Intelligent grid control system and method for hypersonic flow field calculation

Abstract

The invention discloses an intelligent grid control system and method for hypersonic flow field calculation, relates to the technical field of supercomputing parallel calculation and flow field numerical simulation crossing, and adopts an integrated deep coupling architecture of resonance sensing, coupling regulation and control, cooperative adaptation and closed loop verification. Comprising a parallel deviation chaotic characteristic perception and resonance threshold calibration module, a flow field deviation resonance coupling regulation and control module and a resonance coupling cooperative regulation core module, in the invention, chaos characteristics of supercomputing parallel synchronization deviation are captured, a controllable resonance coupling mechanism with flow field fluctuation is established, and integrated cooperative regulation and control are realized through bidirectional gain adjustment, accurate coupling coefficient calculation and closed-loop verification, so that the problems of efficiency loss, linear feedback adjustment lag, fixed threshold redundancy and low efficiency caused by suppression of parallel deviation in the prior art are solved. According to the method, the problems of low flow field calculation precision and stability are solved, and the effects of synchronously improving the hypersonic flow field calculation efficiency, precision and stability, adapting to different working conditions and not needing hardware transformation are achieved.

Description

Technical Field

[0001] This invention relates to the field of supercomputing parallel computing and flow field numerical simulation, specifically to an intelligent grid control system and method for hypersonic flow field calculation. Background Technology

[0002] Hypersonic flow field calculation is a core technology for aircraft design and flow field characteristic analysis in the aerospace field. Its calculation accuracy and efficiency directly determine the development cycle and reliability. Due to the complex unsteady phenomena in hypersonic flow fields, such as shock wave-boundary layer interaction and vortex breaking, it is necessary to rely on intelligent grid control systems to dynamically adjust grid parameters to adapt to the flow field evolution characteristics and ensure calculation accuracy.

[0003] To improve computational efficiency, current technologies generally adopt multi-core parallel computing architectures for supercomputing, which process large-scale grid data through multi-core collaborative parallel processing. However, during parallel computing, there are unavoidable parallel synchronization deviations between multiple cores. Current technologies treat this as disordered interference and suppress it only through hardware optimization and synchronization instruction enhancement, without deeply exploring its intrinsic characteristics and potential value.

[0004] Meanwhile, the stability of hypersonic flow field calculations depends on stability criteria such as the CFL condition. Existing techniques mostly employ fixed thresholds or linear feedback adjustment methods based on flow field physical parameters. Fixed thresholds require a large amount of safety redundancy, resulting in a significant loss of computational efficiency; linear feedback adjustment has obvious lag, which can easily cause computational divergence in critical regions such as the steady-state to unsteady-state transition of the flow field, failing to guarantee computational stability and accuracy.

[0005] Existing technologies have not recognized the chaotic characteristics of parallel synchronization deviation in specific flow field scenarios, nor have they established a correlation mechanism between this characteristic and stability criteria. As a result, the dynamic characteristics of supercomputing parallel architecture cannot provide support for flow field stability control, making it difficult to meet the computational requirements of hypersonic flow fields for high efficiency, accuracy, and stability. In view of this, an intelligent grid control system and method for hypersonic flow field computation is provided to overcome the above problems. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent grid control system and method for hypersonic flow field calculation, so as to solve the problems mentioned in the background art.

[0007] To address the aforementioned technical problems, this invention provides an intelligent grid control system for hypersonic flow field computation. It employs an integrated, deeply coupled architecture of resonance sensing, coupling control, collaborative adaptation, and closed-loop verification. This system includes a parallel deviation chaotic characteristic sensing and resonance threshold calibration module, a flow field deviation resonance coupling control module, a resonance coupling collaborative adjustment core module, an intelligent grid dynamic resonance adaptation module, and a closed-loop resonance stability verification and optimization module. These modules are sequentially connected to form a collaborative control link. By capturing the chaotic characteristics of supercomputing parallel synchronization deviations, a controllable resonance coupling mechanism between the deviations and flow field fluctuations is established. The resonance coupling characteristics are transformed into integrated collaborative control commands for grid parameters and CFL thresholds, enabling efficient, accurate, and stable computation of hypersonic flow fields.

[0008] Furthermore, the parallel deviation chaotic characteristic perception and resonance threshold calibration module includes a deviation chaotic parameter acquisition submodule, a chaotic characteristic extraction submodule, and a deviation flow field resonance threshold calibration submodule. The deviation chaotic parameter acquisition submodule adopts a time step-core node dual-dimensional acquisition strategy, synchronously acquiring the deviation amplitude, fluctuation frequency, and phase difference of each supercomputing core at each time step, with the sampling frequency mapped 1:1 to the flow field calculation time step. The chaotic characteristic extraction submodule extracts three core parameters of deviation: chaotic intensity, fluctuation dominant frequency, and phase stability, based on the improved Lyapunov exponential algorithm. The deviation flow field resonance threshold calibration submodule establishes an effective resonance interval library of deviation chaotic parameters and flow field fluctuation parameters by pre-calculating typical hypersonic flow field cases.

[0009] Furthermore, the flow field deviation resonance coupling control module includes a flow field fluctuation characteristic sensing submodule and a bidirectional resonance gain adjustment submodule. The flow field fluctuation characteristic sensing submodule synchronously collects the flow field shock wave intensity, vortex breaking frequency, and pressure gradient change rate, and extracts the flow field fluctuation dominant frequency and amplitude change rate. The bidirectional resonance gain adjustment submodule adopts a bidirectional gain adjustment design from deviation to flow field and from flow field to deviation, and achieves adaptive correction of the gain coefficient through the following formula:

[0010] ;

[0011] in This is the corrected bidirectional resonant gain coefficient. This is the bidirectional resonant gain coefficient before correction. This is the gain adjustment coefficient. The current chaos intensity is the deviation. This represents the upper limit of the chaos intensity corresponding to the effective resonance interval.

[0012] Furthermore, the core module for resonant coupling coordinated regulation includes a resonant coupling coefficient adaptation submodule and a coordinated instruction generation submodule; the resonant coupling coefficient adaptation submodule calculates the resonant coupling coefficient using the following formula:

[0013] ;

[0014] in The resonant coupling coefficient is... , , The weighting coefficients for chaos intensity, rate of change of flow field fluctuation amplitude, and resonance gain are respectively, and satisfy the following conditions: The degree of chaos is the deviation. The rate of change of the amplitude of the flow field fluctuation. This is the bidirectional resonant gain coefficient. This represents the phase difference between the deviation fluctuation and the flow field fluctuation.

[0015] Furthermore, the intelligent mesh dynamic resonance adaptation module includes a mesh resonance response submodule and a local time step coupling submodule; the local time step coupling submodule calculates the local time step of the mesh refinement region using the following formula:

[0016] ;

[0017] in For the local time step of the mesh densification region, Based on the time step, The resonant coupling coefficient is... This represents the phase difference between the deviation fluctuation and the flow field fluctuation. This represents the safety factor for the time step.

[0018] Furthermore, the closed-loop resonance stability verification and optimization module includes a resonance stability verification submodule and a coupling parameter feedback correction submodule; the resonance stability verification submodule has a residual change rate ≤10 -3 Furthermore, a shock wave position error of ≤3% is used as the criterion for resonance stability; if the residual rate of change is >10... -3 If the error is greater than 3%, the effective resonance interval threshold is corrected. The coupling parameter feedback correction submodule feeds back the verification result to the resonance threshold calibration submodule and the coupling coefficient adaptation submodule to correct the effective resonance interval threshold and the weight of the coupling coefficient calculation.

[0019] A smart grid control method for hypersonic flow field calculation, with controllable resonant coupling as its core, includes the following steps:

[0020] Step (1) Initialize resonance parameters and calculation environment, load typical flow field resonance threshold library, and set the deviation acquisition frequency and flow field calculation time step to be synchronized 1:1;

[0021] Step (2) Start parallel flow field calculation and simultaneously collect parallel deviation chaos parameters and flow field fluctuation parameters;

[0022] Step (3) Achieve controllable resonant coupling between deviation and flow field fluctuation through bidirectional gain adjustment, and calculate the resonant coupling coefficient;

[0023] Step (4) Generate and execute integrated collaborative instructions for mesh parameters, CFL threshold, and core resource allocation based on the resonant coupling coefficient;

[0024] Step (5) Verify the closed-loop resonance stability by comparing the residual change rate with the flow field parameter error, and correct the resonance parameters;

[0025] Step (6) Iterate through steps (2) to (5) until the flow field calculation termination condition is met, and output the calculation results and mesh parameters.

[0026] Furthermore, the parameters initialized in step (1) include: the effective resonance interval parameters are the deviation Lyapunov exponent of 0.45-0.65, the fluctuation main frequency difference ≤0.2Hz, and the weighting coefficient for calculating the resonance coupling coefficient. =0.3、 =0.4、 =0.3, initial value of bidirectional resonance gain =0.7, time step safety factor =0.95, resonance stability verification period is once every 5 time steps.

[0027] Furthermore, in step (3), if the difference between the deviation and the main frequency of the flow field fluctuation is >0.2Hz, the main frequency of the deviation fluctuation is brought closer to the main frequency of the flow field fluctuation by adjusting the interaction delay of the supercomputing core data through software. If the adjustment is ineffective, the non-critical area mesh of the flow field is thinned with a thinning coefficient of 1.1. When the deviation chaos intensity is ≥0.7, the gain coefficient is corrected by the bidirectional resonance gain adjustment formula.

[0028] Compared with the prior art, the beneficial effects of the present invention are:

[0029] 1. Improve parallel computing efficiency: By capturing the chaotic characteristics of supercomputing parallel synchronization deviation, a controllable resonance coupling mechanism with flow field fluctuations is established. Without suppressing the deviation, the core resources of the supercomputing are precisely matched with the flow field computing requirements: core priority is increased in complex regions and resources are released in steady-state regions. Compared with existing technologies, the resonance gain of parallel computing efficiency is achieved, avoiding efficiency loss caused by fixed safety redundancy.

[0030] 2. Improve the accuracy of flow field calculation: Relying on the bidirectional resonant gain adjustment design and precise quantization of the resonant coupling coefficient, the predictive mesh densification of complex flow field regions, namely shock wave and vortex breaking regions, is realized, and the flow field evolution characteristics are captured in advance, so that the shock wave position calculation error is ≤3% and the vortex breaking evolution simulation error is reduced, solving the lag problem of linear feedback adjustment in existing technologies.

[0031] 3. Ensuring computational stability: The resonant coupling characteristics are transformed into an integrated, coordinated command for mesh parameters and CFL thresholds. Combined with precise synchronization design of local time steps and deviation fluctuation phases, the residual change rate is stabilized at 10% in the critical region of steady-state to unsteady-state transition of the flow field. -4 Within this range, there is no risk of computational divergence, achieving self-stabilizing adjustment of the stability criterion.

[0032] 4. Enhanced adaptability and sustainability of regulation: Through the closed-loop resonance stability verification and optimization module, the verification results are used to correct the effective resonance interval threshold and coupling coefficient calculation weight, continuously improving the regulation accuracy, adapting to different typical hypersonic flow field conditions, and achieving full-link coordinated regulation without hardware modification. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of an intelligent grid control system and method for hypersonic flow field calculation according to the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] Please see Figure 1 The present invention provides a technical solution:

[0036] See Figure 1 As shown, an embodiment of an intelligent grid control system and method for hypersonic flow field calculation is presented:

[0037] I. Intelligent Grid Control System:

[0038] An integrated, deeply coupled architecture with resonant sensing, coupling, control, collaborative adaptation, and closed-loop verification is adopted. A controllable resonant coupling mechanism is achieved through the collaborative work of various modules and sub-modules. Each module contains a specific sub-module design, as detailed below:

[0039] 1. Parallel Deviation Chaotic Characteristic Perception and Resonance Threshold Calibration Module:

[0040] The core function is to accurately capture the chaotic characteristics of deviations rather than suppress them, and to define the effective resonance range to avoid excessive resonance leading to uncontrolled fluctuations. This module is implemented through the following three sub-modules:

[0041] The deviation and chaos parameter acquisition submodule employs a dual-dimensional acquisition strategy based on time-step core nodes. Unlike conventional single-dimensional acquisition, it synchronously acquires the synchronous deviation amplitude fluctuation frequency and phase difference of each supercomputing core at each flow field calculation time step. The sampling frequency is strictly synchronized with the flow field calculation time step in a 1:1 mapping to ensure parameter timeliness. Specifically, the acquisition process involves embedding a parameter acquisition interface within the supercomputing parallel computing framework. This interface is bound to the flow field calculation time step trigger signal. Each completed time step calculation triggers a deviation data acquisition of each core node, and the acquired data is transmitted in real-time to the cache unit for subsequent processing.

[0042] The chaotic characteristic extraction submodule, based on an improved Lyapunov exponential algorithm, only optimizes computational efficiency and is not a cutting-edge technology. It extracts three core parameters from the deviation: chaotic intensity, Lyapunov exponential fluctuation main frequency, and phase stability. This avoids the limitation of existing technologies that only extract a single parameter. The specific extraction process is as follows: after denoising the collected deviation data, it is substituted into the improved Lyapunov exponential calculation model. The model improves computational efficiency by simplifying the iteration steps and simultaneously outputs the three parameters: chaotic intensity, fluctuation main frequency, and phase stability.

[0043] The deviation flow field resonance threshold calibration submodule: By pre-calculating typical hypersonic flow field cases, such as shock wave-boundary layer interaction vortex breaking, it establishes a library of effective resonance intervals for deviation chaotic parameters and flow field fluctuation parameters. For example, in the shock wave region, the effective resonance interval is defined as the difference between the dominant frequency of the deviation Lyapunov exponent (0.4-0.7) and the shock wave frequency of the flow field ≤ 0.2Hz. This submodule is a core prerequisite for achieving controllable resonance. The specific implementation process is as follows: Pre-calculate three or more typical hypersonic flow field conditions, record the matching relationship between flow field fluctuation parameters and deviation chaotic parameters under different conditions, determine the effective resonance intervals corresponding to each condition through statistical analysis, and establish an interval library stored in the system storage unit for subsequent use.

[0044] 2. Flow field deviation resonance coupling control module:

[0045] The core function is to achieve controllable resonance between parallel deviations and flow field fluctuations, avoiding disordered superposition. This module is implemented through the following two sub-modules:

[0046] The flow field fluctuation characteristic sensing submodule synchronously collects unsteady parameters such as shock wave intensity, vortex breaking frequency, and pressure gradient change rate from flow field calculations, and extracts the rate of change of the amplitude of the dominant frequency of flow field fluctuations. Specifically, a parameter extraction node is set in the flow field calculation equation solver. This node monitors the changes in flow field physical parameters in real time, obtains the pressure gradient change rate through differential calculation, and obtains the dominant frequency of the fluctuations and the vortex breaking frequency corresponding to the shock wave intensity through spectral analysis. The extracted parameters are synchronized with the acquisition of deviation and chaotic parameters.

[0047] The bidirectional resonance gain adjustment submodule abandons the conventional unidirectional mapping and adopts a bidirectional gain adjustment design from deviation to flow field and from flow field to deviation. When the deviation and flow field fluctuations are not within the effective resonance range, the software adjusts the delay of the supercomputing core data interaction without hardware modification to increase the main frequency of deviation fluctuations, or slightly densifies or thins the local mesh to change the main frequency of flow field fluctuations, so that both enter the effective resonance range. When the resonance intensity is too high, i.e., the deviation Lyapunov exponent > 0.7, the bidirectional gain coefficient is reduced by 0.5-0.8 linearly to avoid uncontrolled fluctuations. In the specific adjustment process, in order to achieve precise dynamic adjustment of the gain coefficient, rather than simple linear interpolation, this invention proposes a bidirectional resonance gain adjustment formula, which achieves adaptive correction of the gain coefficient by quantifying the deviation between the current resonance intensity and the effective range.

[0048] ;

[0049] In the formula:

[0050] The specific adjustment process is as follows: compare the current deviation with the flow field fluctuation parameters and the effective resonance interval library in real time. If they do not match, the adjustment mechanism is triggered. At the software level, the core data interaction delay is adjusted by modifying the delay parameter in the parallel communication protocol. Local grid pre-adjustment is achieved by calling the grid adjustment interface. The gain coefficient is adjusted by calculating the target coefficient using the above formula and replacing the current coefficient.

[0051] The corrected bidirectional resonant gain coefficient, with a value range of 0.5-0.8, is directly used for subsequent resonant coupling strength adjustment.

[0052] The bidirectional resonant gain coefficient before correction is the gain coefficient value of the previous calculation time step, with the initial value set to 0.7.

[0053] Gain adjustment coefficient, an empirical constant, ranging from 0.8 to 1.2, is used to adjust the correction rate of the gain coefficient; in this embodiment, it is set to 1.0.

[0054] The current deviation's chaotic intensity, i.e., the Lyapunov exponent, is calculated in real time through the chaotic characteristic extraction submodule.

[0055] The upper limit of the chaotic intensity corresponding to the effective resonance interval is provided by the interval library of the deviation flow field resonance threshold calibration submodule. In this embodiment, the value of the shock wave region is 0.7.

[0056] 3. Core module for resonant coupling and coordinated regulation:

[0057] This module is the core of the system. Instead of adjusting the CFL threshold and mesh parameters separately, it directly transforms the resonant coupling characteristics into coordinated control commands, which is achieved through the following two sub-modules:

[0058] Resonance Coupling Coefficient Adaptation Submodule: To accurately quantify the degree of synergistic coupling between deviation and flow field fluctuations, and to avoid misjudgment of coupling relationships caused by simple multiplication, this invention proposes a novel formula for calculating the resonance coupling coefficient, introducing a phase matching factor and parameter weights to achieve accurate characterization of the degree of coupling.

[0059] ;

[0060] In the formula:

[0061] The specific calculation process is as follows: call the above parameters from the cache unit, substitute them into the formula for calculation, and retain three decimal places in the calculation result and transmit it to the cooperative instruction generation submodule.

[0062] : Resonance coupling coefficient, with a value range of 0-1, directly reflects the degree of cooperative coupling between deviation and flow field fluctuation, and is subsequently used for cooperative command generation;

[0063] , , : These are the weighting coefficients for the resonant gain of the rate of change of the amplitude of chaotic intensity flow field fluctuations, satisfying Based on the pre-calibration under typical working conditions, the values ​​are 0.3, 0.4, and 0.3 respectively in this embodiment.

[0064] The chaotic intensity of the deviation, i.e. the Lyapunov exponent, is provided by the chaotic feature extraction submodule;

[0065] : Rate of change of flow field fluctuation amplitude, with a value range of 0-1, extracted by the flow field fluctuation characteristic sensing submodule;

[0066] The bidirectional resonant gain coefficient, ranging from 0.5 to 0.8, is provided by the bidirectional resonant gain adjustment submodule, i.e., in the formula above. ;

[0067] The phase difference between the deviation fluctuation and the flow field fluctuation, ranging from 0 to π, is obtained by synchronous acquisition and calculation from the deviation chaotic parameter acquisition submodule and the flow field fluctuation characteristic sensing submodule. This is the phase matching factor, used to quantify the degree of phase matching between the two.

[0068] The collaborative instruction generation submodule generates integrated collaborative instructions based on the K value, without independent adjustment steps. Specifically, it sets up K value judgment logic: when K∈[0.3,0.6] for moderate resonance, corresponding to a complex flow field region, it generates a collaborative instruction with a mesh refinement coefficient of 1.4 + 0.2 × K, a CFL threshold adjustment threshold of 0.7 - 0.1 × K, and a core resource tilt, increasing the core computation priority by 2 levels for complex regions; when K∈[0.1,0.3) for weak resonance, corresponding to a steady-state flow field region, it generates a collaborative instruction with a mesh thinning coefficient of 1.2 + 0.1 × (1-K), a CFL threshold adjustment threshold of 1.1 + 0.1 × (1-K), and a core resource balanced allocation; when K<0.1 or K>0.6 for ineffective resonance, it triggers the resonance gain adjustment submodule for recalibration, temporarily suspending mesh and CFL adjustments. The generated collaborative instructions are transmitted to the corresponding execution module in a standardized format.

[0069] 4. The core function of the intelligent mesh dynamic resonance adaptation module is to enable precise mesh adjustment in response to resonance coupling characteristics, which is achieved through the following two sub-modules:

[0070] The mesh resonance response submodule, upon receiving the coordination command, determines the response accuracy of the mesh adjustment based on the resonance coupling coefficient K. A higher K value results in higher adjustment accuracy; for example, when K=0.5, the mesh refinement accuracy in the shock wave region is 0.01mm, and in the steady-state region it is 0.05mm, which differs from conventional fixed-precision adjustments. The specific implementation process is as follows: after receiving the coordination command, it parses the K value and adjustment type; determines the target accuracy based on the preset K value-adjustment accuracy mapping table; calls the mesh generator's refinement or thinning interface; inputs the target accuracy and adjustment region parameters; and completes the mesh adjustment.

[0071] Local Time Step Coupling Submodule: In the mesh refinement region, to achieve precise synchronization between the local time step and the phase of the deviation fluctuation, while ensuring computational stability, this invention proposes a creative local time step coupling formula that incorporates both the resonant coupling coefficient K and the phase difference into the time step calculation:

[0072] ;

[0073] In the formula:

[0074] Specifically, the implementation involves obtaining the parameters mentioned above from the cache unit, substituting them into the formula to calculate the local time step, and then updating the time step parameters of the mesh densification region to ensure that the local time step and the deviation fluctuation phase remain synchronized with each other with a phase difference of ≤0.1π, thereby further improving computational stability and avoiding the problem of the local time step and parallel deviation becoming disconnected in existing technologies.

[0075] : The local time step of the mesh refinement region, used for iterative calculation of the flow field in that region;

[0076] The base time step is determined by the initial operating conditions of the flow field calculation and is a globally unified initial time step.

[0077] : Resonance coupling coefficient, provided by the resonance coupling coefficient adaptation submodule, is used to correlate the local time step with the degree of coupling cooperation;

[0078] The phase difference between the deviation fluctuation and the flow field fluctuation, and the resonance coupling coefficient formula mentioned above. Use the same parameter to ensure parameter consistency;

[0079] Time step safety factor, with a value range of 0.9-1.0, is used to avoid computational instability caused by excessively large time steps. In this embodiment, it is set to 0.95.

[0080] 5. Closed-loop resonance stability verification and optimization module:

[0081] The core function is to ensure that the resonant coupling is always within the effective range and to correct the control parameters. This is achieved through the following two sub-modules:

[0082] Resonance stability verification submodule: Real-time monitoring of the residual change rate stability index and flow field parameter error accuracy index during the calculation process. If the residual change rate > 10... -3 If the divergence trend or the shock wave position error in the flow field is greater than 3%, it is judged as resonance instability, triggering the gain adjustment submodule to reduce the gain. The specific monitoring process is as follows: during the flow field calculation iteration, the residual change rate is calculated in real time. After every 5 time steps, the error between the calculated flow field shock wave position result and the theoretical value is compared. The monitoring result is then compared with a preset threshold to determine whether instability has occurred.

[0083] The coupling parameter feedback correction submodule feeds back the stability verification results to the resonance threshold calibration submodule and the coupling coefficient adaptation submodule. It corrects the effective resonance range thresholds, such as the Lyapunov exponent threshold in the shock region from 0.4-0.7 to 0.45-0.65, and adjusts the calculation weights of the coupling coefficient K to improve subsequent control accuracy, forming a closed-loop system. The specific correction process is as follows: The correction direction is determined based on the instability type. If instability is caused by excessively high resonance intensity, the lower limit threshold of the effective resonance range is raised. If the error is too large, the calculation weights of each parameter in the coupling coefficient K are adjusted, and the corrected parameters are updated to the corresponding module's storage unit.

[0084] II. Calculation methods for hypersonic flow fields:

[0085] This method takes controllable resonant coupling as its core principle and achieves end-to-end coordination of supercomputing flow field grid stability through the following steps:

[0086] 1. Step 1: Initialize Resonance Parameters and Computation Environment. The core of this step is the pre-calibration of the resonance threshold. The specific operations are as follows: First, start the supercomputing multi-core parallel architecture, load the target hypersonic flow field calculation model including initial physical parameters and initial mesh parameters, allocate computing resources through the supercomputing task scheduling module, and determine the number and distribution of core nodes participating in the calculation; Second, call the parallel deviation chaotic characteristic perception and resonance threshold calibration module, load the pre-calculated typical flow field resonance threshold library, and, based on the target flow field type (e.g., aircraft nose shock wave calculation), initialize the effective resonance interval parameter deviation Lyapunov exponent 0.45-0.65 fluctuation main frequency difference ≤ 0.2Hz, and simultaneously initialize the weight coefficients for the resonance coupling coefficient calculation. =0.3、 =0.4、 =0.3, Initial value of bidirectional resonant gain =0.7, Time step safety factor =0.95 and other parameters, store all initialization parameters in the system cache; third, set the deviation acquisition frequency to be synchronized with the flow field calculation time step at a ratio of 1:1, set the resonance stability verification cycle to be verified once every 5 time steps, and set the trigger signal through the system timer.

[0087] 2. Step 2: Parallel computation to initiate dual sensing of flow field characteristics and deviations:

[0088] The specific operation is as follows: First, start the parallel calculation of the hypersonic flow field. The deviation and chaos parameter acquisition submodule synchronously collects the deviation amplitude fluctuation frequency and phase difference of each core at each time step. The chaotic characteristic extraction submodule calculates the deviation Lyapunov exponent in real time. The first step involves the acquisition and calculation of the dominant frequency of chaotic intensity fluctuations, which is synchronously triggered by the trigger signal set in step 1. The second step involves the flow field fluctuation characteristic sensing submodule synchronously acquiring the flow field shock wave intensity, vortex breaking frequency, and pressure gradient change rate, extracting the amplitude change rate A of the dominant frequency of flow field fluctuations, with the extraction process synchronized with the acquisition of deviation parameters. The third step involves the current set of deviation chaotic parameters. , With flow field fluctuation parameter set The data is transmitted to the resonant coupling control module and stored in the cache unit for backup.

[0089] 3. Step 3: Controllable Resonant Coupling Modulation

[0090] This is the core step, and the specific operations are as follows: First, the bidirectional resonance gain adjustment submodule calculates the difference between the current deviation and the dominant frequency of the flow field fluctuation. ,like ≤0.2Hz is already within the effective resonance range; maintain the current gain coefficient. =0.7; if >0.2Hz, the delay of the supercomputing core data interaction is adjusted by software within the range of 0.01-0.05ms. Specifically, this is achieved by modifying the delay parameters in the parallel communication protocol, so that the main frequency of the deviation fluctuation approaches the main frequency of the flow field fluctuation, until... ≤0.2Hz; if the adjustment still fails to meet the requirement, slightly thin the mesh in non-critical regions of the flow field with a thinning coefficient of 1.1, achieved by calling the mesh thinning interface, to reduce the dominant frequency of flow field fluctuations and ensure entry into the effective resonance range; if the current deviation is chaotic intensity > =0.7, the corrected gain coefficient is calculated using the above bidirectional resonant gain adjustment formula. Replace the original Secondly, the resonant coupling coefficient adaptation submodule is invoked to extract data from the cache unit. , , , and weighting coefficients , , Substituting into the formula for calculating the resonant coupling coefficient, we obtain... Value, Output The value is then transmitted to the cooperative instruction generation submodule.

[0091] 4. Step 4: Execution of resonant coupling coordination instructions:

[0092] This step achieves integrated execution without independent adjustments. The specific operations are as follows: First, if In the region ∈ [0.3, 0.6], characterized by moderate resonance and complex flow field, a cooperative command is executed, and the mesh refinement coefficient is set to 1.4 + 0.2 × ,for example When the value is 0.5, the encryption coefficient is 1.5, achieved by inputting the encryption coefficient and encryption area parameters through the grid encryption interface; CFL threshold adjustment threshold = 0.7 - 0.1 × =0.65, achieved by modifying the time step control parameters of the flow field calculation equation; the core calculation priority for complex regions is increased by 2 levels, achieved by adjusting the priority parameters of the core nodes through the supercomputing task scheduling module; simultaneously, the local time step coupling submodule extracts from the cache unit , , , Substituting into the local time step coupling formula, we obtain... First, adjust the local time step of the encrypted region to ensure that the phase difference with the deviation fluctuation is ≤0.1π; second, if Weak resonance in [0.1, 0.3), steady-state region of flow field, execute cooperative command, mesh thinning coefficient = 1.2 + 0.1 × (1 - ),for example When the coefficient is 0.2, the thinning coefficient is 1.28, achieved through the mesh thinning interface; CFL threshold adjustment threshold = 1.1 + 0.1 × (1 - =1.18, achieved by modifying the time step control parameter; balanced allocation of core resources, achieved by reallocating the computational workload of each core node through the task scheduling module; thirdly, if <0.1 or If the value is greater than 0.6, return to step 3 to readjust the resonance gain, but do not perform the adjustment for now.

[0093] 5. Step 5: Closed-loop resonance stability verification and parameter correction:

[0094] The specific operation is as follows: First, after every 5 time steps, the resonance stability verification submodule calculates the residual change rate and the flow field shock wave position error. If the residual change rate is ≤10... -3 If the error is ≤3%, resonance stability is determined, and calculation continues; if the residual change rate is >10%, then the calculation continues. -3 The divergence trend was observed, so the initial value of the resonance gain coefficient was reduced to 0.5, and the calculation was recalculated. The value is set and the coordination instruction is executed; if the error is >3%, the Lyapunov exponent threshold of the effective resonance interval is lowered by 0.05, for example, from 0.45-0.65 to 0.4-0.6, and the weighting coefficient in the resonance coupling coefficient formula is adjusted, such as by... The weighting of the influence of the flow field fluctuation amplitude change rate is adjusted to 0.45; secondly, the weighting coefficient of the corrected resonance threshold gain coefficient is fed back to the initialization parameters in step 1 to update the parameters in the system cache and complete the update of the computing environment.

[0095] 6. Iterate through step 6 until the calculation is complete:

[0096] Repeat steps 2 to 5 until the flow field calculation reaches the preset termination condition, such as the calculation time reaching the design value, the flow field evolution stabilizing, and the fluctuation amplitude change rate ≤ 0.05%. Stop the iterative calculation, output the final flow field calculation result and the optimized mesh parameters, and store the result in the supercomputing's designated storage unit.

[0097] Summarize:

[0098] The resonant gain effect of supercomputing parallel efficiency: Through controllable resonant coupling, the core resources of supercomputing are precisely matched with the flow field computing requirements, the core priority of complex regions is improved, and the resources of steady-state regions are released. The parallel computing efficiency is improved compared with existing technologies, and this efficiency improvement does not come from hardware upgrades or algorithm optimization, but from resource synergy gains achieved through resonant coupling.

[0099] Predictive and accurate capture of complex flow fields: Resonant coupling enables the early correlation between deviation fluctuations and flow field fluctuations, which is earlier than the judgment based on flow field physical parameters in existing technologies. This achieves predictive mesh densification for shock wave vortex breakup in complex regions, improves the accuracy of shock wave position calculation, and reduces the simulation error of vortex breakup evolution process.

[0100] The resonant self-stabilizing capability of the stability criterion: The CFL threshold is directly related to the resonant coupling coefficient, avoiding efficiency loss caused by fixed redundancy and the hysteresis of linear feedback. In the critical region of steady-state and unsteady-state transition of the flow field, the buffering effect of resonant coupling stabilizes the residual change rate at 10. -4 Within this range, there is no risk of divergence, achieving self-stabilizing adjustment of the stability criterion.

[0101] This approach fundamentally solves problems such as the failure of parallel synchronization deviation suppression, efficiency loss, adjustment lag, and divergence risk. First, there is no need to suppress deviations; deviations become the carrier of collaborative control. Second, there is no fixed redundancy; the CFL threshold adaptively matches the calculation requirements. Third, there is no adjustment lag; resonant coupling enables early prediction. Fourth, there is no divergence risk; the resonant self-stabilizing capability ensures computational stability, ultimately achieving the goal of efficient, accurate, and stable computation of hypersonic flow fields.

Claims

1. An intelligent grid control system for hypersonic flow field calculation, characterized in that, An integrated, deeply coupled architecture of resonance sensing, coupling control, collaborative adaptation, and closed-loop verification is adopted, including a parallel deviation chaotic characteristic sensing and resonance threshold calibration module, a flow field deviation resonance coupling control module, a resonance coupling collaborative adjustment core module, an intelligent grid dynamic resonance adaptation module, and a closed-loop resonance stability verification and optimization module. Each module is sequentially connected to form a collaborative control link. By capturing the chaotic characteristics of supercomputing parallel synchronization deviation, a controllable resonance coupling mechanism between it and flow field fluctuations is established. The resonance coupling characteristics are transformed into integrated collaborative control commands for grid parameters and CFL thresholds, which are used for efficient, accurate, and stable calculation of hypersonic flow fields.

2. The intelligent grid control system for hypersonic flow field calculation as described in claim 1, characterized in that: The parallel deviation chaotic characteristic perception and resonance threshold calibration module includes a deviation chaotic parameter acquisition submodule, a chaotic characteristic extraction submodule, and a deviation flow field resonance threshold calibration submodule. The deviation chaotic parameter acquisition submodule adopts a two-dimensional acquisition strategy of time step and core node, synchronously acquiring the deviation amplitude, fluctuation frequency, and phase difference of each supercomputing core at each time step, with the sampling frequency mapped 1:1 to the flow field calculation time step; the chaotic characteristic extraction submodule extracts three core parameters of deviation: chaotic intensity, fluctuation dominance frequency, and phase stability, based on the improved Lyapunov exponential algorithm; the deviation flow field resonance threshold calibration submodule establishes an effective resonance interval library of deviation chaotic parameters and flow field fluctuation parameters by pre-calculating typical hypersonic flow field cases.

3. The intelligent grid control system for hypersonic flow field calculation as described in claim 1, characterized in that: The flow field deviation resonance coupling control module includes a flow field fluctuation characteristic sensing submodule and a bidirectional resonance gain adjustment submodule; The flow field fluctuation characteristic sensing submodule synchronously collects the flow field shock wave intensity, vortex breaking frequency, and pressure gradient change rate, and extracts the dominant frequency and amplitude change rate of the flow field fluctuation; the bidirectional resonance gain adjustment submodule adopts a bidirectional gain adjustment design from deviation to flow field and from flow field to deviation, and achieves adaptive correction of the gain coefficient through the following formula: ； in This is the corrected bidirectional resonant gain coefficient. This is the bidirectional resonant gain coefficient before correction. This is the gain adjustment coefficient. The current chaos intensity is the deviation. This represents the upper limit of the chaos intensity corresponding to the effective resonance interval.

4. The intelligent grid control system for hypersonic flow field calculation as described in claim 1, characterized in that: The core module for resonant coupling coordinated regulation includes a resonant coupling coefficient adaptation submodule and a coordinated instruction generation submodule; the resonant coupling coefficient adaptation submodule calculates the resonant coupling coefficient using the following formula: ； in The resonant coupling coefficient is... , , The weighting coefficients for chaos intensity, rate of change of flow field fluctuation amplitude, and resonance gain are respectively, and satisfy the following conditions: The degree of chaos is the deviation. The rate of change of the amplitude of the flow field fluctuation. This is the bidirectional resonant gain coefficient. This represents the phase difference between the deviation fluctuation and the flow field fluctuation.

5. The intelligent grid control system for hypersonic flow field calculation as described in claim 1, characterized in that: The intelligent mesh dynamic resonance adaptation module includes a mesh resonance response submodule and a local time step coupling submodule; the local time step coupling submodule calculates the local time step of the mesh refinement region using the following formula: ； in For the local time step of the mesh densification region, Based on the time step, The resonant coupling coefficient is... This represents the phase difference between the deviation fluctuation and the flow field fluctuation. This represents the safety factor for the time step.

6. The intelligent grid control system for hypersonic flow field calculation as described in claim 1, characterized in that: The closed-loop resonance stability verification and optimization module includes a resonance stability verification submodule and a coupling parameter feedback correction submodule; the resonance stability verification submodule has a residual change rate ≤10 -3 Furthermore, a shock wave position error of ≤3% is used as the criterion for resonance stability; if the residual rate of change is >10... -3 Then reduce the resonance gain coefficient; if the error is greater than 3%, then correct the effective resonance interval threshold. The coupling parameter feedback correction submodule feeds back the verification results to the resonance threshold calibration submodule and the coupling coefficient adaptation submodule to correct the effective resonance interval threshold and the weight of the coupling coefficient calculation.

7. A smart grid control method for calculating hypersonic flow fields, characterized in that, With controllable resonant coupling as the core principle, the process includes the following steps: Step (1) Initialize resonance parameters and calculation environment, load typical flow field resonance threshold library, and set the deviation acquisition frequency and flow field calculation time step to be synchronized 1:1; Step (2) Start parallel flow field calculation and simultaneously collect parallel deviation chaos parameters and flow field fluctuation parameters; Step (3) Achieve controllable resonant coupling between deviation and flow field fluctuation through bidirectional gain adjustment, and calculate the resonant coupling coefficient; Step (4) Generate and execute integrated collaborative instructions for mesh parameters, CFL threshold, and core resource allocation based on the resonant coupling coefficient; Step (5) Verify the closed-loop resonance stability by comparing the residual change rate with the flow field parameter error, and correct the resonance parameters; Step (6) Iterate through steps (2) to (5) until the flow field calculation termination condition is met, and output the calculation results and mesh parameters.

8. The intelligent grid control method for hypersonic flow field calculation as described in claim 7, characterized in that: The parameters initialized in step (1) include: the effective resonance interval parameters are the deviation Lyapunov exponent of 0.45-0.65, the fluctuation main frequency difference ≤0.2Hz, and the weighting coefficient for calculating the resonance coupling coefficient. =0.3、 =0.4、 =0.3, initial value of bidirectional resonance gain =0.7, time step safety factor =0.95, resonance stability verification period is once every 5 time steps.

9. The intelligent grid control method for hypersonic flow field calculation as described in claim 7, characterized in that: In step (3), if the difference between the deviation and the main frequency of the flow field fluctuation is >0.2Hz, the main frequency of the deviation fluctuation is brought closer to the main frequency of the flow field fluctuation by adjusting the interaction delay of the supercomputing core data through software. If the adjustment is ineffective, the non-critical area mesh of the flow field is thinned with a thinning coefficient of 1.

1. When the deviation chaos intensity is ≥0.7, the gain coefficient is corrected by the bidirectional resonance gain adjustment formula.

Images