Fault detection method, electronic equipment, storage medium and program product
By creating a fault detection workflow and using deep learning models to optimize task scheduling, the difficult problem of efficiently detecting anomalies in flight parameter data in aviation technology has been solved, achieving high efficiency and accuracy in fault detection, reducing maintenance costs, and improving flight safety.
Patent Information
- Application Number
- CN202510509037.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-22
- Publication Date
- 2025-09-16
AI Technical Summary
In the field of aviation technology, existing technologies are inefficient in efficiently detecting potential anomalies and detecting faults in advance from massive and complex flight parameter data.
By receiving the flight parameter dataset and fault detection tools selected by the user, a fault detection workflow is created. Task scheduling is optimized using directed acyclic graphs (DAGs). Deep learning models such as RNN and LSTM are combined to perform multi-dimensional time series anomaly detection, including data preprocessing, feature extraction, and model training. Finally, the detection results are displayed visually.
It improves the pertinence and efficiency of fault detection, can flexibly schedule data analysis tasks in different workflows, fully utilizes the system computing power, reduces aircraft maintenance costs, and improves flight safety.
Smart Images
Figure CN120654021A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a fault detection method, electronic equipment, storage medium, and program product. Background Art
[0002] The collection and analysis of flight parameter data (FPD) has become crucial in the aviation industry. FPD is a series of discrete time series data collected in real time at a set sampling frequency by a sophisticated sensor network onboard an aircraft. This data records key aircraft indicators during flight, including but not limited to flight attitude, engine performance, and environmental parameters. It is not only a crucial basis for pilots' operational decisions but also a core foundation for aircraft fault detection and health management. However, with the growth in flight missions and flight data volumes, how to more efficiently detect potential anomalies and proactively detect faults from this massive and complex volume of FPD data has become a pressing challenge in aviation technology. Summary of the Invention
[0003] The embodiments of the present application provide a fault detection method, an electronic device, a storage medium, and a program product to alleviate or solve one or more technical problems existing in the prior art.
[0004] In a first aspect, an embodiment of the present application provides a fault detection method, comprising:
[0005] receiving user selections from candidate flight parameter data sets and candidate fault detection tools, and obtaining target flight parameter data sets and target fault detection tools; wherein each candidate flight parameter data set includes flight parameters of any subsystem in any aircraft;
[0006] Creating a fault detection workflow, wherein the fault detection workflow is used to use the target fault detection tool to detect the fault type occurring in the subsystem corresponding to the target flight parameter dataset, and the fault detection workflow is an ordered task combination including multiple data analysis tasks;
[0007] The fault detection workflow is tracked, and a fault detection result obtained after the fault detection workflow is executed is output.
[0008] In some implementations, tracking the fault detection workflow and outputting a fault detection result obtained after the fault detection workflow is executed includes:
[0009] Analyzing the dependencies of the multiple data analysis tasks in the fault detection workflow to obtain a directed acyclic graph corresponding to the fault detection workflow;
[0010] Based on the directed acyclic graph, scheduling the task processing modules corresponding to the data analysis tasks to perform processing in sequence;
[0011] According to the processing results of each data analysis task, the fault detection result is obtained and output.
[0012] In some embodiments, sequentially scheduling the task processing modules corresponding to the data analysis tasks to perform processing based on the directed acyclic graph includes:
[0013] Based on the directed acyclic graph, the multiple data analysis tasks are divided into executable tasks and dependent tasks, wherein the executable tasks are data analysis tasks that do not depend on other data analysis tasks, and the dependent tasks are data analysis tasks that depend on analysis results of other data analysis tasks;
[0014] Put the executable task into the executable task pool, and put the dependent task into the dependent task pool;
[0015] Scheduling the task processing modules corresponding to the executable tasks in the executable task pool to execute processing;
[0016] After any one of the executable tasks in the executable task pool is processed, the dependent tasks in the dependent task pool that depend on the executable task are put into the executable task pool.
[0017] In some implementations, after any one of the executable tasks in the executable task pool is processed, placing the dependent tasks in the dependent task pool that depend on the executable task into the executable task pool includes:
[0018] After any of the executable tasks in the executable task pool is processed, the dependent tasks in the dependent task pool that depend on the executable task are changed into executable tasks;
[0019] When the preset time is reached, the dependent task pool is scanned, and the executable tasks in the dependent task pool are transferred to the executable task pool.
[0020] In some implementations, creating a fault detection workflow includes:
[0021] Obtaining an index identifier of the target flight parameter data set and an index identifier of the target fault detection tool;
[0022] Constructing the plurality of data analysis tasks according to the template of the plurality of data analysis tasks; the plurality of data analysis tasks are obtained by filling the template with the index identifier;
[0023] The fault detection workflow is created according to the multiple data analysis tasks.
[0024] In some implementations, tracking the fault detection workflow includes:
[0025] Acquire the target flight parameter data set according to the index identifier of the target flight parameter data set;
[0026] Acquire the target fault detection tool according to the index identifier of the target fault detection tool;
[0027] Use the target fault detection tool to detect the fault type existing in the subsystem corresponding to the target flight parameter data set.
[0028] In some embodiments, using the target fault detection tool to detect the type of fault occurring in the subsystem corresponding to the target flight parameter dataset includes:
[0029] Performing preprocessing on the target flight parameter data set;
[0030] Constructing statistical features of the target flight parameter data set based on the preprocessing results;
[0031] The statistical features are input into the target fault detection tool to obtain the fault type existing in the subsystem.
[0032] In a second aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor implements any method of the embodiment of the present application when executing the computer program.
[0033] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the method of any one of the embodiments of the present application is implemented.
[0034] In a fourth aspect, an embodiment of the present application provides a computer program product, including a computer program, which implements any method of the embodiments of the present application when executed by a processor.
[0035] Based on the above technical solution, this application has at least the following beneficial effects or advantages:
[0036] By receiving the user's selection from candidate flight parameter data sets and candidate fault detection tools, a target flight parameter data set and a target fault detection tool are obtained, wherein each candidate flight parameter data set includes the flight parameters of any subsystem in any aircraft. Then, a fault detection workflow is created. The fault detection workflow is used to use the target fault detection tool to detect the type of fault that occurs in the subsystem corresponding to the target flight parameter data set. The fault detection workflow is an ordered task combination including multiple data analysis tasks. After the fault detection workflow is created, the fault detection workflow is tracked and the fault detection result obtained after the fault detection workflow is executed is output. The embodiment of the present application improves the targetedness of fault detection through the selection of data sets and detection tools, and by creating a fault detection workflow, different data analysis tasks in different workflows can be flexibly mobilized, thereby improving the efficiency of fault detection and making full use of the computing power of the system.
[0037] The above description is only an overview of the technical solution of this application. In order to more clearly understand the technical means of this application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of this application more obvious and easy to understand, the specific implementation methods of this application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the multiple drawings represent the same or similar components or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments according to the present application and should not be regarded as limiting the scope of the present application.
[0039] Figure 1 A flowchart of a fault detection method provided by an embodiment of the present application is shown;
[0040] Figure 2 A data flow processing diagram of a fault detection method provided by an embodiment of the present application is shown;
[0041] Figure 3 A workflow processing diagram of a fault detection method provided by an embodiment of the present application is shown;
[0042] Figure 4 A schematic block diagram of a fault detection system provided by an embodiment of the present application is shown;
[0043] Figure 5 A schematic diagram showing the processing results of a fault detection method provided in an embodiment of the present application is shown;
[0044] Figure 6 A block diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION
[0045] Hereinafter, only certain exemplary embodiments are briefly described. As will be appreciated by those skilled in the art, the described embodiments may be modified in various ways without departing from the spirit or scope of the present application. Therefore, the drawings and description are to be regarded as illustrative in nature and not restrictive.
[0046] To facilitate understanding of the technical solutions of the embodiments of the present application, the following describes the related technologies of the embodiments of the present application. The following related technologies can be combined with the technical solutions of the embodiments of the present application as optional solutions, and all of them fall within the scope of protection of the embodiments of the present application.
[0047] The following describes in detail the technical solution of this application and how it solves the aforementioned technical problems using specific embodiments. The several specific embodiments listed can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. The following describes the embodiments of this application in detail with reference to the accompanying drawings.
[0048] Flight data is typically time series data. Traditional anomaly detection methods for time series primarily target single-dimensional data and can be broadly categorized as model-based, sample distance-based, and sample density-based. However, flight data is inherently multidimensional, making anomaly detection more complex. In multidimensional time series, it's necessary not only to identify anomalies in a single variable but also to consider the interactions and causal relationships between different variables, all of which are beyond the capabilities of single-dimensional time series methods.
[0049] To address these challenges, ensemble learning and deep learning techniques offer new solutions. These techniques are capable of learning complex patterns in data and deep relationships between variables, making them effective tools for anomaly detection in multidimensional time series. Deep learning models, particularly those capable of processing time series data, such as recurrent neural networks (RNNs) and long short-term memory networks (LSTMs), have demonstrated their potential in capturing the complex dynamics of time series data and detecting future states in a variety of fields.
[0050] In summary, the development of flight parameter fault detection algorithms takes place in an environment where multidimensional time series data is increasing and the requirements for anomaly detection technology are becoming increasingly stringent. Traditional anomaly detection methods have limitations when processing multidimensional time series data, but the emergence of ensemble learning and deep learning technologies has provided new ideas and tools to address this problem.
[0051] This application proposes a fault detection method, which utilizes the above-mentioned technology to develop a method that can effectively process flight parameter data and perform fault detection. This method aims to detect various fault modes of aircraft by analyzing flight parameter data sets. The algorithm includes three functional modules: data set management, fault detection, and algorithm model management. The data set management module is responsible for storing and processing TB-level flight parameter data; the fault detection module uses anomaly detection algorithm models to detect fault modes in the data and display the results through visual charts; the algorithm model management module is responsible for managing various mining algorithm models, including model selection, reasoning, etc. The flight parameter fault detection algorithm of the present invention can improve the accuracy and efficiency of fault detection, reduce aircraft maintenance costs, and improve flight safety.
[0052] See also Figure 1 The flowchart of the fault detection method shown in FIG. 1 includes the following steps:
[0053] Step 101: receiving a user's selection from candidate flight parameter datasets and candidate fault detection tools, and obtaining a target flight parameter dataset and a target fault detection tool; wherein each candidate flight parameter dataset includes flight parameters of any subsystem in any aircraft;
[0054] Step 103: Create a fault detection workflow, wherein the fault detection workflow is used to use the target fault detection tool to detect the fault type of the subsystem corresponding to the target flight parameter dataset. The fault detection workflow is an ordered task combination including multiple data analysis tasks.
[0055] Step 105 : Track the fault detection workflow and output the fault detection result obtained after the fault detection workflow is executed.
[0056] Users can select target datasets and tools from candidate flight parameter datasets and candidate fault detection tools. Flight parameter datasets represent flight parameter data for aircraft subsystems, while tools are models or algorithms used to detect faults. An algorithm is a specific method or procedure for performing a task; it determines how data is processed, features are extracted, and predictions are made. For example, random forests, LSTMs, and TCNs are all different algorithms, each with its own mathematical principles and training methods. A model is a specific instance or product obtained by training an algorithm. Simply put, a model is a summary of the algorithm's application. It is a function or structure derived from training the algorithm on a specific dataset, used to make predictions. It should be understood that not all analytical algorithms require a trained model to make predictions. For example, unsupervised learning algorithms do not rely on a trained explicit model, but instead directly analyze the data to discover patterns. Instead of training a traditional prediction model, they identify anomalies through structural analysis of the data. Some algorithms do not rely on a static model for predictions, but rather on the prediction process itself, dynamically generating the model during the process.
[0057] The user's selection determines the specific object and method of detection. For example, the user can select the flight parameters and corresponding fault detection tools of a certain aircraft engine model to detect engine failure.
[0058] A fault detection workflow is a series of organized tasks and activities that are executed in a specific order and with dependencies to achieve a specific goal or process. A workflow is a concept that encompasses all steps in predicting faults. Create a workflow in the Task Management module. Based on user selections, a fault prediction workflow is created. A workflow includes all steps. For example, the tasks are divided into the following:
[0059] Data selection task: The user selects the dataset for fault prediction.
[0060] Data preprocessing tasks: cleaning, feature extraction, standardization, etc. of data. This may be performed by the data processing module.
[0061] Algorithm selection task: Select a suitable prediction algorithm, such as random forest, LSTM, etc.
[0062] Model training task: Use the selected algorithm to train the data and generate a model. The fault prediction module executes.
[0063] Fault prediction task: Use the trained model to predict faults on new data. This is performed by the fault prediction module.
[0064] Result display task: Display the prediction results to the user in the form of visual charts. This is performed by the visualization module.
[0065] Health record statistics task: Generate and compile health record tables of fault predictions for subsequent analysis. This task is performed by the health record module.
[0066] When you first create a workflow, you need to determine the following: all tasks to be performed; the execution order and dependencies between tasks; the distribution of tasks: different tasks will be performed by different modules or subsystems; and the control of the process: the execution path of the workflow.
[0067] Workflows can be scheduled by a scheduling system. It is understandable that the specific implementation of the scheduling system may include at least one hardware module and / or software module to realize the functions of the scheduling system. The scheduling system can decompose and distribute workflows. Specifically, the decomposition and arrangement of tasks can be based on the following factors: the specific type of task: such as data cleaning, feature extraction, model training, result presentation, etc.; the dependency relationship of tasks: the execution order of tasks is determined based on the dependency relationship between tasks; resource requirements: based on the computing resources and time required for each task, the scheduling system may execute multiple unrelated tasks in parallel, or assign tasks with higher priority to be executed first. Different tasks can be executed by different task processing modules, and different task processing modules can perform at least one data analysis task. The scheduling system can schedule the task processing module corresponding to the data analysis task to perform task processing.
[0068] The embodiments of the present application can flexibly select detection objects and tools according to user needs, thereby improving the pertinence and efficiency of detection.
[0069] In some embodiments, the fault detection workflow is tracked and the fault detection results obtained after the fault detection workflow is executed are output, including: parsing the dependency relationship of multiple data analysis tasks in the fault detection workflow to obtain a directed acyclic graph corresponding to the fault detection workflow; based on the directed acyclic graph, scheduling the task processing modules corresponding to each data analysis task to execute processing in sequence; and obtaining and outputting the fault detection results based on the processing results of each data analysis task.
[0070] During the execution of the fault detection workflow, the scheduling system analyzes the dependencies of the data analysis tasks in the workflow to construct a directed acyclic graph (DAG). A DAG is a graphical structure that clearly represents the order of tasks. For example, if the result of Task A is the input of Task B, then in the DAG, Task A will be executed before Task B. Based on the DAG, the system schedules task processing modules sequentially to ensure that each task is executed correctly. Finally, the fault detection results are output based on the task processing results. Optimizing task scheduling through DAG can improve the accuracy and efficiency of fault detection.
[0071] In some embodiments, based on a directed acyclic graph, the task processing modules corresponding to each data analysis task are scheduled to execute processing in sequence, including: based on the directed acyclic graph, multiple data analysis tasks are divided into executable tasks and dependent tasks, wherein the executable tasks are data analysis tasks that do not depend on other data analysis tasks, and the dependent tasks are data analysis tasks that depend on the analysis results of other data analysis tasks; the executable tasks are placed into an executable task pool, and the dependent tasks are placed into a dependent task pool; the task processing modules corresponding to the executable tasks in the executable task pool are scheduled to execute processing; after any executable task in the executable task pool is processed, the dependent tasks in the dependent task pool that depend on the executable tasks are placed into the executable task pool.
[0072] The embodiment of the present application further refines the DAG-based task scheduling method. First, the data analysis tasks are divided into executable tasks and dependent tasks. Executable tasks can be executed directly, while dependent tasks need to wait for other tasks to be completed before execution. For example, in a fault detection workflow, task A can be executed directly, while task B needs to wait for the result of task A. The scheduling system puts executable tasks into the executable task pool and dependent tasks into the dependent task pool. After executing an executable task, the dependent tasks that depend on the task are moved into the executable task pool. By dynamically managing tasks through the task pool and further optimizing task scheduling, system efficiency can be improved.
[0073] In some embodiments, after any executable task in the executable task pool is processed, the dependent task of the dependent task in the dependent task pool is placed into the executable task pool, including: after any executable task in the executable task pool is processed, the dependent task of the dependent task in the dependent task pool is changed to an executable task; when the preset time is reached, the dependent task pool is scanned and the executable tasks in the dependent task pool are transferred to the executable task pool.
[0074] The embodiment of the present application further introduces a time mechanism. When an executable task is completed, the system will check the dependent task pool and mark the dependent tasks that depend on it as executable tasks. In addition, if the preset time is reached, the system will scan the dependent task pool and move the tasks that can be executed into the executable task pool. For example, if task A is not completed within the specified time, but task B can be executed, the system will move task B into the executable task pool. The time mechanism is used to avoid tasks waiting for too long and improve the flexibility and efficiency of task scheduling.
[0075] In some embodiments, creating a fault detection workflow includes: obtaining an index identifier of a target flight parameter data set and an index identifier of a target fault detection tool; constructing multiple data analysis tasks based on templates of multiple data analysis tasks; the multiple data analysis tasks are obtained by filling the index identifiers into the templates; and creating a fault detection workflow based on the multiple data analysis tasks.
[0076] For example, the process of creating a fault detection workflow can specifically include the following steps: First, obtain the index identifiers of the target flight parameter dataset and the target fault detection tool. The index identifier is a unique identifier used to quickly locate the data and tool. Then, based on the data analysis task template, the index identifier is filled into the template to construct multiple data analysis tasks. These tasks are combined into a fault detection workflow. For example, the template may be a general fault detection algorithm framework, and by filling in the index identifier, tasks for specific datasets and tools are generated. This template-based approach allows for rapid workflow construction, improving the scalability and flexibility of the system.
[0077] In some embodiments, the tracking fault detection workflow includes: obtaining a target flight parameter data set based on the index identifier of the target flight parameter data set; obtaining a target fault detection tool based on the index identifier of the target fault detection tool; and using the target fault detection tool to detect the fault type existing in the subsystem corresponding to the target flight parameter data set.
[0078] During the fault detection workflow tracking process, the target flight parameter dataset is first retrieved based on its index identifier. The target fault detection tool is then retrieved based on its index identifier. The tool is then used to detect the subsystem fault type corresponding to the dataset. For example, the flight parameter data and the corresponding detection tool for an aircraft engine are retrieved, and the tool is used to analyze the data and determine whether the engine is faulty. Rapid data and tool retrieval based on index identifiers ensures accurate and efficient fault detection.
[0079] In some embodiments, a target fault detection tool is used to detect the type of fault occurring in the subsystem corresponding to the target flight parameter data set, including: performing preprocessing on the target flight parameter data set; constructing statistical features of the target flight parameter data set based on the preprocessing results; and inputting the statistical features into the target fault detection tool to obtain the type of fault existing in the subsystem.
[0080] When using a targeted fault detection tool to detect faults, the target flight parameter dataset is first preprocessed, such as by removing noise and standardizing it. Statistical features, such as the mean and variance, are then constructed based on the preprocessing results. Finally, these statistical features are input into the fault detection tool to determine the fault type. For example, after preprocessing engine flight parameter data, features such as speed and temperature are extracted and input into the tool to determine whether the engine is overheating. Preprocessing and feature extraction improve the accuracy and reliability of fault detection.
[0081] In a specific application scenario example, the fault detection method may be applied to a fault detection system, which may include:
[0082] The data management module is used to store flight parameter data uploaded by users and supports TB-level data storage.
[0083] The fault detection module detects various fault modes in the data based on the flight parameter dataset selected by the user through relevant anomaly detection algorithm models, and presents them to the end user through visual charts.
[0084] The algorithm model management module is responsible for the management of various mining algorithm models, including model selection and reasoning.
[0085] The report display module records the results of fault analysis and detection to form an aircraft health record sheet.
[0086] The health management module collects fault information based on the results of the fault detection module to form a health record table.
[0087] The cockpit dashboard report module quickly and conveniently displays the fault status of each data set, analyzes task statistics, and helps users understand the latest analysis trends.
[0088] refer to Figure 2 , the dataset management module can also perform the following steps:
[0089] Enter data source information, including dataset name, description, data address, system name of data source, aircraft number, and aircraft subsystem;
[0090] Upload raw data to the system;
[0091] Automatically identify relevant field data types in uploaded data and perform data consistency checks;
[0092] Perform data cleaning, including checking for outliers and missing values;
[0093] Based on user settings, the system handles outliers and missing values, including discarding records with incorrect data types or obviously abnormal data values, and not processing missing values, discarding records, or filling them with the average value.
[0094] The cleaned data is converted into the specified format;
[0095] The converted data is stored in the database or file system.
[0096] Furthermore, in the data verification and cleaning steps, the system uses the 3sigma method to check whether there are obvious outliers in the numerical data; in the basic preprocessing step, the system normalizes, standardizes and quantitatively binarizes the input data; in the basic preprocessing step, the system uses binning, mean filtering and other methods to reduce the interference caused by noise; in the basic preprocessing step, the system performs operations such as logarithm, inverse, square root, power transformation and other operations according to the characteristics of the data; in the time series feature processing step, the system constructs statistical features, autocorrelation, periodicity, trend, frequency, time dimension features, window dimension features and frequency domain features; in the time series feature processing step, the system constructs features based on the characteristics of the data, including mean, standard deviation, extreme value, quantile, number of spikes, number of missing values, deviation, autocorrelation, periodicity, trend, frequency, duration, time interval, statistical features based on different time granularities, principal component analysis PCA, statistical features within a sliding time window, statistical features within a rolling time window, Fourier transform, short-time Fourier transform, wavelet transform and constructs a spectrum diagram.
[0097] refer to Figure 3 and Figure 4 After receiving the user's selection of a dataset, analysis algorithm, and model, the system automatically creates a fault detection workflow and tracks processing status. Once the workflow is created, it can be broken down into multiple data analysis tasks and executed by the scheduling system. The fault detection workflow is then placed in a waiting queue for scheduling. The scheduler analyzes the dependencies between the processing logic in the workflow and decomposes the workflow into a directed acyclic graph. Executable tasks are placed in the execution task pool, and dependent tasks are placed in the pending task pool (i.e., the dependent task pool). After a task is executed, the scheduler scans the pending task pool and moves the tasks that become executable into the execution task pool.
[0098] Fault detection and analysis dimensions include time, sensors, and faults. By observing the temporal distribution of fault events, a model for the time intervals between equipment faults is established. A time series model of normal sensor observations is built to test observation data at future time points. Specific detection models are developed for typical fault types, and fault diagnosis models are established based on the degree and combination of sensor anomalies.
[0099] A time series model is an artificial intelligence model used to predict future sensor parameter time series based on a time series of sensor parameters over a period of time. This model, based on current changes in sensor distribution, can predict future parameter trends and indicate potential or impending failures before sensor parameters exceed alarm thresholds. Time series models predict future sensor observations and compare them with actual sensor data. By calculating the difference between predicted and actual observations, the degree of sensor anomaly can be quantified. The difference between actual and predicted observations is expressed using error metrics such as absolute error, mean squared error, and standardized error. When the error exceeds a certain threshold, an anomaly is considered to have occurred. The larger the error, the greater the degree of anomaly. The threshold is typically selected based on experience or adaptively adjusted using training data.
[0100] The fault detection process can be divided into three stages:
[0101] In the first stage, an unsupervised algorithm is used to train the reconstruction model of normal data;
[0102] The second stage uses data reconstruction errors to find time series data that may contain fault signs and train a fault recognition model;
[0103] In the third stage, the data to be processed is input into the fault identification model to obtain information about the fault and fault type.
[0104] The algorithm model management module may include performing the following steps:
[0105] Maintain information such as the input data format, output description, parameter settings, application conditions and instructions of the algorithm model;
[0106] Select the corresponding algorithm model in the algorithm model library for data analysis according to different needs.
[0107] All models in the algorithm model management method can be added through configuration. Place the model file in the specified directory, maintain the relevant information in the model master data, and restart the system. All algorithms can be registered through configuration. Place the algorithm executable file in the specified directory, maintain the relevant information in the algorithm master data, and restart the system.
[0108] refer to Figure 4, Spring boot and Spring cloud are the core frameworks in modern microservice architecture. They are used to simplify the development of single services and manage services in distributed systems respectively. Boot: Simplifies the configuration and development process of Java applications, allowing developers to quickly build independent, production-level Spring applications. As the core framework of the application, it is used to develop various service modules, such as task scheduling management, task queues, task execution, etc. Spring Boot is responsible for handling the functions of interacting with the database. Cloud: Built on Spring boot, it aims to simplify the development and management of microservice architecture. It is used for communication, service registration and discovery between multiple service instances to ensure that the task scheduling system can run in a distributed environment, and perform load balancing and fault recovery. Spring Cloud is used to coordinate and manage the execution of algorithm packages. It is understood that all algorithms must comply with specific interface specifications, including but not limited to: the algorithm function must be written in Python, and the third-party packages it references must be within the scope supported by the system and conform to the relevant syntax of Python 3.9; the entry function of the fault detection algorithm is run(input:dataframe, parameter:dict, device:string)->result:Abnormal, where the input field is the data in the dataset, input in dataframe format, the parameter field is the algorithm parameter, input in dictionary format, the device field is the algorithm running device, CPU or CUDA, and the output result is the Abnormal class, which contains the detected fault type and the index where the abnormal data is located, as well as the status information of the algorithm running result. Reference Figure 5 , is a schematic diagram of an example fault detection result output, including the fault records of each aircraft subsystem.
[0109] In this embodiment, a method for processing flight parameter data and detecting faults is provided, and the process includes the following steps:
[0110] In step S1011, the user uploads the flight parameter data set to the flight parameter fault detection system and fills in the data source information, such as the data set name, description information, data address, system name of the data source, aircraft number, aircraft subsystem, etc.
[0111] Step S1012: The fault detection system identifies the data type of the relevant fields according to the uploaded data content and performs a data consistency check.
[0112] Step S1013: perform data cleaning, check for outliers and missing values, and perform relevant processing according to user settings, including discarding records with incorrect data types or obviously abnormal data values, and not processing missing values, discarding records, or filling them with average values.
[0113] Step S1014: convert the cleaned data into a format that is stored in a database or file system.
[0114] Step S1015 : The user selects a data set and a corresponding analysis algorithm or model.
[0115] In step S1016 , the task management module automatically creates a fault detection workflow to track the processing status of the workflow and feeds back the processing results to the user interface.
[0116] In step S1017, the scheduling system decomposes the workflow into multiple data analysis tasks and schedules them for execution.
[0117] In step S1018, the fault detection module detects various fault modes in the data based on the flight parameter data set selected by the user through the relevant anomaly detection algorithm model.
[0118] Step S1021: Display the fault detection result to the end user through a visual chart.
[0119] In step S1022, the health management module collects fault information according to the results of the fault detection module to form a health record table.
[0120] In some embodiments, the following steps may also be included:
[0121] Step S201 : perform basic preprocessing on the input data, including normalization, standardization, quantitative binarization, and noise reduction processing such as binning and mean filtering.
[0122] Step S202: construct time series features, including statistical features, time dimension features, window dimension features, and frequency domain features.
[0123] Step S203 , fault detection and analysis are performed from the dimensions of time, sensor, and fault, to detect abnormal state signs of different sensors and sensor combinations at different time granularities, as well as their possible corresponding fault categories.
[0124] Step S204, establish a time series model of normal sensor observation values, detect observation data at future time points, and when there is a large deviation between the detection data of the time series model and the actual observation data of the sensor, it means that the distribution of sensor data has changed, and there is a high possibility that a fault or a sign of a fault may have occurred.
[0125] Step S205 , establishing a specific detection model for a typical fault type, and establishing a fault diagnosis model based on the degree of sensor abnormality and the combination of sensors in which the abnormality occurs.
[0126] An embodiment of the present invention provides a flight parameter data processing and fault detection method. By integrating functions such as flight parameter data set management, fault detection, algorithm model management, and report display, it can generate high-coverage fault detection that meets specified coverage criteria for the flight parameter data to be tested in a short period of time, thereby improving aircraft maintenance efficiency and flight safety, and reducing maintenance costs.
[0127] The embodiment of the present invention comprehensively compares and analyzes the feasibility of no less than 6 algorithms in fault detection and other aspects, conducts detailed performance analysis and parameter tuning, meets the index requirements of no less than 6 algorithms, and has a detection accuracy rate of no less than 90% for typical faults that have occurred.
[0128] This embodiment of the present invention effectively shortens the runtime of fault detection algorithms through GPU acceleration and parallel computing, ensuring a runtime of less than one second on medium-sized datasets (MB level). Furthermore, model quantization and compression technology can further increase the inference speed of the algorithm model by 2-10 times.
[0129] The embodiment of the present invention supports TB-level data processing. The task scheduling module of the flight parameter fault detection system will split the input data set that is too large, cut the data in batches using an overlapping moving window method with an adjustable step size, and input the data of appropriate size into the algorithm module. Each time the overlapping window slides, the window will retain a portion of the same data as the previous window. Therefore, there is overlap between windows. For example, if the window size is N and the step size is S, each time it slides, the window will move forward S data points, but will retain NS data points to form an overlapping part. The overlapping part helps to maintain the continuity of time series data, especially when capturing data change trends, it can reduce information loss.
[0130] Figure 6 FIG. 1 is a block diagram of an electronic device for implementing an embodiment of the present application. Figure 6 As shown, the electronic device includes a memory 601 and a processor 602. The memory 601 stores a computer program executable by the processor 602. When the processor 602 executes the computer program, the method described in the above embodiment is implemented. The number of the memory 601 and the processor 602 can be one or more. In a specific implementation, the electronic device may also include a communication interface 603 for communicating with external devices and exchanging data.
[0131] In a specific implementation, if the memory 601, the processor 602, and the communication interface 603 are implemented independently, the memory 601, the processor 602, and the communication interface 603 can be connected to each other via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.
[0132] Optionally, in a specific implementation, if the memory 601 , the processor 602 , and the communication interface 603 are integrated on a chip, the memory 601 , the processor 602 , and the communication interface 603 may communicate with each other through an internal interface.
[0133] An embodiment of the present application provides a computer-readable storage medium storing a computer program, which implements the method provided in the embodiment of the present application when the program is executed by a processor.
[0134] An embodiment of the present application provides a computer program product, including a computer program, which implements the method provided in the embodiment of the present application when executed by a processor.
[0135] An embodiment of the present application also provides a chip, which includes a processor for calling and executing instructions stored in the memory from the memory, so that a communication device equipped with the chip executes the method provided in the embodiment of the present application.
[0136] An embodiment of the present application also provides a chip, including: an input interface, an output interface, a processor and a memory. The input interface, the output interface, the processor and the memory are connected through an internal connection path. The processor is used to execute the code in the memory. When the code is executed, the processor is used to execute the method provided in the embodiment of the application.
[0137] It should be understood that the processor may be a CPU (Central Processing Unit), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. It is worth noting that the processor may be a processor that supports the Advanced RISC Machines (ARM) architecture.
[0138] Furthermore, optionally, the above-mentioned memory may include a read-only memory and a random access memory. The memory may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory. Among them, the non-volatile memory may include a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may include a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM) and direct memory bus random access memory (DR RAM).
[0139] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the present application are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.
[0140] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.
[0141] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. Throughout the description of this application, "plurality" means two or more, unless otherwise specifically defined.
[0142] Any process or method described in the flowchart or otherwise described herein can be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a specific logical function or process. The scope of the preferred embodiments of the present application includes other implementations in which the functions may be performed in a different order than shown or discussed, including performing the functions substantially simultaneously or in reverse order depending on the functions involved.
[0143] The logic and / or steps described in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor or other system that can fetch instructions from an instruction execution system, apparatus or device and execute instructions), or used in combination with such instruction execution systems, apparatuses or devices.
[0144] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. All or part of the steps of the above embodiment method can be completed by instructing the relevant hardware through a program, which can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.
[0145] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the aforementioned integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium. The storage medium may be a read-only memory, a magnetic disk, or an optical disk, etc.
[0146] The above is merely an exemplary embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily conceive of various modifications or substitutions within the technical scope described in this application, and such modifications or substitutions should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.
Claims
1. A fault detection method, characterized in that: include: receiving user selections from candidate flight parameter data sets and candidate fault detection tools, and obtaining target flight parameter data sets and target fault detection tools; wherein each candidate flight parameter data set includes flight parameters of any subsystem in any aircraft; Creating a fault detection workflow, wherein the fault detection workflow is used to use the target fault detection tool to detect the fault type occurring in the subsystem corresponding to the target flight parameter dataset, and the fault detection workflow is an ordered task combination including multiple data analysis tasks; The fault detection workflow is tracked, and a fault detection result obtained after the fault detection workflow is executed is output.
2. The method according to claim 1, characterized in that The tracking of the fault detection workflow and outputting the fault detection result obtained after the fault detection workflow is executed include: Analyzing the dependencies of the multiple data analysis tasks in the fault detection workflow to obtain a directed acyclic graph corresponding to the fault detection workflow; Based on the directed acyclic graph, scheduling the task processing modules corresponding to the data analysis tasks to perform processing in sequence; According to the processing results of each data analysis task, the fault detection result is obtained and output.
3. The method according to claim 2, characterized in that The step of sequentially scheduling the task processing modules corresponding to the data analysis tasks to perform processing based on the directed acyclic graph includes: Based on the directed acyclic graph, the multiple data analysis tasks are divided into executable tasks and dependent tasks, wherein the executable tasks are data analysis tasks that do not depend on other data analysis tasks, and the dependent tasks are data analysis tasks that depend on analysis results of other data analysis tasks; Put the executable task into the executable task pool, and put the dependent task into the dependent task pool; Scheduling the task processing modules corresponding to the executable tasks in the executable task pool to execute processing; After any one of the executable tasks in the executable task pool is processed, the dependent tasks in the dependent task pool that depend on the executable task are put into the executable task pool.
4. The method according to claim 3, characterized in that After any one of the executable tasks in the executable task pool is processed, placing the dependent tasks in the dependent task pool that depend on the executable task into the executable task pool includes: After any of the executable tasks in the executable task pool is processed, the dependent tasks in the dependent task pool that depend on the executable task are changed into executable tasks; When the preset time is reached, the dependent task pool is scanned, and the executable tasks in the dependent task pool are transferred to the executable task pool.
5. The method according to claim 1, wherein The creation of a fault detection workflow includes: Obtaining an index identifier of the target flight parameter data set and an index identifier of the target fault detection tool; Constructing the plurality of data analysis tasks according to the template of the plurality of data analysis tasks; the plurality of data analysis tasks are obtained by filling the template with the index identifier; The fault detection workflow is created according to the multiple data analysis tasks.
6. The method according to claim 5, characterized in that Tracking the fault detection workflow includes: Acquire the target flight parameter data set according to the index identifier of the target flight parameter data set; Acquire the target fault detection tool according to the index identifier of the target fault detection tool; Use the target fault detection tool to detect the fault type existing in the subsystem corresponding to the target flight parameter data set.
7. The method according to claim 6, characterized in that The detecting tool using the target fault to detect the type of fault occurring in the subsystem corresponding to the target flight parameter dataset includes: Performing preprocessing on the target flight parameter data set; Constructing statistical features of the target flight parameter data set based on the preprocessing results; The statistical features are input into the target fault detection tool to obtain the fault type existing in the subsystem.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory, wherein the processor implements the method according to any one of claims 1 to 7 when executing the computer program.
9. A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.
10. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.