Aerospace system simulation test data analysis system and analysis method

Abstract

The invention discloses a spaceflight system simulation test data analysis system and analysis method, belongs to the technical field of spaceflight system simulation test data analysis, integrates artificial intelligence and data science, and is particularly suitable for processing complex data generated after a carrier rocket simulation test. According to the method, a plurality of expert networks are integrated, the expert network most suitable for current data characteristics is dynamically selected by utilizing a gating decision module, and decision support is provided in combination with a knowledge base system. The flexibility and expandability of the hybrid expert model are utilized, a large amount of complex data generated after a simulation test is carried out on a typical aerospace system of a carrier rocket is efficiently and accurately analyzed, and the intelligence level of data processing is improved through integration of the knowledge base.

Description

Technical Field

[0001] The present invention relates to the technical field of aerospace system simulation test data analysis, and in particular to an aerospace system simulation test data analysis system and analysis method. Background Art

[0002] In the launch vehicle system simulation test, the launch vehicle system simulation test data changes dynamically over time, which requires the model to have good adaptability. Data analysis and processing is a crucial link.

[0003] At the same time, launch vehicle simulation data is usually high-dimensional, and a single model is difficult to fully capture the data characteristics. The large amount of data generated by the test traditionally requires manual interpretation and analysis by relevant professional and technical personnel, which takes up a lot of manpower and time.

[0004] Traditional data analysis methods, such as single statistical analysis, machine learning algorithms or classical expert systems, although effective in specific situations, require the knowledge and experience of experts in the field to analyze the complex data in the launch vehicle field. Existing solutions make it difficult to achieve effective knowledge integration and cannot effectively analyze the simulation test data of aerospace systems.

[0005] In view of this, it is necessary to provide a new technical solution to solve the above problems. Summary of the invention

[0006] In order to solve the above technical problems, the present application provides a space system simulation test data analysis system and analysis method, which can efficiently and accurately analyze the large amount of complex data generated after simulation tests of a typical aerospace system such as a launch vehicle, and improve the intelligence level of data processing through the integration of a knowledge base.

[0007] A space system simulation test data analysis system, comprising: A simulation test database; the simulation test database is used to store aerospace system simulation test data; Data preprocessing module: the data preprocessing module is used to clean and standardize the original data; Expert model integration module; the expert model integration module includes multiple expert networks for different data analysis tasks; Knowledge base; the knowledge base is used to store and manage the knowledge and experience of experts in the field and provide decision support for the expert network; Gating decision module; the gating decision module is used to dynamically select the most suitable expert network according to the key features extracted by the data preprocessing module; Result fusion module: The result fusion module is used to integrate the analysis results of various expert networks to form a unified analysis report and output it in a readable form.

[0008] According to another aspect of the present invention, a method for analyzing aerospace system simulation test data is provided, wherein the method uses the aerospace system simulation test data analysis system to analyze aerospace system simulation test data, and comprises: Inputting the original simulation test data generated by the aerospace system simulation test into the aerospace system simulation test data analysis system through a database or a user interface; The data preprocessing module cleans the input raw simulation test data, standardizes it and vectorizes it; The gate decision module dynamically selects an expert network for data analysis from the expert model integration module based on the data processed by the data preprocessing module; Analyze the data using the selected expert network and output the analysis results; The analysis results of each expert network are weighted and fused through the result fusion module to form a unified comprehensive analysis report, and the analysis report is output in a readable form.

[0009] Preferably, the original simulation test data generated by the aerospace system simulation test is input into the aerospace system simulation test data analysis system through a database or a user interface, and the data input into the aerospace system simulation test data analysis system includes natural language data, quasi-natural language data and formatted tabular data.

[0010] Preferably, the data preprocessing module cleans the input original simulation test data, standardizes and vectorizes it, including: Using a data preprocessing module to classify natural language data, quasi-natural language data, and formatted tabular data; Abstracting words in natural language data or quasi-natural language data into identifiers, and combining the identifiers into a complete first input sentence; Using each word in the formatted table header as the second input statement; Directly vectorize the data in the formatted table as the first input vector; The non-time column in the formatted table header is vectorized using an existing classic statistical analysis algorithm as the second input vector.

[0011] Preferably, the gating decision module dynamically selects an expert network for data analysis based on the data processed by the data preprocessing module, and the expression of the parameter output by the gating decision module to the expert network is: ; In the formula, Represents the parameters of the gating decision module output to the expert network: represents the normalized input vector of the selected expert module; Represents the normalized output weight of the gating decision module for the input to each expert module; Indicates the corresponding A network of experts.

[0012] Preferably, the expression of the normalized output weight input to each expert network by the gating decision module is: ; In the formula, is the weight vector obtained through training; A weight vector that is manually configured by the user through the user interface; for noise; It is the input parameter of the gating decision module.

[0013] Preferably, the input parameter expression of the gating decision module is: ; in, ; ; In the formula, is the first input parameter, is the second input parameter; To obtain the word frequency probability vector of the input vector in the aerospace professional vocabulary from the knowledge base; is a vector set of the first input sentence and the second input sentence; To hit the corresponding words in the first input sentence and the second input sentence set; is a constant, which is an arbitrary value greater than 0 and less than 1 set artificially based on expert experience; is the matrix composed of the input vector of the gate decision module; the superscript T is the transpose of the matrix; , is the first input vector; is the second input vector; The experimental data for comparison is read from the knowledge base by the gating decision module.

[0014] Preferably, when the selected expert network is used to analyze the data, a query request is sent to the knowledge base according to the requirements of the current analysis task; the knowledge base retrieves and matches relevant knowledge items according to the request, and pushes the matched knowledge items to the expert network.

[0015] Preferably, when the analysis results of each expert network are weighted and fused through the result fusion module to form a unified comprehensive analysis report, and the analysis report is output in a readable form, the result fusion module considers the analysis quality and confidence of different expert networks in the fusion process, and analyzes their relevance to the current analysis task, and presents the analysis results in the form of graphics or charts.

[0016] Preferably, after the analysis results of each expert network are weighted and fused by the result fusion module to form a unified comprehensive analysis report, and the analysis report is output in a readable form, the knowledge base update and the gate decision module update are also included; Among them, the knowledge base update includes updating its content by adding new cases, revising rules or adding new performance indicators based on the experience of this analysis; The update of the gating decision module includes updating and retraining the weights of the gating decision module according to the experience of this analysis and user operations, so that the expert model integration module can meet the update of data changes and user needs.

[0017] Compared with the prior art, this application has at least the following beneficial effects: 1. The present invention utilizes the flexibility and scalability of the hybrid expert model to efficiently and accurately analyze a large amount of complex data generated after simulation tests of a typical aerospace system such as a launch vehicle, and improves the intelligent level of data processing through the integration of a knowledge base.

[0018] 2. The present invention deeply integrates the knowledge and experience of domain experts into the data analysis process to improve the intelligence level of analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Hereinafter, some specific embodiments of the present invention will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the accompanying drawings: Figure 1 A schematic diagram of the system principle of the aerospace system simulation test data analysis system of the present invention; Figure 2 The figure is a schematic diagram of the overall process of the aerospace system simulation test data analysis method of the present invention.

[0020] The above drawings include the following reference numerals: 10. Data preprocessing module; 20. Gating decision module; 30. Expert model integration module; 40. Result fusion module; 50. Knowledge base; 60. Simulation test database; 70. User interface. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0022] like Figure 1 As shown, a space system simulation test data analysis system includes: a data preprocessing module 10, an expert model integration module 30, a knowledge base 50, a gate decision module 20, a result fusion module 40 and a simulation test database 60.

[0023] The simulation test database 60 is used to store aerospace system simulation test data. The data preprocessing module 10 is used to clean and standardize the raw data input by the simulation test database 60. The expert model integration module 30 contains multiple expert networks for different data analysis tasks, and the selected expert network is used to analyze each received data. The knowledge base 50 stores and manages the knowledge and experience of experts in the field, and can provide decision support for the expert network. The gated decision module 20 is used to dynamically select the most suitable expert network based on the key features extracted by the data preprocessing module 10. The result fusion module 40 is used to integrate the analysis results of each expert network, form a unified analysis report, and output it in a readable form.

[0024] Among them, the expert network in the expert model integration module 30 adopts an existing data analysis expert network that can analyze aerospace system simulation test data, and the expert network selected by the gated decision module 20 can perform data analysis according to the type of aerospace system simulation test data.

[0025] The result fusion module 40 can use existing large models to perform result fusion. For example, it can use existing large language models such as chatGPT, Wenxinyiyan, Doubao, etc. for fusion.

[0026] As another embodiment of the present invention, a space system simulation test data analysis system also includes a user interface 70 for providing human-computer interaction. The user interface 70 can be used to input aerospace system simulation test data or manually adjust corresponding parameters in the aerospace system simulation test data analysis system, and can also visualize the obtained analysis report. The analysis report includes key findings, predictions, suggestions, etc. to support the decision-making process, and the visualized analysis report is easy for users to understand intuitively. For example, the trend analysis results can be represented by a time series graph, and the cluster analysis results can be represented by a scatter plot.

[0027] According to another aspect of the present application, a method for analyzing aerospace system simulation test data is also provided, and the aerospace system simulation test data analysis system is used to perform aerospace system simulation test data analysis, comprising the following steps: Step S1: inputting the original simulation test data generated by the aerospace system simulation test into the aerospace system simulation test data analysis system through a database or a user interface.

[0028] Specifically, the data input into the aerospace system simulation test data analysis system includes natural language data, quasi-natural language data, and formatted table data.

[0029] Step S2: The data preprocessing module cleans the input original simulation test data, standardizes it and vectorizes it.

[0030] Specifically, it includes using a data preprocessing module to classify natural language data, quasi-natural language data, and formatted table data; Abstracting words in natural language data or quasi-natural language data into identifiers, and combining the identifiers into a complete first input sentence; Using each word in the formatted table header as the second input statement; Directly vectorize the data in the formatted table as the first input vector; The non-time column in the formatted table header is vectorized using an existing classic statistical analysis algorithm as the second input vector.

[0031] In this embodiment, the data preprocessing module first classifies the input raw data, and the words of the natural language input will be abstracted into tokens, and several tokens will be combined into a complete input sentence S1. The formatted data and table input will be classified and processed by the preprocessing network, and the words in the table header will be used as S2. The data in the table will be directly vectorized and called vector input u1. At the same time, the preprocessing network will identify the time column in the table header, and after executing various classic statistical analysis algorithms on the non-time column, it will be vectorized to become vector input u2.

[0032] Step S3: The gate decision module dynamically selects an expert network for data analysis from the expert model integration module based on the data processed by the data preprocessing module.

[0033] Preferably, the gating decision module dynamically selects an expert network for data analysis based on the data processed by the data preprocessing module, and the expression of the parameter output by the gating decision module to the expert network is: ; In the formula, Represents the parameters of the gating decision module output to the expert network: represents the normalized input vector of the selected expert module; Represents the normalized output weight of the gating decision module for the input to each expert module; Indicates the corresponding A network of experts.

[0034] Preferably, the expression of the normalized output weight input to each expert network by the gating decision module is: ; In the formula, is the weight vector obtained through training; A weight vector that is manually configured by the user through the user interface; for noise; It is the input parameter of the gating decision module.

[0035] Preferably, the input parameter expression of the gating decision module is: ; in, ; ; In the formula, is the first input parameter, is the second input parameter; To obtain the word frequency probability vector of the input vector in the aerospace professional vocabulary from the knowledge base; is a vector set of the first input sentence and the second input sentence; To hit the corresponding words in the first input sentence and the second input sentence set; is a constant, which is an arbitrary value greater than 0 and less than 1 set artificially based on expert experience; is the matrix composed of the input vector of the gate decision module; the superscript T is the transpose of the matrix; , is the first input vector; is the second input vector; The experimental data for comparison is read from the knowledge base by the gating decision module.

[0036] The weights of the gated network can be pre-configured by prior knowledge. In actual use, the knowledge base constructed by simulation test data and aerospace system prior expert knowledge will iteratively update the gated network and expert network based on the accumulation of simulation test data and the input and configuration of users (professionals), dynamically expanding the capabilities of the hybrid expert network.

[0037] Step S4: Analyze the data using the selected expert network and output the analysis results.

[0038] When using the selected expert network to analyze data, a query request is sent to the knowledge base according to the needs of the current analysis task; the knowledge base retrieves and matches relevant knowledge items based on the request, and pushes the matched knowledge items to the expert network.

[0039] Step S5: The analysis results of each expert network are weighted and integrated through the result fusion module to form a unified comprehensive analysis report, and the analysis report is output in a readable form.

[0040] When the analysis results of each expert network are weighted and integrated through the result fusion module to form a unified comprehensive analysis report, the result fusion module considers the analysis quality and confidence of different expert networks in the fusion process, analyzes their relevance to the current analysis task, and presents the analysis results in the form of graphics or charts.

[0041] Preferably, after step 5, the method further includes: updating the knowledge base and updating the gating decision module.

[0042] Among them, knowledge base updating includes updating its content by adding new cases, revising rules or adding new performance indicators based on the experience of this analysis, which helps to improve the accuracy and efficiency of future analysis.

[0043] The update of the gating decision module includes updating and retraining the weights of the gating decision module according to the experience of this analysis and user operations, so that the expert model integration module can meet the update of data changes and user needs.

[0044] An expert model is constructed based on aerospace system simulation test data and actual application needs. For aerospace system simulation, there are many professional categories and a considerable number of professional terms. For the same test data, different professional categories have different requirements for data analysis applications. For example, for a typical aerospace system such as a launch vehicle, there are professions such as trajectory, payload, structure, aerodynamic heat, navigation, guidance, control, and system simulation. There are professional terms and vocabulary such as ballistic inclination, orbital inclination, angle of attack, and pitch angle. These terms can be used by one or more professional categories. The expert model constructed by this patent is constructed based on the classification of aerospace systems and related professional classifications. The training process of a single model is targeted and highly measurable.

[0045] Taking the case of the launch vehicle trajectory professionals in the aerospace system inputting ballistic simulation test data into the network, the preprocessing network cleans the data format commonly used in the profession (removes data anomalies), standardizes and vectorizes it, and inputs it to the gated decision module. Based on the data characteristics, the gated decision module determines that the data should mainly be inferred by the trajectory-related expert model and the system simulation expert model, and adjusts the weights to make the network input more inclined to the corresponding expert model. After the expert model set is output, it is specially delivered to the general large prediction model output to provide users with analysis results (such as the ballistic inclination curve, the target entry point orbit inclination, the maximum angle of attack and the corresponding dynamic pressure, etc., which are of concern to ballistic professionals). At the same time, the gated decision module adds the vectorized input data and the corresponding output weights to the knowledge base to provide support for subsequent users.

[0046] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0047] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0048] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An aerospace system simulation test data analysis system, characterized in that: include: Simulation test database; The simulation test database is used to store aerospace system simulation test data; Data preprocessing module; The data preprocessing module is used to clean and standardize the original data; Expert model integration module; the expert model integration module includes multiple expert networks for different data analysis tasks; Knowledge base; the knowledge base is used to store and manage the knowledge and experience of experts in the field and provide decision support for the expert network; Gating decision module; the gating decision module is used to dynamically select the most suitable expert network according to the key features extracted by the data preprocessing module; Result fusion module: The result fusion module is used to integrate the analysis results of various expert networks to form a unified analysis report and output it in a readable form.

2. A method for analyzing aerospace system simulation test data, characterized in that: The aerospace system simulation test data analysis system according to claim 1 is used to analyze aerospace system simulation test data, comprising: Inputting the original simulation test data generated by the aerospace system simulation test into the aerospace system simulation test data analysis system through a database or a user interface; The data preprocessing module cleans the input raw simulation test data, standardizes it and vectorizes it; The gate decision module dynamically selects an expert network for data analysis from the expert model integration module based on the data processed by the data preprocessing module; Analyze the data using the selected expert network and output the analysis results; The analysis results of each expert network are weighted and fused through the result fusion module to form a unified comprehensive analysis report, and the analysis report is output in a readable form.

3. The aerospace system simulation test data analysis method according to claim 2, characterized in that: The original simulation test data generated by the aerospace system simulation test is input into the aerospace system simulation test data analysis system through a database or a user interface. The data input into the aerospace system simulation test data analysis system includes natural language data, quasi-natural language data and formatted table data.

4. The aerospace system simulation test data analysis method according to claim 3, characterized in that: The data preprocessing module cleans the input raw simulation test data, standardizes and vectorizes it, including: Using a data preprocessing module to classify natural language data, quasi-natural language data, and formatted tabular data; Abstracting words in natural language data or quasi-natural language data into identifiers, and combining the identifiers into a complete first input sentence; Using each word in the formatted table header as the second input statement; Directly vectorize the data in the formatted table as the first input vector; The non-time column in the formatted table header is vectorized using an existing classic statistical analysis algorithm as the second input vector.

5. The aerospace system simulation test data analysis method according to claim 4, characterized in that: The gate control decision module dynamically selects an expert network for data analysis based on the data processed by the data preprocessing module. The expression of the parameter output by the gate control decision module to the expert network is: ； In the formula, Represents the parameters of the gating decision module output to the expert network: represents the normalized input vector of the selected expert module; Represents the normalized output weight of the gating decision module for the input to each expert module; Indicates the corresponding A network of experts.

6. The aerospace system simulation test data analysis method according to claim 5, characterized in that: The expression of the normalized output weight input to each expert network by the gated decision module is: ； In the formula, is the weight vector obtained through training; A weight vector that is manually configured by the user through the user interface; for noise; It is the input parameter of the gating decision module.

7. The aerospace system simulation test data analysis method according to claim 6, characterized in that: The input parameter expression of the gate decision module is: ； in, ； ； In the formula, is the first input parameter, is the second input parameter; To obtain the word frequency probability vector of the input vector in the aerospace professional vocabulary from the knowledge base; is a vector set of the first input sentence and the second input sentence; To hit the corresponding words in the first input sentence and the second input sentence set; is a constant, which is an arbitrary value greater than 0 and less than 1 set artificially based on expert experience; is the matrix composed of the input vector of the gate decision module; the superscript T is the transpose of the matrix; , is the first input vector; is the second input vector; The experimental data for comparison is read from the knowledge base by the gating decision module.

8. The aerospace system simulation test data analysis method according to claim 7, characterized in that: When analyzing the data using the selected expert network, a query request is sent to the knowledge base according to the requirements of the current analysis task; The knowledge base retrieves and matches relevant knowledge items based on the request, and pushes the matched knowledge items to the expert network.

9. The aerospace system simulation test data analysis method according to claim 8, characterized in that: When the analysis results of each expert network are weighted and fused through the result fusion module to form a unified comprehensive analysis report and output the analysis report in a readable form, the result fusion module considers the analysis quality and confidence of different expert networks in the fusion process, analyzes their relevance to the current analysis task, and presents the analysis results in the form of graphics or charts.

10. The aerospace system simulation test data analysis method according to claim 9, characterized in that: After the analysis results of each expert network are weighted and integrated through the result fusion module to form a unified comprehensive analysis report, and the analysis report is output in a readable form, it also includes the update of the knowledge base and the gated decision module; Among them, the knowledge base update includes updating its content by adding new cases, revising rules or adding new performance indicators based on the experience of this analysis; The update of the gating decision module includes updating and retraining the weights of the gating decision module according to the experience of this analysis and user operations, so that the expert model integration module can meet the update of data changes and user needs.

Images