A method, device and equipment for reducing order of Mworks one-dimensional system simulation model

Abstract

The application discloses a method, device and equipment for reducing the order of an Mworks one-dimensional system simulation model, and relates to the technical field of engine one-dimensional system simulation, to solve the problem of low simulation efficiency of the Mworks one-dimensional system simulation model in the prior art. The method comprises the following steps: obtaining a plurality of target design variables in the Mworks one-dimensional system simulation model; the plurality of target design variables are design variables with an influence degree on a rocket engine greater than a preset threshold, and at least comprise an air inlet pressure, fuel injection quantity and rotating speed of the rocket engine; performing dynamic simulation calculation on the plurality of target design variables to obtain a plurality of target physical quantity data; constructing a dynamic reduced-order characteristic model based on the plurality of target design variables and the plurality of target physical quantity data; and integrating the dynamic reduced-order characteristic model into the Mworks one-dimensional system simulation model to obtain an Mworks one-dimensional system simulation reduced-order model; and the simulation efficiency of the Mworks one-dimensional system simulation model is improved.

Description

Technical Field

[0001] This invention relates to the field of engine one-dimensional system simulation technology, and in particular to a method, apparatus and equipment for reducing the order of a Mworks one-dimensional system simulation model. Background Technology

[0002] In the design and development of liquid rocket engines, Mworks one-dimensional system simulation models are widely used to simulate key physical phenomena such as fluid flow, heat transfer, and combustion processes inside the engine.

[0003] In existing technologies, when using Mworks one-dimensional system simulation models to obtain key physical quantities (such as temperature, pressure, and flow rate), the empirical formulas of Mworks are usually used to perform simulation calculations to obtain sample data. This results in excessive time delays in obtaining key physical quantities of the engine, and the amount of data in the rocket engine simulation test process is also large. Statistics show that the simulation calculation time of the existing Mworks one-dimensional system simulation model is about 5 times slower than the actual rocket engine test time. The simulation efficiency is low, the delay is too large, and it cannot meet the real-time data feedback requirements of digital twin technology for physical test rocket engines.

[0004] Therefore, there is an urgent need to establish an advanced technology that can improve the simulation efficiency of Mworks one-dimensional system simulation models in order to solve the problem of low simulation efficiency of Mworks one-dimensional system simulation models in the existing technology. Summary of the Invention

[0005] The purpose of this invention is to provide a method, apparatus, and device for reducing the order of a one-dimensional Mworks system simulation model. It establishes a dynamic order-reduced model by utilizing the correspondence between key physical quantities of a rocket engine and their design variables, and integrates this dynamic order-reduced model into the one-dimensional Mworks system simulation model. This results in a reduced-order Mworks system simulation model, where the output of results can be obtained through data correlation instead of the empirical formula calculations used in the original Mworks one-dimensional system simulation model. This significantly improves simulation efficiency and solves the problem of low simulation efficiency in existing Mworks one-dimensional system simulation models.

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

[0007] In a first aspect, the present invention provides a method for reducing the order of a one-dimensional Mworks system simulation model, which may include:

[0008] Obtain multiple target design variables from the Mworks one-dimensional system simulation model; the multiple target design variables are design variables whose influence on the rocket engine is greater than a preset threshold; the multiple target design variables include at least the rocket engine's inlet pressure, fuel injection quantity, and rotational speed;

[0009] Dynamic simulation calculations are performed on multiple target design variables to obtain multiple target physical quantity data;

[0010] A dynamic order reduction feature model is constructed based on multiple target design variables and multiple target physical quantity data.

[0011] The dynamic reduced-order feature model is integrated into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model.

[0012] Preferably, the step of performing dynamic simulation calculations on multiple target design variables to obtain multiple target physical quantity data may include, prior to:

[0013] Multiple target design variables are sampled according to a preset target experimental design method to obtain multiple sample point data; the target experimental design method includes orthogonal experimental design method, full factorial design method or response surface design method; the multiple sample point data are data that can cover the target experimental design method; the number of multiple sample point data is less than the number of multiple target design variables;

[0014] The dynamic simulation calculation of multiple target design variables to obtain multiple target physical quantity data may include:

[0015] The Mworks one-dimensional simulation model is used to perform simulation calculations on multiple sample point data to obtain multiple physical quantity data corresponding to the multiple sample point data; the multiple physical quantity data are used as multiple target physical quantity data.

[0016] Preferably, the step of constructing a dynamic order reduction feature model based on multiple target design variables and multiple target physical quantity data includes:

[0017] A sample database is constructed based on multiple sample point data and multiple target physical quantity data; a dynamic reduced-order feature model is constructed based on the sample database.

[0018] Preferably, the step of constructing a dynamic reduced-order feature model based on the sample database may include: performing data preprocessing on the data in the sample database, wherein the data preprocessing includes at least one or more of data cleaning, data normalization, and data deduplication.

[0019] Preferably, constructing a dynamic reduced-order feature model based on the sample database may include:

[0020] Multiple feature information of multiple target physical quantity data are extracted from the sample database according to a preset feature extraction strategy; the preset feature extraction strategy includes sliding window processing and / or wavelet transform on the time axis.

[0021] Based on the aforementioned feature information, multiple feature vectors are constructed;

[0022] Based on the multiple feature vectors, the network model is trained to obtain the dynamic reduced-order feature model.

[0023] Preferably, the step of training the network model based on multiple feature vectors to obtain the dynamically reduced-order feature model may include:

[0024] Based on multiple feature vectors, the network model is trained to obtain a first dynamic order reduction model; the prediction performance of the first dynamic order reduction model is tested using a test set to obtain performance test results; the test set is a sample dataset calculated using the Mworks one-dimensional simulation model.

[0025] If the performance test results meet the preset prediction accuracy, the first dynamic reduction model is used as the dynamic reduction feature model; if the performance test results do not meet the preset prediction accuracy and generalization ability, the model parameters are adjusted or samples are re-collected to continue model training until a second dynamic reduction model that meets the preset prediction accuracy is obtained.

[0026] Preferably, constructing multiple feature vectors based on multiple feature information may include: combining multiple feature information in a predetermined order to obtain multiple feature vectors.

[0027] Preferably, integrating the dynamic reduced-order feature model into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model may include:

[0028] The dynamic reduced-order feature model is integrated into the Mworks one-dimensional system simulation model according to the preset model integration method to obtain the Mworks one-dimensional system simulation reduced-order model; the preset model integration method includes at least one of voting, averaging, stacking, bagging and lifting.

[0029] Secondly, the present invention provides a device for reducing the order of a one-dimensional Mworks system simulation model, which may include:

[0030] The acquisition module is used to acquire multiple target design variables in the Mworks one-dimensional system simulation model; the multiple target design variables are design variables whose influence on the rocket engine is greater than a preset threshold; the multiple target design variables include at least the rocket engine's intake pressure, fuel injection quantity, and rotational speed;

[0031] The module is used to perform dynamic simulation calculations on multiple target design variables to obtain multiple target physical quantity data.

[0032] The construction module is used to construct a dynamic reduced-order feature model based on multiple target design variables and multiple target physical quantity data;

[0033] An integration module is used to integrate the dynamic reduced-order feature model into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model.

[0034] Thirdly, the present invention provides a device for reducing the order of a one-dimensional Mworks system simulation model, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it can implement the method for reducing the order of a one-dimensional Mworks system simulation model as described in the first aspect.

[0035] Compared with existing technologies, this invention provides a method for reducing the order of a one-dimensional Mworks system simulation model. This method involves obtaining multiple target design variables from the Mworks one-dimensional system simulation model. These target design variables are those whose impact on the rocket engine exceeds a preset threshold. The target design variables include at least the rocket engine's inlet pressure, fuel injection quantity, and rotational speed. Further, dynamic simulation calculations are performed on these target design variables to obtain multiple target physical quantity data. Based on these target design variables and the target physical quantity data, a dynamic order reduction feature model is constructed. Finally, the dynamic order reduction feature model is integrated into the Mworks one-dimensional system simulation model to obtain the Mworks... A reduced-order model for one-dimensional system simulation is proposed. Based on this, a dynamic reduced-order model is established by utilizing the correspondence between key physical quantities of rocket engines and their design variables. This dynamic reduced-order model is then integrated into the Mworks one-dimensional system simulation model to obtain the reduced-order Mworks one-dimensional system simulation model. This achieves the replacement of the empirical formula calculation method in the original Mworks model with a data association processing method. The output results can be obtained by replacing the empirical formula calculation method in the original Mworks one-dimensional system simulation model with the output results through data association, thus obtaining dynamically changing key physical quantities (simulation results). This greatly improves simulation efficiency and solves the problem of low simulation efficiency in the existing Mworks one-dimensional system simulation model. Attached Figure Description

[0036] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0037] Figure 1 A schematic diagram of the main process of a method for reducing the order of a one-dimensional Mworks system simulation model provided by the present invention;

[0038] Figure 2 This is a schematic diagram of a reduction device for a one-dimensional Mworks system simulation model provided by the present invention.

[0039] Figure 3 This is a schematic diagram of the hardware structure of a device for reducing the order of a one-dimensional system simulation model in Mworks, as provided by the present invention. Detailed Implementation

[0040] To facilitate a clear description of the technical solutions in the embodiments of the present invention, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. For example, the first threshold and the second threshold are merely used to distinguish different thresholds and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0041] It should be noted that in this invention, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0042] In this invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding related objects have an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, a combination of a and b, a combination of a and c, a combination of b and c, or a, b, and c, where a, b, and c can be single or multiple.

[0043] Currently, MWorks one-dimensional system simulation models are widely used to simulate key physical phenomena such as fluid flow, heat transfer, and combustion processes within engines. In MWorks system simulations, each component of a liquid rocket engine is modeled using empirical mathematical formulas. Because the core components of a liquid rocket engine are precise and complex, the resulting models are also very complex, requiring a considerable amount of time for computation during model execution. Although MWorks' empirical formula models are much faster than finite element models, they are still significantly slower than actual rocket engine operation. Statistics show that, on average, MWorks simulations are about five times slower than the actual operating speed of liquid rocket engines currently in use. For example, if a liquid rocket engine's A-condition requires 10 seconds in a physical test, the existing MWorks one-dimensional system simulation model using empirical formulas would only take about one minute to simulate A-condition. This cannot meet the real-time data feedback requirements of digital twin technology during physical rocket engine testing.

[0044] It should be noted that Mworks' empirical formulas are approximate expressions derived from experiments, used to describe the relationships between variables. These formulas are typically based on certain conditions and obtained through the analysis of experimental data, and are significant for both practice and theory. Empirical formulas can help us understand and predict the patterns of change between variables. Although they may not be exact mathematical models, in many cases they provide a sufficiently accurate description, allowing us to predict unknown outcomes based on known data.

[0045] In view of this, the present invention provides a method, apparatus and equipment for reducing the order of a one-dimensional Mworks system simulation model, which is applied to a digital twin platform to solve the problems of excessive delay in obtaining key physical quantities of the engine and low simulation efficiency in the existing Mworks one-dimensional system simulation model, and to meet the real-time data feedback requirements of digital twin technology for physical testing of rocket engines.

[0046] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings:

[0047] Please see Figure 1 , Figure 1 This is a schematic diagram of the main process of a method for reducing the order of a one-dimensional Mworks system simulation model provided by the present invention. The executing entity is a service platform or terminal device equipped with the technical solutions disclosed in the embodiments of the present invention, such as a digital twin platform or a handheld simulation device, etc.

[0048] exist Figure 1 The method may include:

[0049] Step 110: Obtain multiple target design variables from the Mworks one-dimensional system simulation model; the multiple target design variables are design variables whose influence on the rocket engine is greater than a preset threshold; the multiple target design variables include at least the rocket engine's intake pressure, fuel injection quantity, and rotational speed.

[0050] Step 120: Perform dynamic simulation calculations on multiple target design variables to obtain multiple target physical quantity data.

[0051] Step 130: Construct a dynamic reduced-order feature model based on multiple target design variables and multiple target physical quantity data.

[0052] In steps 110 to 130, the aim is to significantly shorten simulation time and improve the real-time performance of data acquisition by constructing a dynamic reduced-order feature model, thereby meeting the requirements of digital twin technology. The system first analyzes the design variables affecting key physical quantities in the one-dimensional engine simulation model, such as intake pressure, fuel injection quantity, or engine speed, and determines the reasonable value range of each design variable. Further, it performs dynamic simulation calculations on the target design variables to obtain multiple target physical quantity data. A preset number of sample points can be selected within the value range of the design variables, as long as they cover the design space. Finally, based on the correspondence between the target design variables and the target physical quantities, the network model is trained to construct the dynamic reduced-order feature model.

[0053] Step 140: Integrate the dynamic reduced-order feature model into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model.

[0054] In step 140, the constructed dynamic reduced-order feature model is integrated into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model, thus realizing the reduction of the order of the Mworks one-dimensional system simulation model. Based on this, when using the Mworks one-dimensional system simulation reduced-order model to obtain the key physical quantities of the engine, it is not necessary to perform complex empirical mathematical formula calculations. Instead, the key physical quantities of the engine are obtained by using the data association relationship output method (pure data processing) to drive the real-time visualization of the engine physical field on the digital twin platform.

[0055] Based on this, the present invention provides a method for reducing the order of a one-dimensional Mworks system simulation model. This method establishes a dynamic order-reduced model by utilizing the correspondence between key physical quantities of a rocket engine and their design variables, and integrates this dynamic order-reduced model into the one-dimensional Mworks system simulation model. This results in a reduced-order Mworks system simulation model, replacing the empirical formula calculation method in Mworks with a pure data processing approach. Simulation results are obtained directly through the correspondence between key physical quantities and their design variables, significantly improving simulation efficiency and solving the problem of low simulation efficiency in existing one-dimensional Mworks system simulation models. It also meets the real-time data feedback requirements of digital twin technology during physical testing of rocket engines.

[0056] Preferably, before step 120, i.e., before performing dynamic simulation calculations on the multiple target design variables to obtain multiple target physical quantity data, the process may include:

[0057] Multiple target design variables are sampled according to a preset target experimental design method to obtain multiple sample point data; the target experimental design method includes orthogonal experimental design method, full factorial design method or response surface design method; the multiple sample point data are data that can cover the target experimental design method; the number of multiple sample point data is less than the number of multiple target design variables.

[0058] Furthermore, dynamic simulation calculations are performed on multiple target design variables to obtain multiple target physical quantity data, which may include:

[0059] The Mworks one-dimensional simulation model is used to perform simulation calculations on multiple sample point data to obtain multiple physical quantity data corresponding to the multiple sample point data; the multiple physical quantity data are used as multiple target physical quantity data.

[0060] Specifically, sampling can be performed using experimental design methods such as orthogonal experimental design (OAT), full factorial design (FFD), or response surface design (RSM). A preset number of sample points can be reasonably selected within the range of design variable values ​​to improve sampling efficiency. For example, if inlet pressure is collected, this pressure can be used to cover the design space. The preset number of sample points represents the minimum number of sample points that can cover the design space. This preset number can be obtained by analyzing historical experimental data or set based on the experience of R&D personnel. No specific limitation is made in this invention.

[0061] It's important to note that orthogonal experimental design (OAT) is a method for studying multi-factor, multi-level experiments. Based on orthogonality, it selects representative points from a comprehensive experiment for further testing. These representative points are characterized by uniform distribution and comparable characteristics. Orthogonal experimental design is a primary method in fractional factorial design. When the experiment involves three or more factors, and there may be interactions between them, the workload becomes very large, even difficult to implement. Orthogonal experimental design is undoubtedly a better choice to address this challenge. The main tool in orthogonal experimental design is the orthogonal array. Experimenters can find the appropriate orthogonal array based on the number of factors, the number of levels of each factor, and whether there are interactions. Then, relying on the orthogonality of the array, representative points are selected from the comprehensive experiment for further testing. This allows for results equivalent to a large number of comprehensive experiments with the fewest possible trials. Therefore, using orthogonal arrays to design experiments is an efficient, fast, and economical method for multi-factor experimental design.

[0062] Full factorial design (FFD) is a method to overcome the interaction problem in single-factor experiments. It calculates all possible combinations of factors in an experiment. This method is called a full factorial experiment. It arranges all possible combinations of factor levels in the experiment. If there are n factors and each factor has 2 levels, such a full factorial experiment is called a 2n factorial experiment.

[0063] Response surface methodology (RSM) is a statistical experimental method for optimizing stochastic processes. Its goal is to find quantitative relationships between experimental indicators and factors, and to identify the optimal combination of factor levels. It actively collects data based on multiple linear regression to obtain regression equations with good properties. The resulting complex multidimensional spatial surfaces closely approximate real-world conditions, require relatively few experimental groups, and are widely used in simulations and system dynamics.

[0064] Furthermore, in step 130, the construction of a dynamic order reduction feature model based on multiple target design variables and multiple target physical quantity data may include: constructing a sample database based on multiple sample point data and multiple target physical quantity data; and constructing a dynamic order reduction feature model based on the sample database.

[0065] Preferably, to ensure data quality, before constructing a dynamic reduced-order feature model based on the sample database, the process may include: performing data preprocessing on the data in the sample database, wherein the data preprocessing includes at least one or more of data cleaning, data normalization, and data deduplication.

[0066] Specifically, data cleaning involves processing and manipulating data to make it suitable for analysis and modeling. Data cleaning includes operations such as removing duplicate data, filling in missing values, handling outliers, and transforming data formats to improve the accuracy and reliability of the data. Data cleaning is usually a necessary step in the data processing process; it can eliminate data errors and noise and improve the accuracy of analysis and modeling.

[0067] Data normalization is a common data preprocessing technique used to convert data into a uniform scale range in order to eliminate dimensional differences between different variables.

[0068] Data deduplication is the process of removing duplicate data records from a dataset and retaining unique data records; this can avoid redundant calculations, reduce storage space usage, and improve data processing efficiency.

[0069] Preferably, the step of constructing a dynamic reduced-order feature model based on the sample database may include: extracting multiple feature information of multiple target physical quantity data from the sample database according to a preset feature extraction strategy; the preset feature extraction strategy includes sliding window processing and / or wavelet transform on the time axis.

[0070] It should be noted that by using feature extraction methods such as sliding window processing or wavelet transform on the time axis to extract the feature information of key physical quantities and construct feature vectors, the temporal characteristics and frequency domain information of the data can be enhanced.

[0071] Furthermore, multiple feature vectors can be constructed based on multiple feature information.

[0072] Specifically, multiple feature information can be combined in a predetermined order to obtain multiple feature vectors. For example, the extracted temperature feature can be combined into a vector in a certain order to obtain a temperature feature vector; the extracted pressure feature can be combined into a vector in a certain order to obtain a pressure feature vector; the extracted flow rate feature can be combined into a vector in a certain order to obtain a flow rate feature vector, and so on.

[0073] Furthermore, based on multiple feature vectors, the network model is trained to obtain the dynamic reduced-order feature model.

[0074] Preferably, the step of training the network model based on multiple feature vectors to obtain the dynamic reduced-order feature model may include: training the network model based on multiple feature vectors to obtain a first dynamic reduced-order model.

[0075] For example, machine learning algorithms can be used to train feature vectors and construct a first dynamic reduced-order model; the machine learning algorithms can include multilayer perceptron (MLP), support vector machine (SVM), and neural networks, etc.

[0076] The predictive performance of the first dynamic reduction model is tested using a test set, which is a sample dataset calculated using a Mworks one-dimensional simulation model. If the performance test results meet the preset prediction accuracy, the first dynamic reduction model is used as the dynamic reduction feature model. If the performance test results do not meet the preset prediction accuracy and generalization ability, the model parameters are adjusted or samples are re-collected to continue model training until a second dynamic reduction model that meets the preset prediction accuracy is obtained.

[0077] Specifically, the trained dynamic order reduction model can be validated using an independent validation dataset to evaluate its prediction accuracy, error, and simulation efficiency. If the evaluation results do not meet the requirements, the model parameters are adjusted or sample collection and training are repeated; if the requirements are met, the dynamic order reduction feature model to be established in this invention is obtained. The dynamically order reduction model constructed in this way can significantly shorten simulation time compared to the existing Mworks one-dimensional simulation model, meeting the real-time data requirements of digital twin technology. Furthermore, the dynamic order reduction model is trained based on machine learning algorithms, thus accurately capturing the dynamic changes of key engine physical quantities and improving the accuracy of data acquisition; thereby supporting rapid iterative optimization of engine design parameters and improving product development efficiency and quality.

[0078] It should be noted that the sample data in both the training and test sets can be constructed using the data from the original surrogate model in Mworks. After training the network model on the training set to obtain the dynamic order reduction model, the test set is used to test the trained dynamic order reduction model and compare it with the output results of the original Mworks model to determine accuracy, error, and simulation efficiency.

[0079] Further, in step 140, integrating the dynamic reduced-order feature model into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model may include:

[0080] The dynamic reduced-order feature model is integrated into the Mworks one-dimensional system simulation model according to the preset model integration method to obtain the Mworks one-dimensional system simulation reduced-order model; the preset model integration method includes at least one of voting, averaging, stacking, bagging and lifting; the lifting method is preferred in this invention for model integration.

[0081] It should be noted that the following is a description of the voting, averaging, stacking, bagging, and boosting methods in model ensemble methods:

[0082] (I) Voting Method

[0083] 1. Absolute majority voting method: A certain mark receives more than half the votes.

[0084] 2. Relative majority voting method: The marker that is predicted to receive the most votes is selected. If multiple markers have the most votes, one of them is randomly selected.

[0085] 3. Weighted voting method: Provides prediction results, similar to the weighted average method.

[0086] (II) Average Method

[0087] Simple average, weighted average.

[0088] Applicable scope: For large-scale ensembles with many learning weights, the weighted average method is prone to overfitting; the weighted average method is suitable when the performance of individual learners differs greatly, and the simple average method is suitable when they are similar.

[0089] (III) Stacking method

[0090] The stacking method combines the predictions of multiple base learners as input to a meta-learner, which then provides the final prediction.

[0091] (iv) Bagged packaging method

[0092] The bagging method improves overall performance and reduces overfitting by randomly sampling from the training set, training multiple models, and then averaging or voting on the predictions of these models. Random Forest is one of the most commonly used bagging algorithms.

[0093] (V) Enhancement Method

[0094] Boosting algorithms improve overall performance by progressively training multiple weak learners, each attempting to correct the errors of the previous learner. Common boosting algorithms include AdaBoost and Gradient Boosting.

[0095] Specifically, the trained reduced-order feature model can be integrated into the Mworks one-dimensional simulation model to obtain the reduced-order Mworks one-dimensional system simulation model, thereby reducing the order of the Mworks one-dimensional simulation model and meeting the real-time data feedback requirements of rocket engine physical testing in digital twin technology. Applying this reduced-order Mworks one-dimensional system simulation model to the rocket engine digital twin system can monitor and predict key physical quantities of the engine in real time, supporting real-time data analysis and optimization design.

[0096] Based on this, the present invention provides a method for reducing the order of a one-dimensional Mworks system simulation model. This method first determines the range of design variables, then uses the Design of Experiments (DOE) method for sampling, performs simulation calculations on the sample points, and constructs a sample database. The sample database is then used to train a feature model, thereby realizing a machine learning-based method for order reduction and prediction to establish a one-dimensional feature model. This significantly improves simulation efficiency and meets the real-time data requirements of digital twin technology. It also solves the problem of low simulation efficiency caused by excessive time delay in acquiring key engine physical quantities in existing Mworks one-dimensional system simulation models. Furthermore, the prediction model obtained based on machine learning algorithms can accurately capture the dynamic changes of key engine physical quantities, improving the accuracy of data acquisition. This supports rapid iterative optimization of engine design parameters, improving product development efficiency and quality.

[0097] Secondly, this invention provides a device for reducing the order of a one-dimensional system simulation model in Mworks. Please refer to [link / reference]. Figure 2 , Figure 2 This is a schematic diagram of a reduction device for a one-dimensional Mworks system simulation model provided by the present invention.

[0098] exist Figure 2 In this context, the device may include:

[0099] The acquisition module 210 is used to acquire multiple target design variables in the Mworks one-dimensional system simulation model; the multiple target design variables are design variables whose influence on the rocket engine is greater than a preset threshold; the multiple target design variables include at least the rocket engine's intake pressure, fuel injection quantity, and rotational speed.

[0100] Module 220 is used to perform dynamic simulation calculations on multiple target design variables to obtain multiple target physical quantity data.

[0101] The construction module 230 is used to construct a dynamic reduced-order feature model based on multiple target design variables and multiple target physical quantity data.

[0102] The integration module 240 is used to integrate the dynamic reduced-order feature model into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model.

[0103] Based on this, the present invention provides a reduction device for an Mworks one-dimensional system simulation model. The device acquires multiple target design variables from the Mworks one-dimensional system simulation model via an acquisition module 210. These target design variables are design variables whose influence on the rocket engine exceeds a preset threshold. The target design variables include at least the rocket engine's intake pressure, fuel injection quantity, and rotational speed. An acquisition module 220 performs dynamic simulation calculations on these target design variables to obtain multiple target physical quantity data. A construction module 230 further constructs a dynamic reduction feature model based on the multiple target design variables and the multiple target physical quantity data. Finally, an integration module 240 integrates the dynamic reduction feature model. By applying the Mworks one-dimensional system simulation model, a reduced-order Mworks one-dimensional system simulation model is obtained. Based on this, different design variables can be input into the reduced-order Mworks one-dimensional system simulation model, and the key physical quantities of the rocket engine can be obtained using the dynamic reduced-order model. Because the dynamic reduced-order model is a reduced-order model obtained by training the network model based on input and output sample data, no mathematical formula calculations are required during model operation. Therefore, compared with the speed at which key physical quantities of the rocket engine are obtained through formula calculations in the existing Mworks one-dimensional system simulation model, the dynamic reduced-order model established in this invention significantly shortens the simulation time, improves the simulation efficiency, and meets the real-time data requirements of digital twin technology. Furthermore, the dynamic reduced-order model can accurately capture the dynamic changes of key physical quantities of the engine, improve the accuracy of data acquisition, and the real-time data feedback supports the rapid iterative optimization of engine design parameters, which helps to improve product development efficiency and quality.

[0104] Thirdly, this invention provides a device for reducing the order of a one-dimensional system simulation model in Mworks. Please refer to [link / reference]. Figure 3 , Figure 3 This is a schematic diagram of the hardware structure of a device for reducing the order of a one-dimensional system simulation model in Mworks, as provided by the present invention.

[0105] exist Figure 3 In this context, the reduction device may include a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the reduction method for a one-dimensional Mworks system simulation model as described in the first aspect.

[0106] Specifically, device 300 is the hardware structure of a device for reducing the order of a one-dimensional Mworks system simulation model provided by the present invention. The device 300 includes a memory 320, a first processor 310, a second processor 350, and a computer program stored in the memory 320 and executable on the first processor 310 and / or the second processor 350. When the first processor 310 and / or the second processor 350 execute the computer program, they implement the method for reducing the order of the one-dimensional Mworks system simulation model described in the first aspect.

[0107] Optionally, the device 300 may include a communication interface 330. There may be one or more communication interfaces 330. The communication interface 330 can use any transceiver-like device for communicating with other devices or communication networks.

[0108] Optionally, the device 300 may also include a communication line 340. The communication line 340 may include a path for transmitting information between the aforementioned components.

[0109] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings, the disclosure, and the appended claims in carrying out the claimed invention. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0110] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely exemplary descriptions of the invention as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A method for reducing the order of a one-dimensional system simulation model in Mworks, characterized in that, include: Obtain multiple target design variables from a one-dimensional system simulation model in Mworks; Several of the target design variables are design variables whose impact on the rocket engine is greater than a preset threshold; The target design variables include at least the rocket engine's intake pressure, fuel injection quantity, and rotational speed; Dynamic simulation calculations are performed on multiple target design variables to obtain multiple target physical quantity data; A dynamic order reduction feature model is constructed based on multiple target design variables and multiple target physical quantity data. The dynamic reduced-order feature model is integrated into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model.

2. The method as described in claim 1, characterized in that, The step of performing dynamic simulation calculations on multiple target design variables to obtain multiple target physical quantity data includes: Multiple target design variables are sampled according to a preset target experimental design method to obtain multiple sample point data; the target experimental design method includes orthogonal experimental design method, full factorial design method or response surface design method; the multiple sample point data are data that can cover the target experimental design method; the number of multiple sample point data is less than the number of multiple target design variables; The dynamic simulation calculation of multiple target design variables yields multiple target physical quantity data, including: The Mworks one-dimensional simulation model is used to perform simulation calculations on multiple sample point data to obtain multiple physical quantity data corresponding to the multiple sample point data. Multiple physical quantity data are used as multiple target physical quantity data.

3. The method as described in claim 2, characterized in that, The construction of a dynamic order reduction feature model based on multiple target design variables and multiple target physical quantity data includes: A sample database is constructed based on the data of multiple sample points and the data of multiple target physical quantities; Based on the sample database, a dynamic reduced-order feature model is constructed.

4. The method as described in claim 3, characterized in that, The construction of the dynamic reduced-order feature model based on the sample database includes the following steps: The data in the sample database is preprocessed, and the preprocessing includes at least one or more of the following: data cleaning, data normalization, and data deduplication.

5. The method as described in claim 3, characterized in that, The construction of a dynamic reduced-order feature model based on the sample database includes: Multiple feature information of multiple target physical quantity data are extracted from the sample database according to a preset feature extraction strategy; the preset feature extraction strategy includes sliding window processing and / or wavelet transform on the time axis. Based on the aforementioned feature information, multiple feature vectors are constructed; Based on the multiple feature vectors, the network model is trained to obtain the dynamic reduced-order feature model.

6. The method as described in claim 5, characterized in that, The step of training the network model based on multiple feature vectors to obtain the dynamically reduced-order feature model includes: Based on the multiple feature vectors, the network model is trained to obtain the first dynamic order reduction model; The predictive performance of the first dynamic order reduction model is tested using a test set, and the performance test results are obtained; the test set is a sample dataset calculated by the Mworks one-dimensional simulation model. If the performance test results meet the preset prediction accuracy, then the first dynamic order reduction model is used as the dynamic order reduction feature model. If the performance test results do not meet the preset prediction accuracy and generalization ability, the model parameters are adjusted or samples are re-collected to continue model training until a second dynamic reduced-order model that meets the preset prediction accuracy is obtained.

7. The method as described in claim 5, characterized in that, The construction of multiple feature vectors based on multiple feature information includes: Multiple feature information items are combined in a predetermined order to obtain multiple feature vectors.

8. The method as described in claim 1, characterized in that, The process of integrating the dynamic order reduction feature model into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation order reduction model includes: The dynamic reduced-order feature model is integrated into the Mworks one-dimensional system simulation model according to the preset model integration method to obtain the Mworks one-dimensional system simulation reduced-order model; the preset model integration method includes at least one of voting, averaging, stacking, bagging and lifting.

9. A device for reducing the order of a one-dimensional system simulation model in Mworks, characterized in that, include: The acquisition module is used to acquire multiple target design variables in the Mworks one-dimensional system simulation model; Several of the target design variables are design variables whose impact on the rocket engine is greater than a preset threshold; The target design variables include at least the rocket engine's intake pressure, fuel injection quantity, and rotational speed; The module is used to perform dynamic simulation calculations on multiple target design variables to obtain multiple target physical quantity data. The construction module is used to construct a dynamic reduced-order feature model based on multiple target design variables and multiple target physical quantity data; An integration module is used to integrate the dynamic reduced-order feature model into the Mworks one-dimensional system simulation model to obtain the Mworks one-dimensional system simulation reduced-order model.

10. A device for reducing the order of a one-dimensional system simulation model in Mworks, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for reducing the order of a one-dimensional Mworks system simulation model as described in any one of claims 1 to 8.

Images