Unmeasurable parameter reconstruction method and system based on high-mode injection dynamics simulation and measured data fusion

Abstract

The application provides a method and system for reconstructing unmeasurable parameters based on high-mode injection dynamics simulation and measured data fusion, and belongs to the technical field of high-altitude simulation test of liquid rocket engines. The method comprises the following steps: constructing a high-mode injection full-system multi-physical-field dynamics simulation model, collecting external measurable parameters in the whole test process to form a measured data set, using the measured data as a driving boundary to calibrate the simulation model in a closed loop to obtain a calibrated simulation model, and using an algorithm for the fusion of ensemble Kalman filtering and four-dimensional variational assimilation to perform data assimilation and optimal inversion on internal implicit state variables to reconstruct unmeasurable parameters such as shock string position, Mach number distribution, wall heat flux and branch flow, and finally output a full-state data set covering the whole running cycle. The internal state of the high-mode injection test is transparently perceived without increasing the number of measuring points or modifying the hardware, which provides core data support for test state evaluation, fault positioning and safety warning.

Description

Technical Field

[0001] This invention relates to the field of high-altitude simulation test technology for liquid rocket engines, and in particular to a method and system for reconstructing unmeasurable parameters based on the fusion of high-mode ejector dynamics simulation and measured data. Background Technology

[0002] High-altitude simulation testing (referred to as high-altitude model testing) is an indispensable and crucial step in the development of rocket engines. Its core task is to create a near-high-altitude, low-pressure environment on ground facilities to accurately measure the engine's thrust, specific impulse, and other performance parameters under high-altitude operating conditions. For high-altitude engines using large-area nozzles, direct ground testing at sea level would cause flow separation within the nozzle, potentially damaging the nozzle structure. Therefore, testing must be conducted using a high-altitude simulation test stand. High-altitude model test stands are mainly divided into two types: passive ejector type and active ejector type. The passive ejector type relies on the ejection effect after the diffuser is activated to create a low-pressure environment, while the active ejector type relies on the active suction effect of the ejector to establish a low-pressure environment.

[0003] The internal flow of a high-mode ejector system is extremely turbulent, involving multi-physics coupling of multiple subsystems, including a steam generator, media delivery network, vacuum chamber, diffuser, and ejector. Inside the diffuser, there is a sudden bottom diffusion, reattached shock waves, and complex shock boundary layer interference, resulting in an extremely complex flow field structure. Currently, on-site testing typically only obtains externally measurable parameters such as total pressure, total temperature, total flow rate, vacuum pressure, and regulating valve opening. Simultaneously, the ejector mixing chamber, the diffuser interior, and the core flow field region of the vacuum chamber are subjected to extreme environments of high temperature, high pressure, and strong scouring. Sensors cannot survive for extended periods, and measurement points cannot be strategically placed, leading to critical internal parameters such as shock train location, mixing chamber Mach number distribution, wall heat flux density, and flow distribution in various branches of the pipeline network remaining "unmeasurable" for extended periods. This predicament of "external data, internal black box" prevents test personnel from accurately grasping the true internal physical state of the system, making fault location difficult and hindering in-depth analysis of the flow mechanism.

[0004] To address the aforementioned issues, existing technologies have included studies using numerical simulation to model the flow field in high-altitude simulation tests. For example, some literature uses FLUENT software to numerically calculate the diffuser in a high-altitude simulated ejection test of a liquid hydrogen / liquid oxygen rocket engine and compares the results with experimental measurements. Other studies employ two-dimensional axisymmetric Reynolds-averaged equations and the Spalart-Allmaras turbulence model to numerically simulate the flow field structure of a rocket engine high-altitude simulated test stand under different ejection schemes. Furthermore, some literature uses the NS equations and SA model to numerically simulate the pressure build-up, stabilization, and depressurization phases of a passive ejection high-altitude simulation test of a solid rocket engine. However, most of these simulation studies remain at the level of steady-state simulation of a single component or local flow field. There is a lack of a closed-loop calibration mechanism between the simulation model and the experimental data. The reliability of the simulation results depends on the accurate determination of the boundary conditions, which are themselves difficult to know precisely due to the lack of internal, unmeasurable parameters. While some studies have compared simulation results with experimental data, the simulation is only used as an independent forward calculation tool and has failed to achieve deep integration and two-way interaction with the experimental data.

[0005] In the field of data assimilation, ensemble Kalman filtering (EnKF) and four-dimensional variational assimilation (4DVar) have become mainstream methods for organically fusing physical models with observational data. The core idea of ​​data assimilation is to use measurement data to optimally estimate the state of a dynamic system, and it has been widely applied in meteorology, oceanography, hydrology, and other fields. In recent years, some scholars have also explored introducing data assimilation methods into fluid mechanics engineering problems, such as using data assimilation combined with physical models and observational data for turbulence reconstruction, and data assimilation of shock wave / boundary layer turbulence models for internal flows. However, the systematic application of data assimilation methods to the reconstruction of internal unmeasurable parameters of the entire high-mode ejection system of liquid rocket engines has not yet been publicly reported.

[0006] In summary, the core contradiction facing existing high-mode ejection testing technology lies in the fact that measurable parameters are limited to the external environment of the system, while a large number of key internal parameters are unavailable due to physical inaccessibility. Furthermore, existing simulation methods lack a closed-loop fusion mechanism with measured data, making it difficult to achieve high-precision inversion and reconstruction of internal, unmeasurable states. How to elevate high-mode ejection testing from "external characteristic observation" to "transparent perception of the entire internal state" without increasing measurement points or modifying hardware is a pressing technical challenge in this field. Summary of the Invention

[0007] To address the aforementioned problems, the present invention aims to provide a method and system for reconstructing unmeasurable parameters based on the fusion of high-mode ejection dynamics simulation and measured data.

[0008] The first aspect of this invention provides a method for reconstructing unmeasurable parameters based on the fusion of high-mode ejection dynamics simulation and measured data, comprising the following steps:

[0009] A multiphysics dynamics simulation model of the entire high-mode ejection system is constructed as a benchmark simulation model.

[0010] Real-time acquisition of external measurable parameters throughout the entire high-mode ejection experiment process to form a time-series dataset of measured external characteristics;

[0011] Using the measured external characteristic time series dataset as the driving boundary, the benchmark simulation model is subjected to parameter closed-loop calibration and state matching to obtain a calibration state simulation model equivalent to the measured environment;

[0012] Based on the measured external characteristic time series dataset, the internal implicit state variables of the calibration state simulation model are assimilated and optimal state inverted to reconstruct the optimal estimated state set of the physical field of the unmeasurable region inside the high-mode ejection system.

[0013] The optimal estimated state set of the physical field is mapped and output globally to form a fully transparent sensing dataset covering the entire operating cycle of the high-mode ejection experiment.

[0014] Furthermore, after collecting the external measurable parameters, the collected external measurable parameters are preprocessed. The preprocessing includes at least outlier removal, time-domain noise filtering, time axis alignment, and dimensional standardization to generate a standardized driving sequence synchronized with the time base of the benchmark simulation model. The standardized driving sequence is used as the measured external characteristic time series dataset for subsequent steps.

[0015] Furthermore, the high-mode ejector system includes a steam generator, a medium delivery pipeline network, a vacuum chamber, a diffuser, and an ejector; the benchmark simulation model has the ability to solve thermal, dynamic, and multi-physics coupled problems across scales.

[0016] Furthermore, the optimal estimation state set of the physical field includes at least one of the following: the spatial location of the shock train inside the ejector, the Mach number distribution in the mixing chamber, the pressure recovery gradient of the diffuser, the heat flux density on the wall, the local flow rate of each branch of the pipeline network, and the implicit flow field distribution inside the vacuum chamber.

[0017] Furthermore, the data assimilation employs a dual-constraint algorithm that combines ensemble Kalman filtering and four-dimensional variational assimilation, using measured data to sequentially update and globally smooth the system state variables.

[0018] Furthermore, the parameter closed-loop calibration specifically includes: correcting the pipeline impedance characteristics based on the total flow rate in the measured external characteristic time series dataset, correcting the ejector and diffuser performance curves based on the vacuum pressure, and correcting the actuator dynamic characteristics based on the control valve response.

[0019] Furthermore, the data assimilation and optimal state inversion include two operating modes:

[0020] In online real-time mode, the measured external characteristic time series dataset is sliced ​​into time windows, and the calibration state simulation model is driven to perform rolling state inversion window by window, and the optimal estimated state set of the physical field for the current time window is output in real time.

[0021] In offline tracing mode, the calibration state simulation model is globally iteratively assimilated using the full-cycle measured external characteristic time series dataset as the overall boundary, and the optimal estimated state set of the physical field for the entire cycle is output.

[0022] Furthermore, the reconstruction accuracy of the optimal estimated state set of the physical field satisfies the following conditions: relative error of pressure field ≤2%, relative error of temperature field ≤3%, and absolute error of shock wave characteristic position ≤5mm.

[0023] Furthermore, the fully transparent perception dataset is used for tracing and locating faults in high-mode ejection experiments, determining system performance boundaries, providing structural safety early warnings, and iteratively evolving the benchmark simulation model.

[0024] A second aspect of the present invention provides a system for reconstructing unmeasurable parameters based on the fusion of high-mode ejection dynamics simulation and measured data, comprising:

[0025] The dynamics simulation kernel module is used to build a multiphysics dynamics simulation model of the entire high-mode ejection system, which serves as a benchmark simulation model.

[0026] The measured data access module is used to collect external measurable parameters in real time throughout the entire process of the high-mode ejection experiment, forming a measured external characteristic time series dataset;

[0027] The model calibration fusion module is used to perform parameter closed-loop calibration and state matching on the benchmark simulation model using the measured external characteristic time series dataset as the driving boundary, so as to obtain a calibration state simulation model that is equivalent to the measured environment.

[0028] The parameter inversion and reconstruction module is used to perform data assimilation and optimal state inversion on the internal implicit state variables of the calibration state simulation model based on the measured external characteristic time series dataset, and reconstruct the optimal estimated state set of the physical field of the unmeasurable region inside the high-mode ejector system.

[0029] The visualization and storage module is used to perform global mapping and output of the optimal estimated state set of the physical field, forming a fully transparent perception dataset covering the entire operation cycle of the high-mode ejection experiment.

[0030] The present invention has the following advantages or beneficial effects:

[0031] By constructing a high-mode ejector full-system multiphysics dynamics simulation model covering the steam generator, medium transport pipeline network, vacuum chamber, diffuser, and ejector, and using external measurable data such as total pressure, total temperature, total flow rate, vacuum pressure, regulating valve opening, and timing signals collected throughout the experiment as driving boundaries, after timing alignment and parameter closed-loop calibration, a data assimilation algorithm that combines ensemble Kalman filtering and four-dimensional variational assimilation is used to perform optimal inversion of the internal implicit state variables. Finally, the output includes flow field, temperature field, pressure field, shock wave characteristics, and pipeline flow distribution data covering the entire operating cycle. Compared with traditional high-mode ejection test methods, this invention can obtain key internal parameters that are completely impossible to obtain by traditional methods, such as the position of the shock train inside the ejector, the Mach number distribution in the mixing chamber, the pressure recovery gradient of the diffuser, the heat flux density on the wall, the local flow rate of each branch of the pipeline network, and the flow field distribution inside the vacuum chamber, without adding any measuring points or modifying any test hardware. The reconstruction accuracy reaches ≤2% for pressure error, ≤3% for temperature error, and ≤5mm for shock position error, fundamentally solving the black box dilemma of high-mode ejection tests caused by the inability to arrange measuring points. Attached Figure Description

[0032] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0033] Figure 1 This is a flowchart illustrating a method for reconstructing unmeasurable parameters based on the fusion of high-mode ejection dynamics simulation and measured data, according to an exemplary embodiment.

[0034] Figure 2 This is a structural diagram of an unmeasurable parameter reconstruction system based on the fusion of high-mode ejection dynamics simulation and measured data, according to an exemplary embodiment.

[0035] Figure 3 This is a schematic diagram of the structure of an electronic device according to the present invention;

[0036] Figure 4 This is a schematic diagram of the structure of a computer-readable storage medium according to the present invention. Detailed Implementation

[0037] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that the invention will be thorough and complete, and the concept of the exemplary embodiments will be fully conveyed to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and therefore their detailed description will be omitted.

[0038] The terms “a,” “one,” “the,” and “the” are used to indicate the existence of one or more elements / components / etc.; the terms “including” and “having” are used to indicate an open-ended meaning of inclusion and to mean that there may be other elements / components / etc. in addition to the listed elements / components / etc.

[0039] Example 1

[0040] like Figure 1 As shown, this embodiment provides a method for reconstructing unmeasurable parameters of high-mode ejectors based on the fusion of simulation and measured data. It can invert and obtain key physical parameters that cannot be directly measured inside the system without adding measurement points or modifying the experimental hardware.

[0041] Step S100: First, a multiphysics dynamics simulation model of the entire high-mode ejector system is built as the baseline simulation model. This model covers the complete physical chain of the experimental system, specifically including five core subsystems: steam generator, medium delivery network, vacuum chamber, diffuser, and ejector. In the spatial dimension, the simulation model takes into account both the independent characteristics of each subsystem and the coupling relationship between subsystems. For example, the steam generator outlet parameters directly determine the ejector inlet boundary conditions, and the diffuser outlet back pressure is affected by the dynamic changes in vacuum chamber pressure. In terms of physical model selection, the baseline simulation model uses the Reynolds-averaged Navier-Stokes equations as the flow control equations, and a turbulence model that can better capture the characteristics of shock waves / boundary layer disturbances is selected. For the phase change process in the steam generator, a two-fluid model is used to describe the mass, momentum, and energy exchange between the gas and liquid phases. For the flow separation and reattachment phenomena that may occur inside the ejector, the model retains the ability to solve large separation flows. The above physical fields are solved through cross-scale multiphysics coupling through the transfer of physical property parameters. The benchmark simulation model has the ability to solve hot, dynamic and cross-scale coupled problems, and can fully reproduce the entire process of steam supply, flow distribution, ejector mixing, pressure build-up and diffusion recovery during the experiment.

[0042] The initial inputs to the benchmark simulation model include: the geometric parameters of the high-mode ejector system, initial environmental conditions, and experimental condition settings. After the benchmark simulation model is built, mesh independence verification is required to ensure that the simulation results do not depend on mesh density; at the same time, preliminary calibration is performed using existing historical experimental data to verify the predictive ability of the simulation model on key external characteristic parameters.

[0043] Step S200: Throughout the entire operation of the high-mode ejector test, various deployed sensors collect real-time external measurable parameters, forming a time-series dataset of measured external characteristics. These parameters include, but are not limited to: total steam pressure, total temperature, and total flow rate at the steam generator outlet; opening signals of regulating valves and valve action sequence switching signals on various media delivery pipelines; vacuum pressure inside the vacuum chamber; diffuser outlet pressure; and steam pressure and temperature before the ejector inlet. All collected data are accompanied by precise timestamps and recorded and stored in a unified data format.

[0044] After acquiring externally measurable parameters, the next step involves data preprocessing. Due to strong electromagnetic interference at the test site, sensor noise, and quantization errors in the data acquisition system, the raw acquired data inevitably contains outliers and random noise. Data preprocessing specifically includes: outlier removal for each channel's acquired data, identifying and removing outliers that significantly deviate from the normal range, and filling the resulting gaps through interpolation; using a low-pass digital filter to filter high-frequency noise in the signal and suppress noise interference; aligning the acquired data from different sensors on the time axis to eliminate time deviations introduced by asynchronous sampling clocks of different sensors, ensuring precise temporal correspondence between the data from each channel; and converting all physical quantities to standard International Units (SI) to ensure dimensional consistency and format standardization. After the above preprocessing, a standardized driving sequence fully synchronized with the time base of the benchmark simulation model is generated. This standardized driving sequence serves as the time-series dataset for the measured external characteristics in subsequent steps.

[0045] Step S300: Using the measured external characteristic time series dataset as the driving boundary, perform parameter closed-loop calibration and state matching on the benchmark simulation model to obtain a calibration-state simulation model equivalent to the measured environment. The essence of this step is to use measured external measurable data to correct key parameters with uncertainties in the benchmark simulation model, ensuring that the simulation model's external characteristic output is consistent with the measured data.

[0046] The specific implementation method of parameter closed-loop calibration is as follows: The total flow rate in the measured external characteristic time-series dataset is used as a benchmark to correct the pipeline network impedance characteristics. Pipeline network impedance characteristics are a key factor determining the pressure loss and flow distribution of steam from the steam generator through each branch to the ejector inlet. In the initial simulation model, the impedance coefficient of each branch is given by empirical formulas based on pipe geometry, number of bends, valve type, etc. However, actual installation conditions, pipe inner wall roughness, local resistance, and other factors can cause deviations between the actual impedance coefficient and the design value. During calibration, the values ​​of the impedance coefficients of each branch are adjusted so that the flow distribution of each branch calculated in the simulation and the total pipe flow rate simultaneously approach the measured values.

[0047] Secondly, the combined performance curves of the ejector and diffuser were corrected using the measured vacuum chamber pressure as a benchmark. The combined working characteristics of the ejector and diffuser determine the maximum mass flow rate that the ejector can pump and the pressure level that the diffuser can recover, which are the core factors determining whether the vacuum chamber pressure can reach the design value. During the calibration process, the pressure recovery coefficient of the diffuser and the efficiency coefficient of the ejector were jointly adjusted using the measured vacuum chamber pressure as a constraint, so that the vacuum chamber pressure obtained by simulation calculation is close to the measured value.

[0048] Secondly, the dynamic characteristics of the actuator are corrected based on the correspondence between the control valve opening command and the actual feedback signal. As the core actuator controlling steam flow, the actual response characteristics of the control valve (including action delay time, action speed, overshoot, etc.) differ from the ideal model. During calibration, the measured control valve opening timing signal is used as the driving input to correct parameters such as the time constant, dead zone width, and saturation limit value in the actuator model, ensuring that the simulated valve action characteristics match the actual response.

[0049] Through the aforementioned multi-parameter joint calibration, the simulation model not only matches the experimental measurements in terms of macroscopic external characteristics, but also makes the state variables of its internal subsystems closer to the real physical state due to the correction of model parameters, providing an accurate benchmark for subsequent data assimilation and parameter inversion. After calibration, a calibration-state simulation model equivalent to the experimental environment is obtained.

[0050] Step S400: Based on the measured external characteristic time series dataset, perform data assimilation and optimal state inversion on the internal implicit state variables of the calibration state simulation model to reconstruct the optimal estimated state set of the physical field of the unmeasurable region inside the high-mode ejector system.

[0051] This step utilizes a data assimilation algorithm to mathematically fuse discrete, finite measured external data with a continuous, high-dimensional simulation model, enabling optimal estimation of state variables within the system that are either unmeasurable or difficult to measure directly. The core idea of ​​data assimilation is to treat the dynamic evolution of the simulation model as a dynamic system. Measured data provides observational information about some of the system's state variables. The assimilation algorithm injects this observational information into the simulation model, allowing the model's internal state to approximate the true state as closely as possible while satisfying physical constraints.

[0052] In this implementation, data assimilation employs a dual-constraint algorithm that combines ensemble Kalman filtering and four-dimensional variational assimilation. Ensemble Kalman filtering is a sequential data assimilation method. Its basic principle is to represent the probability distribution of the state using a finite set of members. When new observation data arrives, the Kalman update formula is used to update the set members, thereby obtaining the optimal estimate of the system state. Its advantages include high computational efficiency, applicability to nonlinear systems, and the ability to provide uncertainty information in state estimation. Four-dimensional variational assimilation is a global optimization method. Its basic principle is to minimize the weighted distance between the simulated trajectory and the observation data within the entire time window by adjusting the initial conditions or parameters of the model within a time window. Its advantages include the ability to utilize all observation information within the time window, ensuring the continuity and smoothness of the reconstructed trajectory in the time dimension.

[0053] The integration of the two methods is as follows: Within each assimilation window, ensemble Kalman filtering is first used to sequentially update the system state, obtaining preliminary estimates of the state variables and their covariance matrices. Then, using the update results of the ensemble Kalman filtering as the initial guess field for four-dimensional variational assimilation, global iterative optimization is performed within the same time window. The covariance matrix of the ensemble Kalman filtering estimates is used as the weight constraint of the four-dimensional variational assimilation background error covariance matrix. Finally, the optimal estimated state after dual constraints is output. This dual-constraint architecture, combining sequential and global approaches, balances the computational efficiency advantages of ensemble Kalman filtering with the global consistency advantages of four-dimensional variational assimilation.

[0054] During state inversion, the selection of state variables includes key physical quantities of each subsystem: pressure, temperature, water level, and steam dryness fraction within the steam generator; pressure, temperature, and flow rate at each node of the pipeline network; pressure, temperature, velocity, and density distribution within the ejector; pressure recovery gradient within the diffuser; and velocity vector and vorticity distribution within the vacuum chamber. These state variables constitute the state vector for data assimilation. Among these, the state variables corresponding to directly measurable external parameters are used to construct the observation operator, serving as constraints for assimilation; while the internal state variables of the unmeasurable region are the target variables for assimilation and reconstruction.

[0055] Through the aforementioned data assimilation and state inversion process, unmeasurable parameters that cannot be obtained through traditional experiments can be reconstructed, including the position of the shock train inside the ejector, the Mach number distribution in the mixing chamber, the flow coefficient of the secondary throat, the pressure recovery gradient inside the diffuser, the wall heat flux density, the local steam flow rate in each branch, and the implicit recirculation zone and flow field distribution inside the vacuum chamber. The reconstruction accuracy meets the requirements of pressure error ≤2%, temperature error ≤3%, and shock wave position error ≤5mm.

[0056] Step S400: Based on the measured external characteristic time series dataset, perform data assimilation and optimal state inversion on the internal implicit state variables of the calibration state simulation model to reconstruct the optimal estimated state set of the physical field of the unmeasurable region inside the high-mode ejector system.

[0057] Data assimilation and optimal state inversion support two operating modes. In online real-time mode, the measured external characteristic time-series dataset is sliced ​​into time windows, driving the calibration-state simulation model to perform rolling state inversion window by window, and outputting the optimal physical field estimation state set for the current time window in real time. Specifically, within each time window, the measured data enters the assimilation system in the order of arrival, achieving the effect of reconstruction while experimenting. This mode is suitable for real-time state monitoring, rapid risk assessment, and operational decision support during the experiment. In offline traceability mode, the calibration-state simulation model is globally iteratively assimilated using the full-cycle measured external characteristic time-series dataset as the overall boundary, outputting the optimal physical field estimation state set for the entire cycle. This mode is suitable for post-experiment troubleshooting, fine performance evaluation, flow mechanism analysis, and final calibration of simulation model parameters.

[0058] Step S500: The optimal estimated state set of the physical field is mapped and output globally to form a fully transparent sensing dataset covering the entire operating cycle of the high-mode ejector experiment. Specifically, the physical quantities (pressure, temperature, velocity, Mach number, density, heat flux density, etc.) of the unmeasurable region inside the system obtained by data assimilation and state inversion are mapped globally according to spatial coordinates and time axis to form a unified format full-state dataset. This dataset covers the complete time history from experiment start-up, steady-state operation to shutdown, and the spatial scope covers the complete physical domain of the steam generator, piping network, ejector, diffuser, and vacuum chamber. This dataset contains reconstructed flow field, temperature field, pressure field, shock wave characteristics, and piping network flow distribution data, realizing the leap from external characteristic observation to internal fully transparent sensing of the high-mode ejector experiment.

[0059] The fully transparent sensing dataset generated in step S500 can be used in the following application scenarios: Source tracing and localization of experimental faults: When an anomaly occurs in a high-mode ejection experiment, the internal flow field evolution before and after the fault is replayed frame by frame using the complete physical field time history, accurately identifying the fault's starting position, propagation path, and development process, thus achieving accurate location of the root cause; System performance boundary determination: By summarizing and analyzing the reconstruction results under different experimental conditions, the system's performance envelope is plotted, providing a clear safety boundary reference for the design of subsequent experimental conditions; Structural safety early warning: When the reconstructed wall heat flux density or pressure load exceeds the design threshold, the system automatically triggers an early warning signal, prompting experimental personnel to pay attention to the structural safety of the corresponding parts; and Iterative evolution of the benchmark simulation model: The fully transparent sensing dataset obtained from each experimental reconstruction is fed back as a high-confidence virtual experimental sample to the benchmark simulation model's modeling process, used to recalibrate the coefficients and empirical parameters in the simulation model, enabling the simulation model to continuously evolve in repeated experiment-reconstruction-iteration cycles.

[0060] Example 1

[0061] like Figure 2 As shown, this embodiment provides a high-mode ejector unmeasurable parameter reconstruction system based on the fusion of simulation and measured data, including:

[0062] The dynamics simulation kernel module is used to build a multiphysics dynamics simulation model of the entire high-mode ejection system, which serves as a benchmark simulation model.

[0063] The measured data access module is used to collect external measurable parameters in real time throughout the entire process of the high-mode ejection experiment, forming a measured external characteristic time series dataset;

[0064] The model calibration fusion module is used to perform parameter closed-loop calibration and state matching on the benchmark simulation model using the measured external characteristic time series dataset as the driving boundary, so as to obtain a calibration state simulation model that is equivalent to the measured environment.

[0065] The parameter inversion and reconstruction module is used to perform data assimilation and optimal state inversion on the internal implicit state variables of the calibration state simulation model based on the measured external characteristic time series dataset, and reconstruct the optimal estimated state set of the physical field of the unmeasurable region inside the high-mode ejector system.

[0066] The visualization and storage module is used to perform global mapping and output of the optimal estimated state set of the physical field, forming a fully transparent perception dataset covering the entire operation cycle of the high-mode ejection experiment.

[0067] The specific methods by which the modules in the above embodiments perform their operations have been described in detail in Embodiment 1, and will not be elaborated upon here.

[0068] Example 3

[0069] like Figure 3 As shown, the present invention also provides an electronic device, including a processor 101, a communication interface 102, a memory 103, and a communication bus 104. The processor 101, communication interface 102, and memory 103 communicate with each other via the communication bus 104. The memory 103 is used to store computer programs. In this embodiment, when the processor 101 executes the program stored in the memory 103, it implements the steps of the method in Embodiment 1.

[0070] The electronic device provided in this embodiment of the invention has a similar implementation principle and technical effect to the above embodiments, and will not be described again here.

[0071] The aforementioned memory 103 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 103 has storage space for program code used to perform any of the method steps described above. For example, the storage space for program code may include individual program codes for implementing the various steps in the methods described above. This program code can be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, optical discs (CDs), memory cards, or floppy disks. Such computer program products are typically portable or fixed storage units. The storage unit may have storage segments or storage spaces arranged similarly to memory 103 in the aforementioned electronic device. The program code may be compressed, for example, in a suitable form. Typically, the storage unit includes programs for performing the method steps according to embodiments of the invention, i.e., code that can be read by a processor such as 101, which, when run by the electronic device, causes the electronic device to perform the various steps in the methods described above.

[0072] Example 4

[0073] like Figure 4 As shown, the present invention also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method described above.

[0074] The computer-readable storage medium may be included in the device / apparatus described in the above embodiments; or it may exist independently and not assembled into the device / apparatus. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of the present invention.

[0075] According to embodiments of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0076] In this embodiment of the invention, the term "multiple" refers to two or more, unless otherwise explicitly defined. The terms "install," "connect," and "fix" should be interpreted broadly. For example, "connect" can mean a fixed connection, a detachable connection, or an integral connection. Those skilled in the art can understand the specific meaning of the above terms in this embodiment of the invention based on the specific circumstances.

[0077] In the description of the embodiments of the present invention, it should be understood that the terms "upper" and "lower" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the device or unit referred to must have a specific orientation or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of the present invention.

[0078] In the description of this specification, the terms "an embodiment," "a preferred embodiment," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0079] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. For those skilled in the art, the embodiments of the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of the present invention should be included within the protection scope of the embodiments of the present invention.

Claims

1. A method for reconstructing unmeasurable parameters based on the fusion of high-mode ejection dynamics simulation and measured data, characterized in that, Includes the following steps: A multiphysics dynamics simulation model of the entire high-mode ejection system is constructed as a benchmark simulation model. Real-time acquisition of external measurable parameters throughout the entire high-mode ejection experiment process to form a time-series dataset of measured external characteristics; Using the measured external characteristic time series dataset as the driving boundary, the benchmark simulation model is subjected to parameter closed-loop calibration and state matching to obtain a calibration state simulation model equivalent to the measured environment; Based on the measured external characteristic time series dataset, the internal implicit state variables of the calibration state simulation model are assimilated and optimal state inverted to reconstruct the optimal estimated state set of the physical field of the unmeasurable region inside the high-mode ejection system. The optimal estimated state set of the physical field is mapped and output globally to form a fully transparent sensing dataset covering the entire operating cycle of the high-mode ejection experiment.

2. The method according to claim 1, characterized in that, After acquiring the external measurable parameters, the acquired external measurable parameters are preprocessed. The preprocessing includes at least outlier removal, time-domain noise filtering, time axis alignment, and dimensional standardization to generate a standardized driving sequence synchronized with the time base of the benchmark simulation model. The standardized driving sequence is used as the measured external characteristic time series dataset for subsequent steps.

3. The method according to claim 1, characterized in that, The high-mode ejector system includes a steam generator, a medium delivery pipeline network, a vacuum chamber, a diffuser, and an ejector; the benchmark simulation model has the ability to solve thermal, dynamic, and multi-physics coupled problems across scales.

4. The method according to claim 1, characterized in that, The optimal estimation state set of the physical field includes at least one of the following: the spatial location of the shock train inside the ejector, the Mach number distribution in the mixing chamber, the pressure recovery gradient of the diffuser, the heat flux density on the wall, the local flow rate of each branch of the pipeline network, and the implicit flow field distribution inside the vacuum chamber.

5. The method according to claim 1, characterized in that, The data assimilation adopts a dual-constraint algorithm that combines ensemble Kalman filtering and four-dimensional variational assimilation, using measured data to sequentially update and globally smooth the system state variables.

6. The method according to claim 1, characterized in that, The parameter closed-loop calibration specifically includes: correcting the pipeline impedance characteristics based on the total flow rate in the measured external characteristic time series dataset, correcting the ejector and diffuser performance curves based on the vacuum pressure, and correcting the dynamic characteristics of the actuator based on the control valve response.

7. The method according to claim 1, characterized in that, The data assimilation and optimal state inversion include two operating modes: In online real-time mode, the measured external characteristic time series dataset is sliced ​​into time windows, and the calibration state simulation model is driven to perform rolling state inversion window by window, and the optimal estimated state set of the physical field for the current time window is output in real time. In offline tracing mode, the calibration state simulation model is globally iteratively assimilated using the full-cycle measured external characteristic time series dataset as the overall boundary, and the optimal estimated state set of the physical field for the entire cycle is output.

8. The method according to claim 1, characterized in that, The reconstruction accuracy of the optimal estimated state set of the physical field satisfies the following conditions: relative error of pressure field ≤2%, relative error of temperature field ≤3%, and absolute error of shock wave characteristic position ≤5mm.

9. The method according to claim 1, characterized in that, The fully transparent perception dataset is used for tracing and locating faults in high-mode ejection experiments, determining system performance boundaries, providing structural safety early warnings, and iteratively evolving the benchmark simulation model.

10. A system for reconstructing unmeasurable parameters based on the fusion of high-mode ejection dynamics simulation and measured data, characterized in that, include: The dynamics simulation kernel module is used to build a multiphysics dynamics simulation model of the entire high-mode ejection system, which serves as a benchmark simulation model. The measured data access module is used to collect external measurable parameters in real time throughout the entire process of the high-mode ejection experiment, forming a measured external characteristic time series dataset; The model calibration fusion module is used to perform parameter closed-loop calibration and state matching on the benchmark simulation model using the measured external characteristic time series dataset as the driving boundary, so as to obtain a calibration state simulation model that is equivalent to the measured environment. The parameter inversion and reconstruction module is used to perform data assimilation and optimal state inversion on the internal implicit state variables of the calibration state simulation model based on the measured external characteristic time series dataset, and reconstruct the optimal estimated state set of the physical field of the unmeasurable region inside the high-mode ejector system. The visualization and storage module is used to perform global mapping and output of the optimal estimated state set of the physical field, forming a fully transparent perception dataset covering the entire operation cycle of the high-mode ejection experiment.

Images