Liquid rocket engine expert fault knowledge base system

Abstract

The invention discloses a liquid rocket engine expert fault knowledge base system, and belongs to the technical field of engine fault diagnosis. The system comprises a data management module and a fault knowledge base module, the data management module carries out data preprocessing on test run simulation data of the liquid rocket engine, and then all the data are stored in a database in a unified mode. The fault knowledge base module constructs and maintains a model fault tree and fault modes and fault cases on the fault tree, and associates the test run simulation data with the fault modes and the fault cases. According to the method, data support is provided for engine fault diagnosis, and potential faults can be identified and processed more accurately and more quickly.

Description

Technical Field

[0001] The present invention relates to an expert fault knowledge base system for liquid rocket engines, belonging to the technical field of engine fault diagnosis. Background Art

[0002] Liquid rocket engines play a crucial role in modern aerospace. However, the high complexity and high risk of liquid rocket engines make them vulnerable to various internal and external factors, leading to faults. To improve the reliability and safety of liquid rocket engine systems, it is urgent to carry out intelligent fault diagnosis.

[0003] An expert system for fault diagnosis is a software system based on artificial intelligence and professional domain knowledge, aiming to use advanced artificial intelligence technologies and big data analysis methods to achieve automatic identification and accurate diagnosis of mechanical equipment faults. The system includes an expert knowledge base, which collects rich knowledge and rule sets provided by experts in related fields, describing various possible fault situations, symptoms, causes, and solutions. This knowledge may come from literature, experimental data, case studies, and expert experience. The expert system uses an inference engine to analyze the fault symptoms and characteristics provided by the user and reason based on the predefined knowledge base to determine the most likely fault cause, providing operators with fast and accurate fault diagnosis results and solution suggestions. Some systems also have the capabilities of real-time data monitoring and adaptive learning, and can continuously update the knowledge base according to the continuously accumulated data and real-time feedback to improve the accuracy and precision of diagnosis.

[0004] The expert system for liquid rocket engines is a specific-domain expert system that provides a powerful tool for liquid rocket engine operation and maintenance personnel, enabling them to more quickly and accurately identify and handle potential faults, ensuring the safe operation of spacecraft. However, the existing expert systems for liquid rocket engines lack a fault knowledge base. The data volume of liquid rocket engines is large and difficult to manage, and it is impossible to conduct detailed analysis on representative fault sample data. It is necessary to establish a database, including test run data, simulation data, and experimental data. For the problem that there are many fault modes in liquid rocket engines and there is a tree-like hierarchical relationship between various fault modes, making it difficult to locate and analyze the fault modes from the appearance, it is necessary to construct a fault mode library; in case of a fault, non-professionals cannot carry out correct disposal measures for various fault modes. Summary of the Invention

[0005] The technical problem solved by the present invention is: overcoming the deficiencies of the prior art, an expert fault knowledge base system for liquid rocket engines is proposed. By establishing a data management module and a fault knowledge base module, it provides data support for engine fault diagnosis and promotes faster and more accurate identification and handling of potential faults.

[0006] The technical solution of the present invention is as follows:

[0007] A liquid rocket engine expert fault knowledge base system, comprising a data management module and a fault knowledge base module;

[0008] The fault knowledge base module constructs and maintains fault trees of liquid rocket engines of multiple models, as well as fault modes and fault cases on the fault trees;

[0009] The data management module performs data preprocessing on the test run simulation data of the liquid rocket engine, then uniformly stores all the preprocessed data in a database, and matches and associates the data marked as faulty in the database with the fault modes and fault cases.

[0010] Further, the data management module includes a test run simulation data management function module and a training and validation set management function module;

[0011] The test run simulation data management function module accesses the test run simulation data under each model, performs preprocessing and storage, and tags the stored data according to the model, operating conditions, and occurrence of faults;

[0012] The training and validation set management function module accesses the data stored in the test run simulation data management function module and is used to maintain and manage the data sets required for training and validation

[0013] Further, the test run simulation data of the liquid rocket engine includes test run data, simulation data, slow-varying parameters, and fast-varying parameters, which are semi-structured and unstructured data from various stages of the mission. Sparse noise reduction or sparse morphological component analysis is used for data preprocessing of the above data.

[0014] Further, all the preprocessed data is uniformly stored in a database, and the data storage structure includes distributed file storage, time-series data storage, and relational data storage;

[0015] The distributed file storage uses HDFS and is used to store time-series data source files and file data; among them, the time-series data source files include slow-varying data and fast-varying data, and the file data includes test outlines, audio and video, picture data, and diagnostic analysis reports;

[0016] The time-series data storage uses HBASE and is used to store the structured data after parsing the test run time-series files;

[0017] The relational data storage uses MySQL and is used to store conventional file description information, structured management data, and description information of time-series files.

[0018] Further, when the newly generated test run simulation data is stored in the test run simulation data management function module, measurement point associated parameter mapping and classification processing is performed to make the measurement point parameters of the newly generated test run simulation data correspond one by one with the model measurement points of the test run simulation data management function module.

[0019] Further, the method for measurement point associated parameter mapping and classification is as follows: through the mapping table of file parameters and measurement points, the file parameter name is matched with the name and code of the model measurement point. If the match is successful, the mapped parameter column is automatically filled with the measurement point name; if there are multiple, the name of the model measurement point is preferentially used for matching, and the user can modify it by pulling down; if the mapping is not successful, check whether there is a matching relationship for the file parameter name in the historical matching library of this test run. If there is, the mapped parameter column is automatically filled with the measurement point name that has been historically mapped. If there are multiple, any one of them is randomly filled, and the user can modify it by pulling down. If there is no such match, manual mapping needs to be performed by selecting the mapped parameter.

[0020] Further, a multi-level indexing mechanism is adopted. All preprocessed data is stored in a distributed database, and the data that needs to establish an index is stored in a relational database. When querying data, according to the pointer in the database pointing to the distributed database, the query planner queries the distributed database, and the relational database returns the query result set. The query result set and the query result of the distributed database are merged to obtain the final query result.

[0021] Further, the fault knowledge base module includes a fault tree module, a fault mode library module, and a fault case library module;

[0022] The fault tree module is provided with multiple model liquid rocket engine fault trees. Each fault tree node provides fault mode information and related fault cases to form the integration of various fault information.

[0023] The fault mode library module manages and maintains fault modes, associates with fault tree nodes, and provides various fault mode attribute information and handling solutions. Among them, the fault mode attribute information includes fault items, fault data characteristics, historical fault test runs, and fault photos.

[0024] The fault case library module manages and maintains fault cases and associates with fault tree nodes.

[0025] Further, the fault tree module includes a fault tree application sub-module and a fault tree design sub-module. The fault tree application sub-module is used to store multiple existing model liquid rocket engine fault trees. The fault tree design sub-module is used to design a liquid rocket engine fault tree for a new model, establish an association with the fault mode and fault case, and store the built liquid rocket engine fault tree in the fault tree application sub-module.

[0026] The advantages of the present invention compared with the prior art are as follows:

[0027] (1) The present invention establishes a liquid rocket engine database for integrating the test run data, simulation data, experimental data, etc. of the engine, storing and managing them uniformly, and making it more convenient to find representative fault sample data for detailed analysis.

[0028] (2) The present invention constructs a fault knowledge base and designs a fault tree for integrating various fault mode information and fault case information of liquid rocket engines under various models, obtaining the tree-like hierarchical relationship between various fault modes, and enabling the location analysis of fault modes from the appearance. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0030] Figure 1 is the functional logic flowchart of the fault knowledge base of the embodiment of the present invention;

[0031] Figure 2 is the functional logic flowchart of the test run simulation data management of the embodiment of the present invention;

[0032] Figure 3 is the functional logic flowchart of the training verification set management of the embodiment of the present invention;

[0033] Figure 4 is the fault tree structure diagram of the embodiment of the present invention;

[0034] Figure 5 is the functional logic flowchart of the fault mode library of the embodiment of the present invention;

[0035] Figure 6 is the functional logic flowchart of the fault case library of the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] Hereinafter, the exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0037] The present invention provides a liquid rocket engine expert fault knowledge base system, including a data management module and a fault knowledge base module.

[0038] The data management module accesses and integrates the test run data, simulation data, etc. of the liquid rocket engine input, receives the access of slow-varying parameters (100Hz, 1kHz, 5kHz) and fast-varying parameters (25.6kHz), and preprocesses the input data using more than 10 general tools such as sparse noise reduction and sparse morphological component analysis, and uniformly stores the data in the database. This module includes a test run simulation data management functional sub-module and a training verification set management functional sub-module. The test run simulation data is the source of the training verification set data and also an important input data for training strategies and diagnostic analysis.

[0039] The test run simulation data management functional sub-module accesses and archives and stores the test run simulation data under each model, such as Figure 2 shown, supports local data import, synchronizes the data of the test data management system for data parsing and warehousing, and supports establishing data tags, data preview, download, query, and viewing diagnostic records, etc. The test run simulation data management functional sub-module also classifies and processes the test run data by tagging the data according to the operating conditions and fault occurrence situations, as well as mapping the associated parameters of the measurement points. Among them, the operating condition tags, such as start-up condition, 50% condition, and mutation condition. The fault situation tags, such as short circuit fault, short circuit fault.

[0040] The method for mapping the associated parameters of the measurement points is as follows: through the mapping table of file parameters and measurement points, the system will automatically match the file parameter name with the name and code of the model measurement points. If the match is successful, the mapped parameter column will be automatically filled with the name of the measurement point; if there are multiple, the measurement point name will be matched preferentially, and the user can pull down to modify; if the mapping is not successful, the system will search the historical match library of this test run for whether there is a match relationship for this file parameter name. If there is, the mapped parameter column will be automatically filled with the name of the measurement point that has been historically mapped. If there are multiple, one will be randomly filled, and the user can pull down to modify. If there is no match, the user needs to click the select mapped parameter in the operation column on the right side of the mapped parameter to perform manual mapping.

[0041] The training verification set management functional sub-module accesses and archives and stores all the test run historical data, and at the same time evaluates the corresponding strategies according to the dataset association strategy, providing data for the system's fault diagnosis strategy development, training verification and evaluation, such as Figure 3 shown, the training verification set management functional module can create a dataset, and at the same time needs to set the target fault data and the proportion of the target fault data, and can associate existing strategies, evaluate strategies, view evaluation records, etc.

[0042] The stored data mainly comes from a large amount of semi-structured and unstructured test run data in various stages of the task. When storing the original data, a distributed file system (HDFS), big data redundancy elimination, and efficient and low-cost big data storage technology are adopted to store the massive manufacturing data. A multi-level index mechanism is adopted. The data itself is stored in a distributed database, and the data that needs to establish an index is stored in a relational database.

[0043] The data storage structure includes distributed file storage, time-series data storage, and relational data storage. The distributed file storage uses HDFS and mainly stores source files, including time-series data source files (slow-changing data, fast-changing data), and file data (test outlines, audio and video, picture data, diagnostic analysis reports). The time-series data storage uses HBASE and mainly stores the structured data after parsing the test run time-series files; the relational data storage uses MySQL and mainly stores conventional file description information, structured management data (engine model results, test run times, fault cases, knowledge bases), and description information of time-series files.

[0044] When querying data, according to the "pointer" pointing to the distributed database in the database, the query planner queries the distributed database, and the relational database returns the query result set. The combination of the query result set and the query result of the distributed database is the final query result.

[0045] The fault knowledge base module structurally processes various types of data of liquid rocket engines, constructs a model fault tree and the fault modes and fault cases on the fault tree. An association relationship is established between the fault tree and the fault mode and fault case information and stored in the corresponding fault knowledge base. The fault knowledge base module includes a fault tree application sub-module, a fault tree design sub-module, a fault mode library sub-module, and a fault case library sub-module.

[0046] The fault tree application sub-module provides the fault tree of the liquid rocket engine after the model design is completed. Each fault tree node provides fault mode information (including short circuit, open circuit, series circuit, non-explosion of electric detonators, etc.) and related fault cases, forming an integration of each fault information. The fault tree of the YF101 engine is as Figure 4 shown. The first level is the combustion chamber fault, the turbopump fault, and the oxygen pump fault; the second level further decomposes the first-level faults, and so on, forming the nodes of the entire fault tree.

[0047] The fault tree design sub-module accepts the fault tree of the new model liquid rocket engine established by the user, as Figure 1 shown, and binds the relevant fault modes and fault cases.

[0048] The fault mode library sub-module accepts the fault mode information established by the user under the model and the integration of the fault mode knowledge, and manages and maintains the connected fault modes, asFigure 5 As shown in the figure. The fault mode library file recording format includes various information formats such as data, images, and audio. The fault mode library contains a standard library with known fault modes, provides more than 10 kinds of fault mode attribute information and its handling solutions, provides a structural crack fault database based on the wavelet finite element method, and provides indicators such as natural frequencies and modal vibration modes. The fault mode attribute information includes: fault entries, fault data characteristics, historical fault occurrences, fault photos, fault handling solutions, etc. The fault mode library provides comparative analysis of data from multiple occurrences under a single fault mode, and obtains statistical analysis and comparison displays of fault mode information in various ways. It provides a basis for the application of fault trees and the selection of fault labels in the test run simulation data module.

[0049] The fault case library sub-module accepts the fault case information under the established model by the user and the integration of fault case knowledge, and manages and maintains the connected fault cases, such as Figure 6 As shown in the figure. It associates with the fault tree nodes, provides basic auxiliary fault diagnosis for the application of fault trees, and forms in-depth application knowledge based on fault historical data.

[0050] Among them, the fault mode library and the fault case library are two databases with different query methods, and the data cross-interconnects. The fault mode library makes list records, model tree displays, adds, edits, deletes, details, and queries of fault modes according to different fault modes. The fault case library makes list records, model tree displays, edits of fault cases, details, and log queries according to each fault case.

[0051] The above-described embodiments are only relatively preferred specific implementation manners of the present invention, and the general changes and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included in the protection scope of the present invention.

Claims

1. A knowledge base system for expert faults of liquid rocket engines, characterized in that, It includes a data management module and a fault knowledge base module; The fault knowledge base module constructs and maintains fault trees of liquid rocket engines of multiple models, as well as fault modes and fault cases on the fault trees; The data management module preprocesses the test run simulation data of the liquid rocket engine, then uniformly stores all the preprocessed data in the database, and matches and associates the data marked as faulty in the database with the fault modes and fault cases.

2. The liquid rocket engine expert fault knowledge base system according to claim 1, wherein The data management module includes a test run simulation data management function module and a training and validation set management function module; The test run simulation data management function module accesses the test run simulation data under each model, preprocesses and stores it, and tags the stored data according to the model, operating conditions, and occurrence of faults; The training and validation set management function module accesses the data stored in the test run simulation data management function module and is used to maintain and manage the data sets required for training and validation.

3. The liquid rocket engine expert fault knowledge base system according to claim 1 or 2, characterized in that, The test run simulation data of the liquid rocket engine includes test run data, simulation data, slow-varying parameters, and fast-varying parameters. Sparse denoising or sparse morphological component analysis is used to preprocess the above data.

4. The liquid rocket engine expert fault knowledge base system according to claim 3, characterized in that All the preprocessed data is uniformly stored in the database. The data storage structure includes distributed file storage, time series data storage, and relational data storage; The distributed file storage uses HDFS and is used to store time series data source files and file data; among them, the time series data source files include slow-varying data and fast-varying data, and the file data includes test outlines, audio and video, picture data, and diagnostic analysis reports; The time series data storage uses HBASE and is used to store the structured data parsed from the test run time series files; The relational data storage uses MySQL and is used to store conventional file description information, structured management data, and description information of time series files.

5. The liquid rocket engine expert fault knowledge base system according to claim 2, characterized in that, When the newly generated test run simulation data is stored in the test run simulation data management function module, measurement point associated parameter mapping and classification processing are performed to make the measurement point parameters of the newly generated test run simulation data correspond one by one with the model measurement points of the test run simulation data management function module.

6. The liquid rocket engine expert fault knowledge base system according to claim 5, characterized in that The method for measurement point associated parameter mapping and classification is as follows: through the mapping table of file parameters and measurement points, match the file parameter names with the names and codes of the model measurement points. If the match is successful, the mapped parameter column is automatically filled with the measurement point name; if there are multiple, first match through the name of the model measurement point; if the mapping is not successful, check whether there is a matching relationship for the file parameter name in the historical matching library of this test run. If there is, the mapped parameter column is automatically filled with the measurement point name that has been mapped historically. If there are multiple, randomly fill in any one of them; If not, select the mapped parameter for manual mapping.

7. The liquid rocket engine expert fault knowledge base system according to claim 1, characterized in that, A multi-level index mechanism is adopted to store all the preprocessed data in the distributed database, and the data that needs to establish an index is stored in the relational database; when querying data, according to the pointer in the database pointing to the distributed database, the query planner queries the distributed database, and the relational database returns the query result set. The query result set and the distributed database query result are merged to obtain the final query result.

8. The liquid rocket engine expert fault knowledge base system according to claim 1, characterized in that The fault knowledge base module includes a fault tree module, a fault mode library module and a fault case library module; The fault tree module is provided with multiple models of liquid rocket engine fault trees, each fault tree node provides fault mode information and related fault cases, forming an integration of various fault information; The fault mode library module manages and maintains the fault mode, associates the fault tree nodes, and provides a variety of fault mode attribute information and disposal solutions; wherein the fault mode attribute information includes fault items, fault data characteristics, historical fault numbers and fault photos; The fault case library module manages and maintains fault cases and associates fault tree nodes.

9. The liquid rocket engine expert fault knowledge base system according to claim 8, characterized in that, The fault tree module includes a fault tree application submodule and a fault tree design submodule; The fault tree application submodule is used to store fault trees of multiple existing liquid rocket engine models. The fault tree design submodule is used to design a liquid rocket engine fault tree for a new model, establish associations with fault modes and fault cases, and store the constructed liquid rocket engine fault tree in the fault tree application submodule.

Images