a basic control instruction to a fuel injection or an expansion wall surface adjustment system; inputting the physical parameters into a fuzzy control algorithm sequence to solve an output comprehensive state evaluation factor; extracting evaluation factors of a plurality of continuous time steps to construct a time sliding window, inputting a neural network algorithm sequence to calculate a predicted state evaluation factor of a next time step; and generating a parameter update instruction according to a prediction deviation amount to adjust fuel injection parameters and expansion wall surface angles. The application combines fuzzy control and neural networks to realize advanced prediction and self-adaptive closed-loop control of the evolution state of a knock wave, and improves operation stability and control reliability.

Description

Technical Field

[0001] This invention relates to the field of detonation combustion control technology, and in particular to an adaptive detonation wave enhancement method and system. Background Technology

[0002] Scramjet engines have application potential under high Mach number flight conditions and have become a major development direction for hypersonic vehicle propulsion systems. However, existing scramjet engine demonstrators suffer from limited acceleration performance and low net thrust in actual operation. Traditional scramjet engines mostly employ isobaric combustion cycle designs, whose thermodynamic cycle efficiency is significantly lower than that of near-isochoric detonation combustion. To overcome the thrust performance bottleneck, introducing local or global detonation combustion in supersonic airflow has become an effective way to improve the engine's thermodynamic cycle efficiency.

[0003] Although detonation combustion possesses a high heat release rate, in practical combustion chamber applications, the propagation of the detonation wave is highly susceptible to interference from non-ideal factors in the flow field environment. Fluid boundary layers, lateral expansion, and spatial curvature effects continuously consume the energy behind the wave, leading to a decrease in the detonation wave propagation speed and a loss in peak pressure. In severe cases, this can cause decoupling between the shock wave and the combustion surface, or even detonation failure. Current research on detonation combustion and thermal jet initiation largely relies on basic experimental pipelines that independently supply fuel and oxidizer, failing to effectively integrate with the highly dynamic and nonlinear supersonic flow field environment inside a real engine, and lacking systematic control schemes for engineering applications.

[0004] More significantly, existing combustion control methods lack real-time monitoring and intelligent closed-loop adjustment capabilities for the propagation state of detonation waves. Under the complex operating conditions of scramjet engines, it is impossible to detect the forward or backward trends of detonation waves in a timely manner based on multi-source physical quantity data within the flow field. Due to the lack of a systematic adaptive intervention mechanism, the system cannot dynamically optimize fuel injection parameters or automatically adjust the geometric expansion angle of the combustion chamber transition section before the detonation wave becomes unstable and decays. This lag and simplistic control approach results in the engine's inability to effectively suppress the weakening effect of flow field boundary losses and expansion on the detonation wave, making it difficult to maintain the self-sustaining stability of the detonation wave within the predetermined space, ultimately causing propulsion system failure. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive detonation wave enhancement method and system, which solves the problem that scramjet engines in complex airflow environments are prone to decoupling instability or even detonation due to the lack of advanced prediction and intelligent closed-loop intervention mechanism for the evolution of detonation waves.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The first aspect of this invention provides an adaptive detonation wave enhancement method, comprising the following steps:

[0008] The multi-source physical quantities of the scramjet engine combustion chamber are obtained, and the detonation wavefront position, peak pressure parameter, actual propagation velocity parameter, and characteristic temperature parameter of the combustion chamber at the current time step are extracted.

[0009] The overdrive parameter is calculated based on the actual propagation speed parameter. Based on the overdrive parameter and the position of the detonation wave front, basic control commands are generated and sent to the fuel injection system or the expansion wall adjustment and control system.

[0010] The peak pressure parameter of the knock wave, the actual propagation velocity parameter, and the characteristic temperature parameter of the combustion chamber are input into the fuzzy control algorithm sequence for calculation, and the comprehensive state evaluation factor is output.

[0011] Extract the comprehensive state evaluation factors from multiple consecutive time steps, construct a time sliding window, and input it into a neural network algorithm sequence to calculate and output the predicted state evaluation factors for the next time step.

[0012] Based on the deviation between the predicted state assessment factor and the reference benchmark, parameter update instructions are generated and sent to the fuel injection system and the expansion wall adjustment control system to adjust the fuel injection parameters and the expansion wall angle.

[0013] In one feasible implementation, the multi-source physical quantities include a spatial pressure distribution sequence, the optical observation propagation velocity output by a high-speed camera system, and a temperature field distribution sequence. The specific process of acquiring the multi-source physical quantities of the scramjet engine combustion chamber and extracting parameters for the current time step includes: locating the axial physical coordinates corresponding to the largest value in the spatial pressure distribution sequence as the detonation wavefront position, and extracting the corresponding absolute pressure value as the peak pressure parameter of the detonation wave; calculating the quotient of the spatial difference between the current and previous time step detonation wavefront positions and the time step, and performing an arithmetic average with the optical observation propagation velocity output by the high-speed camera system to generate the actual propagation velocity parameter; using the detonation wavefront position as the target input value, performing a spatial interpolation algorithm in the temperature field distribution sequence to calculate the temperature value at the corresponding coordinates as the combustion chamber characteristic temperature parameter.

[0014] In one feasible implementation, the specific process of calculating the overdrive parameter based on the actual propagation velocity parameter and generating basic control commands based on the overdrive parameter and the detonation wavefront position includes: calculating the dimensionless relative ratio of the actual propagation velocity parameter to the preset ideal Chapman-Jouguet detonation wave velocity constant as the overdrive parameter; when the overdrive parameter is less than a preset overdrive critical threshold and the detonation wavefront position reaches a preset detonation wave backpropagation threshold position coordinate, generating a first basic control command and sending it to the fuel injection system; when the overdrive parameter is greater than or equal to the overdrive critical threshold and the detonation wavefront position reaches a preset detonation wave forward propagation threshold position coordinate in the memory, generating a second basic control command and sending it to the expansion wall adjustment control system; when the detonation wavefront position is between the detonation wave forward propagation threshold position coordinate and the detonation wave backpropagation threshold position coordinate, determining that the detonation wave is in a stationary state, and generating a maintenance command to maintain the current basic physical state of the fuel injection system and the expansion wall adjustment control system.

[0015] The first basic control command is used to drive the injection orifice opening adjustment valve of the fuel injection system to increase the physical flow area and drive the injection pressure adjustment valve to increase the absolute pressure in the fuel injection pipeline; the second basic control command is used to drive the motor drive system of the expansion wall adjustment control system to output rotational displacement, causing the expansion wall to rotate around the movable joint in a direction away from the center line of the combustion chamber.

[0016] The technical principle behind the above process lies in obtaining overdrive parameters characterizing the detonation wave state through multi-source physical quantity calculations. When the detonation wave is in an underdriven state and exhibits a backward propagation trend, the control system increases fuel supply through a first basic control command to improve the combustion chamber equivalence ratio and prevent detonation wave attenuation and flameout. When the detonation wave is in a strong overdriven state and exhibits forward propagation, the control system increases the physical expansion angle of the combustion chamber transition section through a second basic control command to reduce the back pressure of the flow field and prevent the detonation wave from being swallowed into the isolation section. The system constructs a dual-channel, low-level safety intervention mechanism for the location and intensity of the detonation wave through the above logic.

[0017] In one feasible implementation, the specific process of inputting a fuzzy control algorithm sequence, solving it, and outputting a comprehensive state evaluation factor includes: calculating the membership floating-point values ​​of the three parameters in fuzzy sets in different numerical intervals to generate an input fuzzy set vector; substituting the input fuzzy set vector into the antecedent of the fuzzy control rule base to calculate the activation intensity, and performing a logical intersection operation with the corresponding rule consequent fuzzy set to generate a fuzzy subset of the rule output results; performing a logical union operation on the fuzzy subset of the rule output results that generate activation responses to generate a total output fuzzy set; and using the discrete centroid method to perform multiplication-addition-integral operations on the total output fuzzy set at discrete sampling points in the output universe to solve and generate the comprehensive state evaluation factor.

[0018] By using fuzzy control logic to perform nonlinear mapping and fusion on multidimensional heterogeneous physical parameters, a single-dimensional state evaluation factor is output, which is convenient for subsequent algorithm processing.

[0019] In one feasible implementation, the specific process of extracting comprehensive state evaluation factors into the neural network algorithm sequence and predicting the state evaluation factors for the next time step includes: reading comprehensive state evaluation factors from a consecutive set integer number of time steps in chronological order, constructing an input feature column vector in column vector form; performing matrix multiplication on the weight matrix from the input layer to the hidden layer and the input feature column vector, performing matrix addition on the result and the bias vector, and substituting it into the first nonlinear activation function to generate the hidden layer output state vector; subsequently performing matrix multiplication on the weight matrix from the hidden layer to the output layer and the hidden layer output state vector, and adding the output bias scalar; substituting the result into the second nonlinear activation function for numerical range-limited calculation, generating the predicted state evaluation factors for the next time step. The above process utilizes a neural network algorithm combined with a sliding time window data structure to mine the time series features of historical state evaluation factors, thereby outputting the combustion state evolution results for future time steps in advance.

[0020] In one feasible implementation, the specific process of generating parameter update instructions includes: using the nozzle opening degree, fuel injection pressure, and expansion wall angle of the current time step as iterative reference values; substituting the deviations into the opening degree control mapping function, pressure control mapping function, and angle control mapping function respectively, and multiplying them by the corresponding adaptive learning gain coefficients to generate opening degree compensation values, pressure compensation values, and angle compensation values; adding the iterative reference values ​​to the corresponding compensation values ​​to generate nozzle opening degree update instructions, fuel injection pressure update instructions, and expansion wall angle update instructions for the next time step. By superimposing deviation compensations, the system constitutes a closed-loop adaptive iterative adjustment architecture.

[0021] In one feasible implementation, before acquiring multi-source physical quantities, the system performs an initialization setting step: loading pre-calibrated fuzzy control inference rules into the memory, establishing a floating-point storage array of weight matrix and bias vector for the neural network algorithm sequence, and writing Gaussian distribution values ​​that follow the set mean and variance; driving the regulating valve of the fuel injection system to the initial open state and initial fuel injection pressure, and driving the motor drive system of the expansion wall adjustment control system to rotate to the set initial expansion angle.

[0022] A second aspect of the present invention provides an adaptive detonation wave enhancement system, the system comprising:

[0023] An intelligent sensor monitoring system is installed in the combustion chamber of the scramjet engine to simultaneously acquire multiple physical quantities.

[0024] The control algorithm unit communicates with the intelligent sensor monitoring system and includes a hardware processor and a memory with a built-in information analysis module. The hardware processor reads and executes the information analysis module to: receive multi-source physical quantities, perform calculations, and output parameter update instructions.

[0025] The fuel injection system is located at the front end of the combustion chamber and is electrically connected to the control algorithm unit. It receives parameter update instructions and is used to adjust the state of the fuel fluid entering the combustion chamber.

[0026] The expansion wall adjustment control system includes an expansion wall disposed inside the combustion chamber and a motor drive system. The motor drive system is electrically connected to the control algorithm unit to receive parameter update commands and drive the expansion wall to rotate, thereby adjusting the physical expansion angle of the combustion chamber transition section.

[0027] In summary, the present invention has at least one of the following beneficial technical effects:

[0028] 1. This invention acquires multi-source physical quantities such as pressure, velocity, and temperature in the combustion chamber of a scramjet engine, extracts the detonation wavefront position, peak pressure, and actual propagation velocity, and inputs these heterogeneous physical parameters into a fuzzy control algorithm sequence for nonlinear mapping, outputting a single-dimensional comprehensive state evaluation factor. This transforms complex dynamic flow field characteristics into calculable quantitative indicators, solving the problem that single-dimensional physical parameters cannot accurately reflect the comprehensive evolution state of the detonation wave, and establishing a real-time evaluation system for the combustion chamber airflow environment and detonation wave intensity.

[0029] 2. This invention extracts comprehensive state evaluation factors from multiple consecutive time steps to construct a time sliding window, and inputs it into a neural network algorithm sequence to calculate the predicted state evaluation factors for the next time step. Subsequently, compensation values ​​are generated based on the prediction deviation and superimposed on the iterative baseline value. This logic utilizes historical time series features to mine the evolution patterns of the detonation wave state. Before the detonation state undergoes substantial decay or instability, it generates update instructions for fuel injection parameters and expansion wall angles in advance, constructing an adaptive closed-loop iterative architecture with time-advanced intervention capabilities to maintain the operational stability of the scramjet engine under different operating conditions.

[0030] 3. This invention calculates overdrive parameters based on actual propagation velocity parameters. When the detonation wave is in an underdriven state and exhibits a backward propagation trend, a basic control command is generated to adjust the fuel injection system to increase the absolute pressure and physical flow area of ​​the pipeline. When it is in a strong overdriven state and exhibits forward propagation, a basic control command is generated to drive the motor to rotate and increase the expansion wall angle. The above process establishes a dual-channel bottom-level safety intervention mechanism for different detonation wave instability modes. By locally increasing fuel supply or changing the physical geometric expansion back pressure of the transition section, the heat release from the chemical reaction is adjusted and the propagation path is optimized, reducing energy dissipation caused by the fluid boundary layer and lateral expansion. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0032] Figure 2 This is a schematic diagram of the system architecture of the present invention;

[0033] Figure 3 This is a schematic diagram comparing the propagation velocity of detonation waves under different control methods in a specific application embodiment of the present invention;

[0034] Figure 4 This diagram illustrates the comparison of peak combustion chamber pressures under different control methods in specific application embodiments of the present invention. Detailed Implementation

[0035] See attached document Figure 2 The present invention provides an adaptive detonation wave enhancement system, which may include: an intelligent sensor monitoring system, a control algorithm unit, a fuel injection system, and an expansion wall adjustment control system.

[0036] The intelligent sensor monitoring system is located in and around the combustion chamber of the scramjet engine and is communicatively connected to the control algorithm unit. The control algorithm unit is electrically connected to both the fuel injection system and the expansion wall adjustment control system. The control algorithm unit receives physical quantity data collected by the intelligent sensor monitoring system and sends control commands to both the fuel injection system and the expansion wall adjustment control system.

[0037] The intelligent sensor monitoring system includes a pressure sensor array, a temperature sensor array, and a high-speed camera system. The pressure sensor array contains multiple individually numbered pressure sensors. The first pressure sensor is installed at the combustion chamber inlet, and the final pressure sensor is installed at the combustion chamber outlet. The remaining pressure sensors are installed at equal intervals along the length of the combustion chamber wall. The pressure sensor array is used to collect wall pressure distribution data at different axial positions within the combustion chamber and transmits this data to the control algorithm unit in real time.

[0038] The temperature sensor array contains multiple individually numbered temperature sensors. These sensors are fixedly mounted on the combustion chamber wall and along the combustion chamber centerline. The temperature sensor array is used to collect temperature field distribution data at different spatial nodes inside the combustion chamber and transmit this data to the control algorithm unit.

[0039] The high-speed camera system is located outside the combustion chamber, near the observation window. The system's frame rate is configured to 1000Hz. It acquires images of the spatial shape of the detonation wavefront within the combustion chamber and combines these images with continuously acquired timestamps to generate propagation velocity data. The high-speed camera system then transmits the spatial shape images and propagation velocity data to the control algorithm unit.

[0040] The control algorithm unit includes a hardware processor and memory. The memory contains a built-in information analysis module. This module is configured to receive wall pressure distribution data, temperature field distribution data, spatial shape images, and propagation velocity data. Embedded within the information analysis module are fuzzy control algorithm sequences and neural network algorithm sequences. Based on the aforementioned multi-source input data, the information analysis module calculates the current propagation position and state of the detonation wave, generating adjustment and control commands for the fuel injection system and the expansion wall adjustment and control system.

[0041] The fuel injection system is located at the front end of the combustion chamber and includes a fuel injection orifice, an injection pressure regulating valve connected to the fuel injection orifice pipeline, and an injection orifice opening regulating valve that controls the flow area of ​​the fuel injection orifice. Both the injection pressure regulating valve and the injection orifice opening regulating valve are directly connected to the control algorithm unit. The fuel injection system receives regulation control commands and changes the fuel injection quantity and injection pressure into the combustion chamber by adjusting the mechanical opening of the valves.

[0042] The expansion wall adjustment control system is located inside the combustion chamber. An adjustable expansion wall is mounted below the combustion chamber wall. The front end of the expansion wall is mechanically hinged to the combustion chamber inlet via a movable joint. A mounting base for supporting the movable joint is fixed inside the combustion chamber base.

[0043] The expansion wall adjustment control system also includes a motor drive system mechanically connected to the back of the expansion wall. The motor drive system is electrically connected to the control algorithm unit. The motor drive system receives adjustment control commands and drives the expansion wall to physically rotate around the axis of the movable joint. This physical rotation manifests as a lifting or lowering motion of the expansion wall, which changes the physical expansion angle of the combustion chamber expansion section, thereby altering the geometric parameters of the combustion chamber transition section.

[0044] The control algorithm unit's internal memory pre-configured and stores spatial coordinate data, including waveface boundary position coordinates, dynamic stationary position coordinates, detonation wave forward propagation threshold position coordinates, and detonation wave backward propagation threshold position coordinates. The information analysis module compares the detonation wave position coordinates located in real time by the pressure sensor array with the aforementioned pre-stored spatial coordinate data and uses this as a trigger condition to invoke subsequent control algorithm sequences.

[0045] See attached document Figure 1 The adaptive detonation wave enhancement method provided by the present invention relies on the above-mentioned adaptive detonation wave enhancement system, and uses a preset discrete time step as the control period to execute a closed-loop feedback control process in sequence.

[0046] During the method startup phase, the system as a whole is initialized. The control algorithm unit sends hardware reset commands to the fuel injection system and the expansion wall adjustment control system, driving the injection orifice opening adjustment valve, the injection pressure adjustment valve, and the motor drive system connected to the expansion wall to position themselves to the preset physical zero reference. Simultaneously, the control algorithm unit loads the preset fuzzy control rule matrix and the initial weight parameters of the neural network nodes from the memory into the running memory, and clears the historical state cache data remaining in the memory due to previous operation.

[0047] After entering the periodic control loop, within a single time step, the control algorithm unit controls the intelligent sensor monitoring system to synchronously acquire multi-source data. Addressing the hardware differences in sampling frequencies among the high-speed camera system, pressure sensor array, and temperature sensor array, the control algorithm unit configures a global hardware clock to generate a timestamp sequence based on a specific frequency reference. The control algorithm unit performs linear interpolation calculations on the heterogeneous data acquired by different sensors according to the timestamp sequence, achieving data alignment in the time dimension, and ultimately generating the spatial pressure distribution matrix, temperature field distribution matrix, and spatial coordinates of the detonation wavefront within the current control cycle.

[0048] After data alignment is completed, the control algorithm unit performs state feature parameter extraction. The control algorithm unit locates the maximum pressure point in the spatial pressure distribution matrix through numerical comparison and extracts the corresponding axial physical position as the current wavefront position. The control algorithm unit uses the spatial difference between the current wavefront position and the wavefront position at the previous time step, divided by the discrete time step length, to calculate the actual propagation velocity of the detonation wave.

[0049] Based on the acquired actual propagation velocity, the control algorithm unit performs overdrive calculation and basic response logic judgment. The control algorithm unit retrieves the theoretical velocity value of the ideal detonation wave from the pre-stored memory, calculates the relative ratio of the actual propagation velocity to the theoretical velocity value of the ideal detonation wave, and defines this ratio as the overdrive parameter. The control algorithm unit compares the overdrive parameter with a preset overdrive threshold. If the overdrive parameter is less than the preset overdrive threshold, the control algorithm unit activates the fuel injection system parameter adjustment program; if the overdrive parameter is greater than or equal to the preset overdrive threshold, the control algorithm unit activates the geometric path adjustment program of the expansion wall adjustment control system.

[0050] After completing the basic response logic allocation, the control algorithm unit calls the fuzzy control algorithm sequence and the neural network algorithm sequence to calculate the adaptive adjustment amount. The control algorithm unit loads the peak pressure, actual propagation velocity, and combustion chamber centerline temperature of the current cycle as input variables into the fuzzy control algorithm sequence, and outputs a quantitative evaluation factor characterizing the current combustion state through preset fuzzy rules. Subsequently, the control algorithm unit extracts the quantitative evaluation factors from multiple consecutive time steps to form a time-sliding window vector. The control algorithm unit inputs the time-sliding window vector into the neural network algorithm sequence, performs matrix multiplication and activation function operations, and outputs a predicted state factor for the next time step.

[0051] The control algorithm unit performs parameter compensation calculations based on the numerical deviation between the predicted state factor and the preset system stable operation reference value. The calculation results are output as the step compensation amount of the injection orifice opening regulating valve, the pressure increase / decrease of the injection pressure regulating valve, and the angular displacement of the motor drive system. The control algorithm unit converts the above compensation amounts into pulse width modulation signals and sends them to the drive end of each actuator to control the mechanical components to reach the target physical position. After the command is sent, the control algorithm unit updates the system global clock count value, advances to the next control cycle, and re-executes all process steps from data synchronization acquisition to drive command issuance.

[0052] This embodiment details the system startup time node (defined as time step). The hardware actuator reference positioning and software algorithm parameter allocation process.

[0053] The control algorithm unit performs parameter initialization operations for the algorithm model. Specifically, this includes setting the fuzzy control rule base and neural network model parameters for the detonation wave enhancement control algorithm. The control algorithm unit allocates variable registers in memory and binds the physical addresses of the fuzzy controller's input variables to the detonation wave pressure collected by the pressure sensor array. The detonation wave velocity extracted by the high-speed camera system and the combustion chamber temperature collected by the temperature sensor .

[0054] The control algorithm unit maps the physical addresses of the fuzzy controller's output variables to the compensation amounts for the nozzle opening degree, injection pressure, and transition section geometry parameters. The control algorithm unit loads 30 pre-calibrated fuzzy control inference rules into the memory. These fuzzy control inference rules are fixed in a two-dimensional data matrix using an If-Then conditional judgment structure.

[0055] The control algorithm unit performs memory allocation and parameter initialization for the neural network model. It sets the number of input layer nodes and configures the number of hidden layer nodes and output layer nodes based on the time sliding window length. The control algorithm unit stores the weight matrix between each layer in physical memory. and bias vector Establish a floating-point number storage array. The control algorithm unit calls the random number generation function to generate numbers with a mean of 0 and a variance of 0. The Gaussian distribution numerical matrix is ​​obtained, and the data of the numerical matrix is ​​written into the aforementioned storage array. The mathematical calculation logic for initialization is as follows:

[0056] ;

[0057] ;

[0058] in, It follows a normal distribution (Gaussian distribution). The mean, Let Variance be the variance.

[0059] This randomized distribution assignment operation is used to set the initial boundary conditions for the neural network computation parameters, breaking the output symmetry state of the hidden layer nodes in the early stage of matrix multiplication.

[0060] The control algorithm unit initializes the electromechanical parameters of the fuel injection system. It generates a first control signal and sends it to the electromagnetic drive coil of the injection orifice opening regulating valve, driving the internal valve core to produce a mechanical linear displacement, thus adjusting the initial opening state of the injection orifice. The physical limit is 50% of the maximum mechanical stroke. Subsequently, the control algorithm unit generates a second control signal and sends it to the control terminal of the injection pressure regulating valve, driving the pneumatic damping element to change the flow channel cross-sectional area, establishing an initial set back pressure in the fuel supply line, and setting the initial fuel injection pressure. Physically stabilized to 0.8MPa.

[0061] The control algorithm unit performs geometric initialization of the expansion wall adjustment control system. Based on the physical space constraints of the combustion chamber, the control algorithm unit sets the initial geometric characteristic parameters of the transition section in the register. The control algorithm unit sets the effective axial reference length of the transition section. Write the value as 0.3m, and set the maximum physical radius of the transition section at the end of the flow direction cross-section. The writing length is 0.15m. The control algorithm unit calculates the initial expansion angle by combining the reference length and the maximum physical radius. The control algorithm unit sends the initial expansion angle to the motor drive system. The servo pulse command drives the motor to output physical torque, causing the expansion wall below the combustion chamber wall to mechanically rotate around the axis of the front movable joint until it contacts the preset angle limit point, thus completing the solid construction of the reference geometric boundary of the internal pneumatic channel of the combustion chamber.

[0062] The control algorithm unit controls each discrete time step The data acquisition bus of the intelligent sensor monitoring system is triggered to simultaneously read the thermodynamic and kinematic physical quantities in the combustion chamber and perform data structuring processing to extract the physical characteristic parameters of the knock wave.

[0063] The control algorithm unit reads the electrical signals collected by the pressure sensor array via the data bus. The pressure sensor array contains eight physically fixed pressure sensors, each corresponding to one of the axial coordinates of the combustion chamber. , where the index number The value range is 1 to 8. The first pressure sensor is fixed at the combustion chamber inlet coordinate system. The eighth pressure sensor is fixed at the coordinate position of the combustion chamber outlet. The second to seventh pressure sensors are arranged at equal physical intervals. and Between. The control algorithm unit converts the acquired analog electrical signals into digital quantities to generate the current time step. Spatial pressure distribution sequence .

[0064] The control algorithm unit executes an array traversal algorithm to locate the spatial pressure distribution sequence through numerical comparison. The maximum value in the data is used to extract the axial physical coordinates corresponding to the maximum value as the current position of the detonation wavefront. Its mathematical extraction logic is expressed as:

[0065] ;

[0066] in, This indicates that the parameters are refined.

[0067] The control algorithm unit simultaneously extracts the absolute pressure value corresponding to that coordinate position and stores it as the peak pressure parameter of the detonation wave at the current moment. .

[0068] The control algorithm unit retrieves the previous time step from the cache in memory. Location of the blast wave front Combined with the fixed sampling period set by the system The actual propagation velocity of the detonation wave in the combustion chamber axial direction was calculated using the backward difference method. The calculation process is expressed as follows:

[0069] ;

[0070] A synchronously operating high-speed camera system performs continuous optical imaging of the combustion chamber interior at a fixed physical frame rate of 1000Hz. The control algorithm unit receives the two-dimensional pixel matrix sequence output by the high-speed camera system and extracts the set of consecutive pixels whose luminous intensity exceeds a preset grayscale threshold as the optical spatial shape of the wavefront. The control algorithm unit uses an image recognition algorithm to determine the pixel translation amount of the aforementioned consecutive pixel set between two adjacent frames and, combined with the physical calibration parameters of the camera lens, converts the pixel translation amount into an optical observation propagation velocity. The control algorithm unit then compares the optical observation propagation velocity with the actual propagation velocity calculated based on pressure difference. Perform an arithmetic average and overwrite the resulting value with the current time step. The final actual propagation speed parameter.

[0071] The control algorithm unit reads the analog signals output by the temperature sensor array via a multiplexer. The temperature sensor array contains five independent temperature sensors, each embedded in a mounting hole in the combustion chamber wall and mounted on a measuring bracket fixed to the centerline of the combustion chamber cross-section. Their corresponding spatial axial coordinates are marked as follows: , where the index number The value ranges from 1 to 5. The control algorithm unit converts the above analog signal into a temperature value and generates the current time step. Temperature field distribution sequence The control algorithm unit will determine the position of the detonation wavefront. Using the target input value, a spatial interpolation algorithm is invoked on the temperature field distribution sequence. The temperature value at the corresponding coordinate is calculated and output as the characteristic temperature parameter of the combustion chamber at that moment. .

[0072] After completing the feature parameter extraction, the control algorithm unit sets the current time step. Extracted peak pressure parameters of detonation wave Actual propagation speed parameters and combustion chamber characteristic temperature parameters Write to the data register as the basic boundary condition for triggering subsequent logical judgments and adaptive calculation mechanisms.

[0073] The control algorithm unit obtains the current time step. After obtaining the physical characteristic parameters of the detonation wave, a numerical determination program for the propagation state of the detonation wave is executed, and the corresponding hardware actuator is triggered to perform basic response actions based on the determination result.

[0074] The control algorithm unit retrieves the time step from the register. Actual propagation speed parameters It also reads the ideal Chapman-Jouguet detonation wave velocity constant, pre-calculated and solidified based on the combustion chamber intake conditions, from the read-only memory. The control algorithm unit performs floating-point division and subtraction operations to calculate the deviation of the current detonation wave from the ideal Chapman-Jouguet detonation wave velocity constant. The dimensionless relative ratio is used to generate the overdrive parameter. Overdrive parameters The calculation process is expressed as follows:

[0075] ;

[0076] After the calculation is completed, the control algorithm unit retrieves the preset overdrive critical threshold. and overdrive parameters Overdrive Critical Threshold Perform numerical comparisons. The control algorithm unit synchronously retrieves the time step. Location of the blast wave front The wavefront position deviation is calculated by performing a difference operation between the wavefront position deviation and the preset dynamic stationary position coordinates, the detonation wave forward threshold position coordinates, and the detonation wave backward threshold position coordinates in the memory. The control algorithm unit then combines the comparison result of the overdrive parameters with the wavefront position deviation and executes a binary logic branch routing.

[0077] When the logical condition satisfies the overdrive parameter Less than the critical threshold of overdrive When the wavefront position deviation indicates that the wavefront has reached the detonation wave backpropagation threshold position coordinates, the control algorithm unit determines that the detonation wave is in a physical state of propagation velocity attenuation. In this state, the control algorithm unit generates a first basic control command sequence and sends it to the fuel injection system. This first basic control command sequence drives the stepper motor of the injection orifice opening regulating valve to rotate forward, increasing the physical flow area of ​​the fuel injection orifice; simultaneously, it drives the throttling component of the injection pressure regulating valve to actuate, increasing the absolute pressure within the fuel injection pipeline. This physical action increases the fuel mass flow rate entering the combustion chamber, replenishing reactants to maintain the self-sustaining propagation of the detonation wave.

[0078] When the logical condition satisfies the overdrive parameter Greater than or equal to the overdrive critical threshold When the wavefront position deviation indicates that the wavefront has reached the detonation wave propagation threshold position coordinates, the control algorithm unit determines that the detonation wave is in an overdrive propagation physical state. In this state, the control algorithm unit generates a second basic control command sequence and sends it to the expansion wall adjustment control system. The second basic control command sequence drives the motor drive system to output a set number of rotational displacements, causing the expansion wall to rotate around the axis of the front movable joint in a direction away from the combustion chamber centerline. This physical action increases the physical expansion angle of the expansion wall, expands the internal cross-sectional area of ​​the combustion chamber in this axial section, and reduces the airflow back pressure to promote the rearward propagation of the detonation wave.

[0079] When the wavefront position deviation indicates that the wavefront coordinates are equal to the preset dynamic stationary position coordinates, the control algorithm unit outputs a stationary maintenance command. This command controls the motor drive system to execute electromagnetic braking, locking the current mechanical position of the expansion wall and maintaining the current valve opening setting of the fuel injection system. The control algorithm unit writes the valve openings and motor displacements after the basic response logic has been executed into a cache as the current system state update values, for use by the adaptive depth control algorithm at subsequent time steps.

[0080] After executing the basic response logic, the control algorithm unit further calls the fuzzy control algorithm sequence stored in the memory to perform a quantitative numerical evaluation of the mixture state and the physical state of the knock wave propagation in the combustion chamber.

[0081] The control algorithm unit reads the current time step in batches from the data register. The three core physical characteristic parameters extracted are the peak pressure parameter of the detonation wave. Actual propagation speed parameters and combustion chamber characteristic temperature parameters The control algorithm unit uses the three precise numerical parameters mentioned above as physical input variables for the fuzzy inference engine and performs fuzzy numerical mapping operations.

[0082] During the fuzzification phase, the control algorithm unit retrieves the domain range and corresponding membership functions of each input variable from the pre-stored memory. The physical distribution domain is divided into three consecutive numerical intervals: low, normal, and high. The actual propagation speed parameter is... Combustion chamber characteristic temperature parameters All are divided into corresponding numerical intervals according to the same logic. The control algorithm unit adopts a mathematical model that combines triangular membership functions and Gaussian membership functions to calculate the floating-point membership values ​​of the three input variables in fuzzy sets in different numerical intervals, and generates fuzzy set vectors of input quantities.

[0083] Upon entering the rule inference phase, the control algorithm unit invokes the fuzzy control rule library loaded into the runtime memory. This library contains 30 Mamdani conditional judgment instruction sequences constructed based on historical experimental calibration data. The control algorithm unit substitutes the fuzzy set vector of the input variable into the conditional antecedents of the 30 instruction sequences. For each instruction sequence, the control algorithm unit determines the activation strength triggered by the instruction sequence by calculating the minimum value of the floating-point membership degree of the antecedent variable. Subsequently, the control algorithm unit performs a logical intersection operation on the activation strength of each instruction sequence and its corresponding rule consequent fuzzy set, obtaining a fuzzy subset of the output result for each rule.

[0084] The control algorithm unit performs a logical union operation, taking the maximum value, on the fuzzy subsets of the output results of all activated instruction sequences, aggregating the dispersed fuzzy subset envelopes into a continuous overall output fuzzy set. The overall output fuzzy set is distributed within the defined system stable state universe of discourse.

[0085] During the defuzzification phase, the control algorithm unit uses the discrete centroid method to solve the overall output fuzzy set, converting it into a single-dimensional precise output value, which is defined as the current comprehensive state evaluation factor. The control algorithm unit performs multiplication-addition-integral operations on discrete sampling points in the output universe of discourse. The specific computational logic for its centroid-based defuzzification is as follows:

[0086] ;

[0087] In the formula, Denotes the first state set within the system's stable state domain. Discrete physical sampling points This indicates the overall output fuzzy set at discrete physical sampling points. The membership degree value at the location, This outputs the total number of sampling points within the universe of discourse.

[0088] After completing the above division and summation calculations, the control algorithm unit will calculate the current comprehensive state evaluation factor. The output is written to the system state buffer register. This factor, as a normalized quantitative floating-point number, directly reflects the physical coupling result between aerodynamic and thermodynamic parameters and the detonation wave propagation state within the current time step, and serves as the basic input data for subsequent neural network sequence execution of time series trend prediction.

[0089] After completing the fuzzy inference current state assessment, the control algorithm unit calls the feedforward neural network algorithm sequence stored in memory and uses historical state data within the continuous time step to calculate the predicted value of the detonation wave physical state for the next time step.

[0090] The control algorithm unit performs the construction operation of the time sliding window data. The control algorithm unit reads past consecutive data from the system state buffer register in chronological order. Comprehensive state assessment factors within a time step. This is a preset positive integer constant in a fixed storage area, representing the physical length of the time observation window. The control algorithm unit will read... The floating-point values ​​are arranged in chronological order and constructed as a dimension in the runtime memory. Input feature column vector Its mathematical structure is defined as follows:

[0091] ;

[0092] In the formula, The comprehensive state assessment factor is calculated for the current time step. The comprehensive state evaluation factor is the one with the furthest time step within the time sliding window. This indicates transpose.

[0093] The control algorithm unit inputs the feature column vector. The boundary conditions for the neural network input layer are loaded into the arithmetic logic unit. The control algorithm unit retrieves the weight matrix from the input layer to the hidden layer from the weight register. and the corresponding bias vector The control algorithm unit controls the weight matrix. With input feature column vector Perform matrix multiplication, then combine the resulting column vector with the bias vector. Perform matrix addition.

[0094] After completing the above linear algebra operations, the control algorithm unit calls the first nonlinear activation function stored in memory. The control algorithm unit substitutes each component of the column vector of the addition operation result into the first nonlinear activation function. Numerical mapping is performed to generate the hidden layer output state vector. The logical representation of this step is as follows:

[0095] ;

[0096] Subsequently, the control algorithm unit performs the forward computation of the output layer. The control algorithm unit retrieves the weight matrix from the hidden layer to the output layer. and output bias scalar The control algorithm unit controls the weight matrix. With the hidden layer output state vector Perform matrix multiplication and add the output bias scalar. .

[0097] The control algorithm unit calls the second nonlinear activation function stored in memory. Substitute the scalar result of the above operation into the second nonlinear activation function. Perform numerical interval-constrained calculations to generate values ​​for the next time step. Predictive state evaluation factors The output layer operation logic is represented as follows:

[0098] ;

[0099] After the control algorithm unit completes the calculation of all the matrices and nonlinear functions mentioned above, it will generate the predicted state evaluation factor. This predicted state evaluation factor is written as a single floating-point number to the specified physical address of the control execution buffer. Leading the current physical time step in the time dimension It is used to trigger and calculate the adaptive parameter update amount of subsequent actuators to compensate for the physical movement delay of mechanical actuators.

[0100] After completing the propagation trend prediction of the neural network, the control algorithm unit executes the algorithm for the next time step based on the prediction state evaluation factors written in the memory. The calculation and distribution of physical control commands for mechanical actuators.

[0101] The control algorithm unit reads the preset expected optimal stable state reference factor from the read-only memory area. The control algorithm unit will run the predicted state evaluation factors in memory. With the expected optimal steady-state reference factor A floating-point subtraction operation is performed to generate the system state deviation. The control algorithm unit calculates the fuel injection orifice opening compensation value based on the system state deviation. The control algorithm unit then retrieves the preset adaptive learning gain coefficient for the opening degree. and the corresponding opening degree control mapping function The control algorithm unit sets the current time step. injection orifice opening Using this as a baseline, perform iterative update calculations to generate the next time step. Injection orifice opening update command Its mathematical calculation logic is expressed as follows:

[0102] ;

[0103] In the formula, To limit the proportional constant of the physical stroke of a stepper motor in a single operation, This is a mathematical function that converts dimensionless system state deviations into physical dimensions of mechanical opening degree.

[0104] Simultaneously, the control algorithm unit calculates the pressure compensation value for the fuel injection pipeline based on the system state deviation. The control algorithm unit retrieves the preset injection pressure adaptive learning gain coefficient. and the corresponding pressure control mapping function The control algorithm unit sets the current time step. fuel injection pressure Using this as a baseline, perform iterative update calculations to generate the next time step. Fuel injection pressure update command Its mathematical calculation logic is expressed as follows:

[0105] ;

[0106] In the formula, To limit the proportional constant of the rate of change of the throttling area of ​​the pressure regulating valve during a single operation, To convert the dimensionless system state deviation into a mathematical function of fluid pressure (a physical quantity), the control algorithm unit further calculates the geometric angle compensation value of the combustion chamber transition section expansion wall based on the system state deviation. The control algorithm unit then retrieves a preset expansion angle adaptive learning gain coefficient. and the corresponding angle control mapping function The control algorithm unit sets the current time step. transition section expansion wall angle Using this as a baseline, perform iterative update calculations to generate the next time step. Transition segment expansion wall angle update command Its mathematical calculation logic is expressed as follows:

[0107] ;

[0108] In the formula, To limit the proportional constant of the number of rotation steps in a single cycle of the motor drive system, This is a mathematical function that converts dimensionless system state deviations into spatial geometric angular physical dimensions.

[0109] After completing the above iterative update calculations, the control algorithm unit will generate the injection orifice opening update command. Fuel injection pressure update command and the transition section expansion wall angle update command This is converted into a corresponding pulse width modulation (PWM) electrical signal. The control algorithm unit sends the aforementioned PWM electrical signals in parallel to the drive end of the injection orifice opening regulating valve, the control end of the injection pressure regulating valve, and the motor drive system in the expansion wall adjustment control system via the output interface board.

[0110] Each mechanical actuator receiving the signal generates a corresponding physical displacement based on the aforementioned pulse width modulated electrical signal. The physical actions of the injection orifice opening regulating valve and the injection pressure regulating valve change the mass flow rate and fluid momentum of fuel entering the combustion chamber; the physical actions of the expansion wall regulating control system change the physical flow area of ​​the combustion chamber transition section. The above sequence of actions completes the adaptive closed-loop regulation of this control cycle, the system global clock counter increments, and the control algorithm unit prepares to enter the next time step. Multi-source data acquisition and iterative calculation.

[0111] Specific application examples:

[0112] To further clarify the technical solution of the present invention, taking the operation of a certain type of scramjet engine at Mach 6 as an example, the working process of the adaptive detonation wave enhancement method is explained in detail:

[0113] Once the engine is in operation, the intelligent sensor monitoring system located in the combustion chamber synchronously samples data with a time step of 0.1 milliseconds. At a certain moment, the system extracts the multi-source physical quantities of the current time step: the detonation wavefront is located at 0.85 meters axially in the combustion chamber, the peak pressure parameter of the detonation wave is 1.2 MPa, the actual propagation speed parameter is 1750 m / s, and the characteristic temperature parameter of the combustion chamber is 2600 Kelvin.

[0114] The system calculates the overdrive parameter based on the pre-calibrated ideal Chapman-Jouguet detonation wave velocity constant (set to 2000 m / s), obtaining a dimensionless relative ratio of 0.875. Since this overdrive parameter is less than the preset critical threshold of 1.0 in memory, and the wavefront position has reached the preset backpropagation threshold position coordinates, it indicates that the detonation wave is currently in an underdriven state and has a tendency to attenuate towards the isolation section. The control algorithm unit then generates the first basic control command, driving the regulating valve of the fuel injection system to increase the absolute pressure of the injection orifice by 15%, thereby increasing the local fuel equivalence ratio.

[0115] Simultaneously, the system substitutes the aforementioned physical parameters into the fuzzy control algorithm sequence. By calculating membership degrees and performing rule intersection and union operations, the system uses the discrete centroid method to calculate a dimensionless comprehensive state evaluation factor, currently valued at 0.45 (1.0 represents the ideal stable state). The system further extracts the evaluation factor sequence from the past 10 consecutive time steps, constructs an input feature column vector, and inputs it into an initialized feedforward neural network algorithm sequence for forward propagation calculation, predicting that the state evaluation factor for the next time step will decrease to 0.41.

[0116] For the deviation of 0.44 between the predicted state evaluation factor and the system's set reference benchmark (0.85), the control algorithm unit substitutes this deviation into the internal control mapping function to calculate and generate opening degree compensation values ​​and angle compensation values. The system adds the iterative benchmark value and the compensation value, outputs an update command and sends it to the underlying actuator, increasing the physical flow area of ​​the fuel injection orifice by 5%, while the drive motor contracts the expansion wall angle by 0.5 degrees towards the centerline to increase the back pressure of the flow field. Under closed-loop adaptive iterative adjustment, the actual propagation speed of the detonation wave rapidly recovers to 1950 m / s within the subsequent three time steps, successfully avoiding decoupling and detonation failure.

[0117] Experimental verification and effect comparison:

[0118] To verify the effectiveness of the present invention, a combustion chamber bench test was conducted under the same supersonic inflow conditions, using both the traditional constant-parameter isobaric injection control method and the adaptive enhancement method provided by the present invention.

[0119] Appendix Figure 3 and attached Figure 4 The dynamic evolution of the detonation wave propagation state under two different control methods is shown within a 10-millisecond test period. (Appendix) Figure 3 For comparison of the actual propagation speed of detonation waves, see attached. Figure 4 Comparison of peak combustion chamber pressures.

[0120] like Figure 3 As shown by the dark gray dashed line, under traditional control methods, due to the inability to dynamically adjust based on aerodynamic boundary layer losses and lateral expansion effects, the actual propagation speed of the detonation wave exhibits a significant and continuous decay trend after a brief initial maintenance. Around 6 milliseconds into the process, influenced by spatial curvature and energy dissipation, the propagation speed experiences a precipitous drop, rapidly falling below the speed of sound from 1600 m / s. Simultaneously, the corresponding peak pressure curve (dark gray dashed line) in the lower half of the graph also sharply decreases from 1.4 MPa to 0.3 MPa. Aerodynamic-thermodynamic decoupling occurs in the flow field, and the detonation wave completely dissipates and extinguishes.

[0121] In comparison, such as Figure 4As shown by the solid black line, after adopting the method of this invention, the system predicts the state of the detonation wave in advance through a fuzzy neural network and performs closed-loop updates on the injection pressure and wall expansion angle. Although the actual propagation speed experiences high-frequency small-amplitude oscillations caused by aerodynamic disturbances, the overall trajectory remains closely aligned with the baseline of 2000 m / s; simultaneously, the peak pressure in the combustion chamber (marked with dots on the solid black line) is stably maintained in the high operating range of 1.4 to 1.6 MPa. The above data demonstrate that the dual-channel bottom-level intervention mechanism and feedforward prediction compensation architecture established in this invention can effectively overcome non-ideal flow field disturbances and maintain the high-intensity, self-sustaining, and stable propagation of the detonation wave.

Claims

1. An adaptive detonation wave enhancement method, characterized in that, Includes the following steps: The multi-source physical quantities of the scramjet engine combustion chamber are obtained, and the detonation wavefront position, peak pressure parameter, actual propagation velocity parameter, and characteristic temperature parameter of the combustion chamber at the current time step are extracted. Based on the actual propagation speed parameters, the overdrive parameters are calculated. Based on the overdrive parameters and the position of the detonation wave front, basic control commands are generated and sent to the fuel injection system or the expansion wall adjustment control system. The peak pressure parameter of the detonation wave, the actual propagation velocity parameter, and the characteristic temperature parameter of the combustion chamber are input into the fuzzy control algorithm sequence for calculation, and the comprehensive state evaluation factor is output. Extract the comprehensive state evaluation factors from multiple consecutive time steps to construct a time sliding window and input it into a neural network algorithm sequence to calculate and output the predicted state evaluation factors for the next time step. Based on the deviation between the predicted state evaluation factor and the reference benchmark, a parameter update command is generated and sent to the fuel injection system and the expansion wall adjustment control system to adjust the fuel injection parameters and the expansion wall angle.

2. The adaptive detonation wave enhancement method according to claim 1, characterized in that, The multi-source physical quantities include the spatial pressure distribution sequence, the optical observation propagation speed output by the high-speed camera system, and the temperature field distribution sequence; The steps of obtaining multi-source physical quantities of the scramjet engine combustion chamber and extracting the detonation wavefront position, peak pressure parameter, actual propagation velocity parameter, and characteristic temperature parameter of the combustion chamber at the current time step specifically include: The axial physical coordinates corresponding to the largest value in the spatial pressure distribution sequence are used as the position of the detonation wave front, and the corresponding absolute pressure value is extracted as the peak pressure parameter of the detonation wave. Calculate the spatial difference between the current and previous time step positions of the detonation wavefront and the quotient of the time step, and perform an arithmetic average with the optical observation propagation speed output by the high-speed camera system to generate the actual propagation speed parameter. Using the position of the detonation wavefront as the target input value, a spatial interpolation algorithm is executed in the temperature field distribution sequence to calculate the temperature value at the corresponding coordinates as the characteristic temperature parameter of the combustion chamber.

3. The adaptive detonation wave enhancement method according to claim 1, characterized in that, The steps of calculating the overdrive parameter based on the actual propagation velocity parameter, and generating and sending basic control commands to the fuel injection system or the expansion wall adjustment control system based on the overdrive parameter and the position of the detonation wave front, specifically include: The dimensionless relative ratio of the actual propagation velocity parameter to the preset ideal Chapman-Jouguet detonation wave velocity constant is calculated as the overdrive parameter. When the overdrive parameter is less than the preset overdrive critical threshold and the position of the detonation wave front reaches the preset detonation wave back propagation threshold position coordinate, a first basic control command is generated and sent to the fuel injection system. When the overdrive parameter is greater than or equal to the overdrive critical threshold, and the position of the detonation wave front reaches the preset detonation wave forward propagation threshold position coordinates in the memory, a second basic control command is generated and sent to the expansion wall adjustment control system. When the position of the detonation wavefront is between the coordinates of the detonation wave forward propagation threshold position and the coordinates of the detonation wave backward propagation threshold position, it is determined that the detonation wave is in a stationary state, and a maintenance command is generated to maintain the current basic physical state of the fuel injection system and the expansion wall adjustment control system.

4. The adaptive detonation wave enhancement method according to claim 3, characterized in that, The first basic control command is used to drive the injection orifice opening adjustment valve of the fuel injection system to increase the physical flow area and drive the injection pressure adjustment valve to increase the absolute pressure in the fuel injection pipeline. The second basic control command is used to drive the motor drive system of the expansion wall adjustment control system to output rotational displacement, causing the expansion wall to rotate around the movable joint in a direction away from the center line of the combustion chamber.

5. The adaptive detonation wave enhancement method according to claim 1, characterized in that, The specific steps of inputting the peak pressure parameter of the detonation wave, the actual propagation velocity parameter, and the characteristic temperature parameter of the combustion chamber into the fuzzy control algorithm sequence for calculation and outputting the comprehensive state evaluation factor include: Calculate the floating-point membership degrees of the three parameters in fuzzy sets with different numerical ranges, and generate the input fuzzy set vector. The input fuzzy set vector is substituted into the condition antecedent of the fuzzy control rule base to calculate the activation intensity, and a logical intersection operation is performed with the corresponding rule consequent fuzzy set to generate a fuzzy subset of the rule output result. A logical union operation is performed on the fuzzy subset of the activated rule output results to generate a total output fuzzy set. The discrete centroid method is then used to perform multiplication-addition-integral operations on the total output fuzzy set at discrete sampling points in the output universe to calculate and generate the comprehensive state evaluation factor.

6. The adaptive detonation wave enhancement method according to claim 1, characterized in that, The step of extracting the comprehensive state evaluation factors for multiple consecutive time steps, constructing a time sliding window, and inputting it into the neural network algorithm sequence specifically includes: Read the comprehensive state evaluation factors within a consecutive preset integer number of time steps in chronological order, and construct an input feature column vector in column vector form; Perform matrix multiplication on the weight matrix from the input layer to the hidden layer and the input feature column vector, perform matrix addition on the result and the bias vector, and substitute it into the first nonlinear activation function to generate the hidden layer output state vector.

7. The adaptive detonation wave enhancement method according to claim 6, characterized in that, The specific steps for calculating and outputting the predicted state evaluation factor for the next time step include: Perform matrix multiplication on the weight matrix from the hidden layer to the output layer and the output state vector of the hidden layer, and add the output bias scalar; The calculation result is substituted into the second nonlinear activation function for numerical interval restriction calculation to generate the predicted state evaluation factor for the next time step.

8. The adaptive detonation wave enhancement method according to claim 1, characterized in that, The step of generating parameter update instructions based on the deviation between the predicted state evaluation factor and the reference benchmark and sending them to the fuel injection system and the expansion wall adjustment control system specifically includes: The injection orifice opening, fuel injection pressure, and expansion wall angle at the current time step are used as the iterative baseline values. The deviations are substituted into the opening degree control mapping function, the pressure control mapping function, and the angle control mapping function, respectively, and multiplied by the corresponding adaptive learning gain coefficients to generate opening degree compensation values, pressure compensation values, and angle compensation values. The iterative baseline value is added to the corresponding compensation value to generate the injection orifice opening update command, fuel injection pressure update command, and expansion wall angle update command for the next time step, respectively.

9. The adaptive detonation wave enhancement method according to claim 1, characterized in that, Before acquiring the multi-source physical quantities of the scramjet engine combustion chamber, an initialization setup step is also included: Load the pre-calibrated fuzzy control inference rules into the memory, establish a floating-point storage array of weight matrix and bias vector for the neural network algorithm sequence, and write the values ​​that follow a Gaussian distribution with a set mean and variance. The regulating valve of the fuel injection system is limited to the initial opening state and the initial fuel injection pressure, and the motor drive system of the expansion wall adjustment control system is driven to rotate to the set initial expansion angle.

10. An adaptive detonation wave enhancement system, applied to the adaptive detonation wave enhancement method according to any one of claims 1-9, characterized in that, include: An intelligent sensor monitoring system is installed in the combustion chamber of the scramjet engine to simultaneously acquire multiple physical quantities. The control algorithm unit is communicatively connected to the intelligent sensor monitoring system and includes a hardware processor and a memory with a built-in information analysis module. The hardware processor reads and executes the information analysis module to: receive the multi-source physical quantities, perform calculations, and output parameter update instructions. A fuel injection system is located at the front end of the combustion chamber and electrically connected to the control algorithm unit. It receives the parameter update command and is used to adjust the state of the fuel fluid entering the combustion chamber. The expansion wall adjustment control system includes an expansion wall disposed inside the combustion chamber and a motor drive system. The motor drive system is electrically connected to the control algorithm unit to receive the parameter update command and drive the expansion wall to rotate, thereby adjusting the physical expansion angle of the combustion chamber transition section.

Images