CN114417501A - Airborne deployment-oriented health management predictive modeling method - Google Patents
Airborne deployment-oriented health management predictive modeling methodDownload PDFInfo
- Publication number
- CN114417501A CN114417501ACN202111663607.4ACN202111663607ACN114417501ACN 114417501 ACN114417501 ACN 114417501ACN 202111663607 ACN202111663607 ACN 202111663607ACN 114417501 ACN114417501 ACN 114417501A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- airborne
- phm
- operator
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Automation & Control Theory (AREA)
- Aviation & Aerospace Engineering (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Description
Translated fromChinese技术领域technical field
本发明涉及但不限于航空系统可靠性领域，涉及地面/机载健康管理，尤指一种面向机载部署的健康管理预测性建模方法。The present invention relates to, but is not limited to, the field of aviation system reliability, relates to ground/airborne health management, and particularly relates to an airborne deployment-oriented predictive modeling method for health management.
背景技术Background technique
航空健康管理与故障诊断系统中，机载和地面平台的计算资源、约束条件不同，预测与健康管理(PHM)处理的不同阶段需要在不同的硬件环境下运行；具体的，地面平台具有不受限的计算性能、存储空间、以及丰富的数据分析工具包等，而机载环境计算资源有限，要求功耗低，存储空间小。In the aviation health management and fault diagnosis system, the computing resources and constraints of the airborne and ground platforms are different, and different stages of prediction and health management (PHM) processing need to run in different hardware environments; However, the airborne environment requires limited computing performance, storage space, and rich data analysis toolkits, which requires low power consumption and small storage space.
目前，还没有一套完整的适用于航空机载部署的PHM建模方法和相应的算子库。因此，针对航空系统的PHM需要建立一套完整的建模策略，实现规范的数据采集分析，发挥地面平台优势进行数据处理和模型训练，在模型评估及模型决策时保证机载运行时间和准确度，最后将选出的最优模型部署在机载端。At present, there is no complete set of PHM modeling method and corresponding operator library suitable for airborne deployment. Therefore, for the PHM of the aviation system, it is necessary to establish a complete set of modeling strategies to realize standardized data collection and analysis, to take advantage of the ground platform for data processing and model training, and to ensure the airborne running time and accuracy during model evaluation and model decision-making. , and finally deploy the selected optimal model on the airborne side.
发明内容SUMMARY OF THE INVENTION
本发明的目的：本发明针对面向机载的PHM部署需求，考虑地面与机载环境的差异，提出一种面向机载部署的健康管理预测性建模方法，通过在地面阶段进行数据处理及模型训练，在原理样机中进行评估选择和决策，最终部署于机载的全周期PHM建模策略。Purpose of the present invention: The present invention proposes an airborne deployment-oriented health management predictive modeling method for airborne PHM deployment requirements, considering the difference between the ground and airborne environments, by performing data processing and modeling in the ground phase Training, evaluation selection and decision-making in principle prototypes, and eventual deployment in airborne full-cycle PHM modeling strategies.
本发明的技术方案：本发明实施例提出一种面向机载部署的健康管理预测性建模方法，包括：Technical solution of the present invention: The embodiment of the present invention proposes an airborne deployment-oriented health management predictive modeling method, including:
步骤1，在地面平台对机载数据和地面数据进行数据处理及模型训练，得到多个训练模型；Step 1: Data processing and model training are performed on the airborne data and ground data on the ground platform to obtain multiple training models;
步骤2，在原理样机中对步骤1中得到的训练模型进行评估选择和决策，得到符合任务需求的训练模型；In step 2, the training model obtained in step 1 is evaluated, selected and decided in the principle prototype to obtain a training model that meets the requirements of the task;
步骤3，将步骤2得到的符合任务需求的训练模型部署于机载中；Step 3, deploying the training model that meets the mission requirements obtained in step 2 in the airborne;
通过上述步骤1～3，得到机载的全周期PHM建模策略。Through the above steps 1 to 3, the airborne full-cycle PHM modeling strategy is obtained.
可选地，如上所述的面向机载部署的健康管理预测性建模方法中，所述步骤1包括：运行于地面平台的数据导入阶段、数据处理阶段和模型训练阶段；具体包括：Optionally, in the above-mentioned airborne deployment-oriented health management predictive modeling method, the step 1 includes: a data import stage, a data processing stage and a model training stage running on the ground platform; specifically:
步骤1.1，数据导入阶段，包括：在地面高性能平台中，导入各类结构化数据及非结构化数据；Step 1.1, the data import stage, includes: importing various structured data and unstructured data into the ground high-performance platform;
步骤1.2，数据处理阶段，包括：数据探索分析子阶段、数据预处理子阶段和特征工程子阶段；Step 1.2, data processing stage, including: data exploration and analysis sub-stage, data preprocessing sub-stage and feature engineering sub-stage;
步骤1.3，模型训练阶段，包含：通过基于物理或经验模型的模型训练和基于数据驱动的模型训练，训练得到针对特定的PHM任务的多个算法模型。Step 1.3, the model training phase, includes: through model training based on physical or empirical models and data-driven model training, training to obtain multiple algorithm models for specific PHM tasks.
可选地，如上所述的面向机载部署的健康管理预测性建模方法中，所述步骤1.2中，Optionally, in the above-mentioned airborne deployment-oriented health management predictive modeling method, in step 1.2,
数据探索分析子阶段，包括：对导入数据进行直观探索和简单挖掘，得到样本数量、特征数量，数据分布特征、特征趋势性、相关性；Data exploration and analysis sub-stage, including: intuitive exploration and simple mining of imported data to obtain the number of samples, the number of features, data distribution characteristics, feature trends, and correlations;
数据预处理子阶段用于实现对数据质量的提升，包括：数据清洗、数据去噪、数据标准化；The data preprocessing sub-stage is used to improve data quality, including: data cleaning, data denoising, and data standardization;
特征工程子阶段，包含：对数据的特征挖掘、特征选择、特征提取，以提取出对具体任务(如故障诊断、寿命预测)最有用的特征。The feature engineering sub-stage includes: feature mining, feature selection, and feature extraction of data to extract the most useful features for specific tasks (such as fault diagnosis, life prediction).
可选地，如上所述的面向机载部署的健康管理预测性建模方法中，所述步骤1.3中的PHM任务包括：状态监测、故障诊断、故障预测、剩余寿命预测；Optionally, in the above airborne deployment-oriented health management predictive modeling method, the PHM tasks in step 1.3 include: condition monitoring, fault diagnosis, fault prediction, and remaining life prediction;
针对特定的PHM任务类型，对其进行问题分析及解构的方式为：For a specific PHM task type, the problem analysis and deconstruction methods are as follows:
方式1，根据对分析对象的认知，在对象存在容易求解的物理或经验模型时，则针对具体任务，对该对象建立对应的物理模型，以进行模型训练和求解；Method 1: According to the cognition of the analysis object, when the object has a physical or empirical model that is easy to solve, a corresponding physical model is established for the object according to the specific task, so as to carry out model training and solution;
方式2，当对象结构复杂，失效或退化机理难以得到，则采用数据驱动的方法，利用相应机器学习，统计分析等方法选择算法并进行模型训练；Method 2: When the structure of the object is complex, and the failure or degradation mechanism is difficult to obtain, a data-driven method is adopted, and the corresponding machine learning, statistical analysis and other methods are used to select algorithms and conduct model training;
其中，所述步骤1.3的模型训练阶段，采取多种算法，进行多个模型的训练，以得到多个训练好的算法模型。Wherein, in the model training phase of step 1.3, multiple algorithms are adopted to train multiple models, so as to obtain multiple trained algorithm models.
可选地，如上所述的面向机载部署的健康管理预测性建模方法中，所述步骤2包括：运行于与机载软硬件环境相似的原理样机的模型评估阶段和模型决策阶段；具体包括：Optionally, in the above-mentioned airborne deployment-oriented health management predictive modeling method, the step 2 includes: a model evaluation stage and a model decision stage that operate on a principle prototype similar to the airborne software and hardware environment; specifically include:
步骤2.1，模型评估阶段，在原理样机中对多个训练好的算法模型进行评估，根据算法模型类型以及任务要求，选择多种合适的模型评估指标，将多个在地面训练好的算法模型运行于原理样机中，得到模型评估表；Step 2.1, model evaluation stage, evaluate multiple trained algorithm models in the principle prototype, select a variety of appropriate model evaluation indicators according to the type of algorithm model and task requirements, and run multiple algorithm models trained on the ground In the principle prototype, the model evaluation table is obtained;
步骤2.2，模型决策阶段，根据指定的任务要求制定模型决策规则，并结合得到的模型评估表，对模型进行最终决策，选择出对于该具体任务的“最优模型”。Step 2.2, the model decision-making stage, formulate model decision-making rules according to the specified task requirements, and combine the obtained model evaluation table to make a final decision on the model, and select the "optimal model" for the specific task.
可选地，如上所述的面向机载部署的健康管理预测性建模方法中，所述步骤3包括：Optionally, in the above airborne deployment-oriented health management predictive modeling method, the step 3 includes:
在模型部署阶段，将选出的“最优模型”采用机载硬件支持的语言进行封装，并将封装后的“最优模型”部署于机载端。In the model deployment stage, the selected "optimal model" is packaged in a language supported by the airborne hardware, and the packaged "optimal model" is deployed on the airborne end.
可选地，如上所述的面向机载部署的健康管理预测性建模方法中，还包括：Optionally, the above airborne deployment-oriented health management predictive modeling method further includes:
构建机载PHM算子库，包括：根据步骤1到步骤3中的各阶段，梳理得到PHM算子库；Constructing the airborne PHM operator library, including: sorting out the PHM operator library according to each stage in step 1 to step 3;
其中，所述机载PHM算子库按照数据处理流程分类，包含：流程模块和支撑性模块；Wherein, the airborne PHM operator library is classified according to the data processing flow, and includes: a flow module and a supporting module;
所述机载PHM算子库按照算子功能分类，包含：数据处理通用算子、PHM任务专用算子和集成化模块化的飞机部件级/系统级/全机级PHM算子。The airborne PHM operator library is classified according to operator functions, including: data processing general operators, PHM task-specific operators, and integrated and modular aircraft component-level/system-level/full-aircraft-level PHM operators.
可选地，如上所述的面向机载部署的健康管理预测性建模方法中，Optionally, in the above-mentioned airborne deployment-oriented health management predictive modeling method,
所述数据处理通用算子，包括：数据导入算子单元、数据基本操作算子单元、数据预处理算子单元、数据探索分析算子单元、特征工程算子单元、机器学习算子单元和超参数优化算子单元；The general data processing operator includes: a data import operator unit, a data basic operation operator unit, a data preprocessing operator unit, a data exploration and analysis operator unit, a feature engineering operator unit, a machine learning operator unit, and a supercomputer unit. Parameter optimization operator unit;
所述PHM任务专用算子，包括：专家系统算子单元、异常监测算子单元、故障诊断算子单元、寿命预测算子单元等PHM任务专用算子；其中，专家系统算子单元包含专家知识，基于经验的模型，基于物理失效机制的模型等。The PHM task-specific operators include: an expert system operator unit, an anomaly monitoring operator unit, a fault diagnosis operator unit, a life prediction operator unit and other PHM task-specific operators; wherein, the expert system operator unit contains expert knowledge , experience-based models, models based on physical failure mechanisms, etc.
所述集成化模块化的飞机部件级/系统级/全机级PHM算子，为针对飞机特定部件、特定系统或全机的高集成化状态监测、故障诊断及寿命预测算子；所述飞机部件级/系统级/全机级PHM算子为根据全流程机载PHM算法处理策略，在对应的各功能算子中选择特定的数据处理通用算子和PHM专用算子，搭建而成的部件或系统的模块化算子。The integrated and modularized aircraft component-level/system-level/full-aircraft-level PHM operator is a highly integrated condition monitoring, fault diagnosis and life prediction operator for specific parts, specific systems or the whole aircraft of the aircraft; the aircraft Component-level/system-level/full-machine-level PHM operators are components constructed by selecting specific general-purpose operators for data processing and PHM-specific operators from the corresponding functional operators according to the processing strategy of the airborne PHM algorithm in the whole process. Or the modular operator of the system.
可选地，如上所述的面向机载部署的健康管理预测性建模方法中，Optionally, in the above-mentioned airborne deployment-oriented health management predictive modeling method,
所述流程模块包括：数据导入算子单元，数据探索分析算子单元，数据预处理算子单元，特征工程算子单元，模型训练算子单元，模型评估算子单元，模型决策算子单元，模型部署算子单元；The process module includes: a data import operator unit, a data exploration and analysis operator unit, a data preprocessing operator unit, a feature engineering operator unit, a model training operator unit, a model evaluation operator unit, and a model decision operator unit, Model deployment operator unit;
所述支撑性模块包括：数据基本操作算子单元、机器学习算子单元、专家系统算子单元和超参数优化算子单元；The supporting module includes: a basic data operation operator unit, a machine learning operator unit, an expert system operator unit and a hyperparameter optimization operator unit;
其中，数据基本操作算子单元作为支撑整个流程模块的基础算子；机器学习算子单元、专家系统算子单元和超参数优化算子单元作为支撑模型训练算子单元的支撑性算子。Among them, the basic data operation operator unit is used as the basic operator supporting the entire process module; the machine learning operator unit, the expert system operator unit and the hyperparameter optimization operator unit are used as the supporting operators supporting the model training operator unit.
本发明的有益效果：本发明实施例提供的面向机载部署的健康管理预测性建模方法，一方面，通过借助地面平台的高性能计算机，以及与机载环境相同的原理样机，实现利用海量异构数据的分析，问题建模以及模型决策，最终选择面向具体任务的“最优模型”进行模型部署，实现面向机载部署的全周期PHM建模策略；另一方面，通过将面向机载部署的全周期PHM建模策略具象化为具体的算子，提出PHM算子库，完整支撑面向机载的PHM建模过程。Beneficial effects of the present invention: The airborne deployment-oriented predictive modeling method for health management provided by the embodiments of the present invention, on the one hand, realizes the utilization of massive Analysis of heterogeneous data, problem modeling and model decision-making, and finally select the "optimal model" for specific tasks for model deployment, and realize the full-cycle PHM modeling strategy for airborne deployment; The deployed full-cycle PHM modeling strategy is visualized as specific operators, and a PHM operator library is proposed to fully support the airborne-oriented PHM modeling process.
附图说明Description of drawings
附图用来提供对本发明技术方案的进一步理解，并且构成说明书的一部分，与本申请的实施例一起用于解释本发明的技术方案，并不构成对本发明技术方案的限制。The accompanying drawings are used to provide a further understanding of the technical solutions of the present invention, and constitute a part of the specification. They are used to explain the technical solutions of the present invention together with the embodiments of the present application, and do not limit the technical solutions of the present invention.
图1为本发明实施例提供的一种面向机载部署的健康管理预测性建模方法的原理示意图；FIG. 1 is a schematic diagram of the principle of an airborne deployment-oriented predictive modeling method for health management provided by an embodiment of the present invention;
图2为本发明实施例中模型评估阶段与模型决策阶段的原理示意图；2 is a schematic diagram of the principles of a model evaluation stage and a model decision stage in an embodiment of the present invention;
图3为本发明实施例中机载PHM算子库的示意图。FIG. 3 is a schematic diagram of an airborne PHM operator library in an embodiment of the present invention.
具体实施方式Detailed ways
为使本发明的目的、技术方案和优点更加清楚明白，下文中将结合附图对本发明的实施例进行详细说明。需要说明的是，在不冲突的情况下，本申请中的实施例及实施例中的特征可以相互任意组合。In order to make the objectives, technical solutions and advantages of the present invention clearer, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that, the embodiments in the present application and the features in the embodiments may be arbitrarily combined with each other if there is no conflict.
本发明提供以下几个具体的实施例可以相互结合，对于相同或相似的概念或过程可能在某些实施例不再赘述。The present invention provides the following specific embodiments that can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.
本发明实施例的目的是建立一种面向机载部署的PHM建模方法并搭建PHM算子库，可以完整支撑航空地面/机上PHM数据分析和建模过程，快速便捷实现飞机部件级/系统级/全机级成员的故障诊断与寿命分析，对飞机的健康状态做出评估。The purpose of the embodiments of the present invention is to establish a PHM modeling method for airborne deployment and build a PHM operator library, which can fully support the aviation ground/onboard PHM data analysis and modeling process, and quickly and conveniently realize the aircraft component level/system level. /Fault diagnosis and life analysis of all aircraft-level members to evaluate the health status of the aircraft.
图1为本发明实施例提供的一种面向机载部署的健康管理预测性建模方法的原理示意图。本发明实施例提供的面向机载部署的健康管理预测性建模方法采取以下的技术方案来实现；具体包括两部分内容，即，形成机载PHM算法处理策略和PHM算子库。FIG. 1 is a schematic schematic diagram of an airborne deployment-oriented predictive modeling method for health management provided by an embodiment of the present invention. The airborne deployment-oriented health management predictive modeling method provided by the embodiments of the present invention adopts the following technical solutions; it specifically includes two parts, namely, forming an airborne PHM algorithm processing strategy and a PHM operator library.
第一部分：形成机载PHM算法处理策略：The first part: forming the airborne PHM algorithm processing strategy:
该部分的实施方式包括：Implementations of this section include:
步骤1，在地面平台对机载数据和地面数据进行数据处理及模型训练，得到多个训练模型；Step 1: Data processing and model training are performed on the airborne data and ground data on the ground platform to obtain multiple training models;
步骤2，在原理样机中对步骤1中得到的训练模型进行评估选择和决策，得到符合任务需求的训练模型。In step 2, the training model obtained in step 1 is evaluated, selected and decided in the principle prototype to obtain a training model that meets the requirements of the task.
步骤3，将步骤2得到的符合任务需求的训练模型部署于机载中。In step 3, the training model obtained in step 2 that meets the mission requirements is deployed in the aircraft.
上述第一部分具体包括以下步骤：运行于地面平台的数据导入阶段、数据处理阶段和模型训练阶段；以及运行于原理样机中的模型评估阶段和模型决策阶段，以及运行于机载中的模型部署阶段；如图1所示上述各阶段的运行。The above-mentioned first part specifically includes the following steps: the data import stage, data processing stage and model training stage running on the ground platform; as well as the model evaluation stage and model decision stage running on the principle prototype, and the model deployment stage running on the airborne ; Figure 1 shows the operation of the above-mentioned stages.
对于上述步骤1中的上述各阶段，即数据导入阶段、数据处理阶段和模型训练阶段均运行于地面高性能计算平台(本文中可简称为：地面平台)，该步骤1中，通过针对机载运行历史数据、实验台架产生的试验数据、仿真模型平台产生的仿真数据、维护维修数据等海量异构数据，利用地面平台丰富的数据分析工具包，借助地面平台的高性能计算机，实现数据的深入挖掘分析；针对具体PHM任务(如故障诊断、寿命预测等)的模型训练，为机上的具体PHM任务提供训练好的模型，为实现机上的PHM任务提供基础。以下分别对上述步骤1中各阶段的具体实施方式进行说明。For the above-mentioned stages in the above step 1, that is, the data import stage, the data processing stage and the model training stage, all run on the ground high-performance computing platform (herein referred to as: ground platform), in this step 1, by targeting the airborne Operational historical data, test data generated by experimental bench, simulation data generated by simulation model platform, maintenance and repair data and other massive heterogeneous data, using the rich data analysis toolkit of the ground platform, with the help of the high-performance computer of the ground platform, realize the data analysis. In-depth mining and analysis; model training for specific PHM tasks (such as fault diagnosis, life prediction, etc.) Specific implementations of each stage in the above step 1 will be described below.
步骤1.1：数据导入阶段。Step 1.1: Data Import Phase.
该数据导入阶段运行于地面高性能平台，可以包含各类结构化数据(如txt，csv等格式的数据文件)及非结构化数据(如表格，图像等文件)的导入。The data import stage runs on the ground high-performance platform, which can include the import of various structured data (such as data files in txt, csv and other formats) and unstructured data (such as tables, images and other files).
步骤1.2：数据处理阶段。Step 1.2: Data processing stage.
该数据处理阶段运行于地面高性能平台，包含:数据探索分析子阶段、数据预处理子阶段和特征工程子阶段。The data processing stage runs on the ground high-performance platform, including: data exploration and analysis sub-stage, data preprocessing sub-stage and feature engineering sub-stage.
数据探索分析子阶段可以通过许多成熟算法实现，如数据的相关性分析可以通过计算数据两两特征之间的皮尔逊相关系数、肯德尔相关系数等实现，同时通过画散点对图、相关性热区图等，也可以直观展示数据两两特征之间的线性相关性；趋势性分析可以通过绘制每个特征的折线图来直观展示。数据的分布情况初探可以通过计算数据的各类统计特征(均值、方差、众数，峰度、偏度、均方根等)实现，也可绘制数据箱线图，频率直方图等直观显示数据分布情况。The sub-stage of data exploration and analysis can be realized by many mature algorithms. For example, the correlation analysis of data can be realized by calculating the Pearson correlation coefficient and Kendall correlation coefficient between the two features of the data. Heat map, etc., can also visually display the linear correlation between two features of the data; trend analysis can be visually displayed by drawing a line chart of each feature. Preliminary exploration of data distribution can be achieved by calculating various statistical characteristics of data (mean, variance, mode, kurtosis, skewness, root mean square, etc.), and can also draw data box plots, frequency histograms, etc. Distribution.
数据预处理子阶段可以实现对数据质量的提升，包括：数据清洗、数据去噪、数据标准化等。其中，数据清洗主要从以下三个方面进行：一致性检查，无效值和缺失值的处理、重复值处理。对于无效值和缺失值，根据不同情况采取三种处理方法：当缺失值少且属性重要程度低时，若该属性为数值型数据，则根据数据的分布情况采用均值、中位数等进行简单填充；若缺失率高且属性重要程度低，则可直接删除该属性；若缺失率高且属性重要程度高，则采用插补法与建模法。对于重复值的判断，先将数据集中的记录按一定规则排序，然后通过比较邻近记录看是否相似，以此检测记录是否重复。数据标准化常用方法为min-max归一化(将不同特征的数据归一化为[0,1]区间内)和z-score归一化(将不同特征数据均归一化为服从标准正态分布的数据)。The data preprocessing sub-stage can improve the data quality, including: data cleaning, data denoising, data standardization, etc. Among them, data cleaning is mainly carried out from the following three aspects: consistency check, processing of invalid and missing values, and processing of duplicate values. For invalid values and missing values, three processing methods are adopted according to different situations: when there are few missing values and the importance of the attribute is low, if the attribute is numerical data, the mean, median, etc. are used for simple processing according to the distribution of the data. Fill; if the missing rate is high and the attribute importance is low, the attribute can be deleted directly; if the missing rate is high and the attribute importance is high, imputation and modeling methods are used. For the judgment of duplicate values, first sort the records in the data set according to certain rules, and then compare the adjacent records to see if they are similar, so as to detect whether the records are duplicates. The common methods of data normalization are min-max normalization (normalize the data of different features into the [0,1] interval) and z-score normalization (normalize the data of different features to obey the standard normality) distributed data).
特征工程子阶段可以采用许多算法，具体分为基于经验及物理模型的方法和基于数据驱动的方法。其中，基于经验及物理模型的方法是针对特定对象，利用已有的经验知识，或者为其建立对应的物理模型，从而选择或构建得到对目标任务最有用的特征；基于数据驱动的方法仅从数据的角度出发，通过分析不同特征数据的趋势性、与目标标签的关联性等选择或构建对目标最有意义的特征，具体的特征选择方法有：基于方差的方法，relief，fisher-score，基于稀疏学习的方法等，特征提取方法包括主成分分析PCA，线性判别分析LDA，独立成分分析ICA，局部线性嵌入LLE，等距映射Isomap等。The feature engineering sub-phase can use many algorithms, which can be divided into empirical and physical model-based methods and data-driven methods. Among them, the methods based on experience and physical models are for specific objects, using existing empirical knowledge, or establishing corresponding physical models for them, so as to select or construct the most useful features for the target task; data-driven methods only From the data point of view, select or construct the most meaningful features for the target by analyzing the trend of different feature data and the correlation with the target label. The specific feature selection methods are: variance-based method, relief, fisher-score, Based on sparse learning methods, etc., feature extraction methods include principal component analysis PCA, linear discriminant analysis LDA, independent component analysis ICA, local linear embedding LLE, isometric mapping Isomap, etc.
步骤1.3：模型训练阶段。Step 1.3: Model training phase.
该模型训练阶段运行于地面高性能平台，包含：通过基于物理或经验模型的模型训练和基于数据驱动的模型训练，训练得到针对特定的PHM任务的多个算法模型。The model training phase runs on the ground high-performance platform, including: through model training based on physical or empirical models and data-driven model training, training to obtain multiple algorithm models for specific PHM tasks.
该步骤中的PHM任务例如包括：状态监测、故障诊断(识别是否存在故障、识别故障类型、识别故障部位)、故障预测(预测何时会出现故障)、剩余寿命预测(预测部件退化趋势、预测部件何时失效)等。The PHM tasks in this step include, for example: condition monitoring, fault diagnosis (identifying whether there is a fault, identifying the type of fault, identifying the fault location), fault prediction (predicting when a fault will occur), remaining life prediction (predicting component degradation trends, predicting when components fail), etc.
针对具体的PHM任务类型，对其进行问题分析及解构的方式可以为：首先根据对分析对象的认知，如对象存在容易求解的物理或经验模型，则针对具体任务，对该对象建立对应的物理模型，以进行模型训练和求解；若对象结构复杂，失效或退化机理难以得到，则采用数据驱动的方法，利用相应机器学习，统计分析等方法选择算法并进行模型训练。For a specific type of PHM task, the method of problem analysis and deconstruction can be as follows: First, according to the cognition of the object to be analyzed, if the object has a physical or empirical model that is easy to solve, then for the specific task, establish the corresponding object for the object. The physical model is used for model training and solution; if the structure of the object is complex, and the failure or degradation mechanism is difficult to obtain, a data-driven method is adopted, and corresponding machine learning, statistical analysis and other methods are used to select algorithms and conduct model training.
需要说明的是，本发明实施例在步骤1.3的的模型训练阶段，可以采取多种算法，进行多个模型的训练，得到多个训练好的模型。It should be noted that, in the embodiment of the present invention, in the model training stage of step 1.3, multiple algorithms may be adopted to train multiple models, and multiple trained models may be obtained.
对于上述步骤2中的上述各阶段，即模型评估阶段和模型决策阶段运行于与机载软硬件环境相似的原理样机中。完成步骤1的模型训练阶段后，得到了几种训练好的模型，需要对这几种模型进行决策以得到最终部署于机载的“最优模型”。由于“最优模型”的概念与实际应用密切相关，以机载部署为目的，因此对模型进行评估需要指定模型运行的硬件环境，构建与机载软/硬件环境相同的原理样机模拟机载运行环境。以下分别对上述步骤2中各阶段的具体实施方式进行说明。For the above-mentioned stages in the above-mentioned step 2, that is, the model evaluation stage and the model decision-making stage, operate in a principle prototype similar to the airborne software and hardware environment. After completing the model training phase in step 1, several trained models are obtained, and decisions need to be made on these models to obtain the "optimal model" that is finally deployed on the airborne. Since the concept of "optimal model" is closely related to practical applications and is aimed at airborne deployment, the evaluation of the model requires specifying the hardware environment in which the model runs, and constructing the same principle prototype as the airborne software/hardware environment to simulate the airborne operation. surroundings. Specific implementations of each stage in the above step 2 will be described below.
步骤2.1：模型评估阶段。Step 2.1: Model evaluation phase.
该模型评估运行于原理样机，对于不同的任务类型，可采用多种反映模型预测准确度的模型评估指标进行模型评估。具体实施方式为：对于回归问题模型，常用的模型评估指标有均方根误差RMSE，平均绝对误差MAE，拟合优度R2等；对于分类问题模型，常用的模型评估指标有准确率acc，查准率，召回率，F1分数，ROC曲线等。除反映模型预测准确度的指标外，还有反应模型运行效率的指标：模型预测时间，该指标对于强实时任务非常必要。在原理样机中进行多个训练好的模型的评估，并得到模型评估表，如图2所示，为本发明实施例中模型评估阶段与模型决策阶段的原理示意图。The model evaluation runs on the principle prototype. For different task types, a variety of model evaluation indicators that reflect the model prediction accuracy can be used for model evaluation. The specific implementation is: for regression problem models, commonly used model evaluation indicators include root mean square error RMSE, mean absolute error MAE, goodness of fit R2, etc.; for classification problem models, commonly used model evaluation indicators include accuracy rate acc, check Accuracy, recall, F1 score, ROC curve, etc. In addition to the indicators reflecting the prediction accuracy of the model, there is also an indicator reflecting the operating efficiency of the model: the model prediction time, which is very necessary for strong real-time tasks. A plurality of trained models are evaluated in the principle prototype, and a model evaluation table is obtained, as shown in FIG. 2 , which is a schematic diagram of the model evaluation stage and the model decision stage in the embodiment of the present invention.
步骤2.1：模型决策阶段。Step 2.1: Model decision stage.
该模型决策阶段运行于原理样机。首先根据具体的任务要求制定模型决策规则，并结合得到的模型评估表，对模型进行最终决策，选择出对于该具体任务的“最优模型”，即符合任务需求的训练模型。The model decision phase runs on a prototype. First, the model decision rules are formulated according to the specific task requirements, and combined with the obtained model evaluation table, the final decision is made on the model, and the "optimal model" for the specific task is selected, that is, the training model that meets the task requirements.
模型评估阶段与模型决策阶段的过程如图2所示。The process of the model evaluation stage and the model decision stage is shown in Figure 2.
本发明实施例的步骤3具体为模型部署阶段。Step 3 in the embodiment of the present invention is specifically a model deployment stage.
在该模型部署阶段，将步骤2中所选出的“最优模型”采用机载硬件支持的语言进行封装，并将封装后的“最优模型”部署于机载端。In the model deployment stage, the "optimal model" selected in step 2 is packaged in a language supported by the airborne hardware, and the packaged "optimal model" is deployed on the airborne end.
通过上述步骤1～3，得到第一部分中机载的全周期PHM建模策略。Through the above steps 1 to 3, the airborne full-cycle PHM modeling strategy in the first part is obtained.
第二部分：构建机载PHM算子库；The second part: build the airborne PHM operator library;
根据上述第一部分中机载PHM算法处理策略的各阶段和各子阶段，梳理相应内容得到PHM算子库，如图3所示，为本发明实施例中机载PHM算子库的示意图。According to each stage and each sub-stage of the airborne PHM algorithm processing strategy in the first part, the corresponding content is sorted out to obtain a PHM operator library, as shown in FIG. 3 , which is a schematic diagram of the airborne PHM operator library in the embodiment of the present invention.
本发明实施例中机载PHM算子库按照数据处理流程分类，可以包含：流程模块和支撑性模块。In the embodiment of the present invention, the airborne PHM operator library is classified according to the data processing flow, and may include: a flow module and a supporting module.
本发明实施例中机载PHM算子库按照算子功能分类，可以包括：数据处理通用算子、PHM任务专用算子和集成化模块化的飞机部件级/系统级/全机级PHM算子。In the embodiment of the present invention, the airborne PHM operator library is classified according to operator functions, and may include: general data processing operators, PHM task-specific operators, and integrated and modular aircraft component-level/system-level/full-aircraft-level PHM operators .
如图3所示，机载PHM算子库中含有：数据基本操作算子单元、机器学习算子单元、专家系统算子单元和超参数优化算子单元四个支撑性模块。As shown in Figure 3, the airborne PHM operator library contains four supporting modules: the basic data operation operator unit, the machine learning operator unit, the expert system operator unit and the hyperparameter optimization operator unit.
数据基本操作算子单元作为支撑整个流程的基础算子。包含逻辑运算、按元素运算、选择/替换行/列等。The data basic operation operator unit serves as the basic operator supporting the entire process. Contains logical operations, element-wise operations, select/replace rows/columns, etc.
机器学习算子单元、专家系统算子单元和超参数优化算子单元作为支撑模型训练算子单元的支撑性算子。机器学习算子单元为后续的故障诊断和寿命预测提供多样的可供灵活选择的分类，回归和人工神经网络算子。专家系统算子单元包含针对特定具体部件的经验解决方案。超参数优化是为模型训练中的超参数提供的参数调优方法，常见模型超参数有神经网络中的网络层数、节点个数；决策树模型中的最大树深，正则化系数等；超参数优化方法包含粒子群优化、遗传算法、模拟退火算法等。The machine learning operator unit, the expert system operator unit and the hyperparameter optimization operator unit are used as supporting operators to support the model training operator unit. The machine learning operator unit provides a variety of flexibly selected classification, regression and artificial neural network operators for subsequent fault diagnosis and life prediction. Expert system operator units contain empirical solutions for specific specific components. Hyperparameter optimization is a parameter tuning method provided for hyperparameters in model training. Common model hyperparameters include the number of network layers and nodes in the neural network; the maximum tree depth in the decision tree model, the regularization coefficient, etc.; Parameter optimization methods include particle swarm optimization, genetic algorithm, simulated annealing algorithm, etc.
按照机载PHM算法处理方法策略，按照算子库中流程模块的顺序，在算子库中选择特定的数据处理通用算子和PHM专用算子，搭建出针对特定部件或特定系统的全流程状态监测、故障诊断和寿命预测过程，作为该部件或系统的集成化模块化算子。According to the processing method and strategy of the airborne PHM algorithm, according to the sequence of the process modules in the operator library, select the specific data processing general operator and PHM special operator in the operator library, and build the whole process state for specific components or specific systems. Monitoring, fault diagnosis and life prediction process as an integrated modular operator for the component or system.
本发明实施例提供的技术方案，针对航空预测性维护，解决航空机载和地面完整的故障预测与健康管理(简称PHM)预测性维护规范不统一的问题，包含地面平台、原理样机、机载部署的全周期PHM算法策略。形成支持数据分析与建模的PHM算子库，包括数据导入阶段、数据处理阶段、模型训练阶段、模型评估阶段、模型决策阶段及模型部署阶段，并实现为具体的PHM算子库。该PHM算子库完整支撑机上PHM建模和数据分析过程，同时兼顾灵活性与集成模块化。经理论分析和试验，在本发明专利下开发的算子库框架能够满足航空PHM需求，基于算子库框架可实现飞机部件级/系统级/全机级成员面向机载部署的PHM处理。The technical solutions provided by the embodiments of the present invention are aimed at aviation predictive maintenance, and solve the problem that the aviation airborne and ground complete fault prediction and health management (referred to as PHM) predictive maintenance specifications are not unified, including ground platform, principle prototype, airborne Deployed full-cycle PHM algorithm strategy. A PHM operator library supporting data analysis and modeling is formed, including the data import stage, data processing stage, model training stage, model evaluation stage, model decision stage and model deployment stage, and is implemented as a specific PHM operator library. The PHM operator library fully supports the on-board PHM modeling and data analysis process, while taking into account flexibility and integrated modularity. After theoretical analysis and tests, the operator library framework developed under the patent of the present invention can meet the needs of aviation PHM, and based on the operator library framework, the PHM processing for aircraft component-level/system-level/full-aircraft-level members can be implemented for airborne deployment.
本发明实施例提供的面向机载部署的健康管理预测性建模方法，一方面，通过借助地面平台的高性能计算机，以及与机载环境相同的原理样机，实现利用海量异构数据的分析，问题建模以及模型决策，最终选择面向具体任务的“最优模型”进行模型部署，实现面向机载部署的全周期PHM建模策略；另一方面，通过将面向机载部署的全周期PHM建模策略具象化为具体的算子，提出PHM算子库，完整支撑面向机载的PHM建模过程。The airborne deployment-oriented predictive modeling method for health management provided by the embodiments of the present invention, on the one hand, realizes the analysis using massive heterogeneous data by using the high-performance computer of the ground platform and the same principle prototype as the airborne environment. Problem modeling and model decision-making, and finally select the “optimal model” for specific tasks for model deployment, and realize the full-cycle PHM modeling strategy for airborne deployment; The modulo strategy is visualized as a specific operator, and a PHM operator library is proposed to fully support the airborne-oriented PHM modeling process.
以下通过一个具体实施例对本发明实施例提供的面向机载部署的健康管理预测性建模方法的实施方式进行示意性说明。An implementation manner of the airborne deployment-oriented predictive modeling method for health management provided by the embodiment of the present invention is schematically described below through a specific embodiment.
针对飞机的特定部件，如发动机，滑油模块，旋转部件等，使用集成化模块化的飞机部件级算子，即可快速便捷对飞机部件的健康状态进行评估。For specific parts of the aircraft, such as engines, lubricating oil modules, rotating parts, etc., the integrated and modular aircraft part-level operators can be used to quickly and easily evaluate the health status of aircraft parts.
针对机上某类不常出现的数据及故障，按照本发明实施例所提出的机载PHM算法处理策略，支持自行使用算子库中丰富的数据处理通用算子及PHM任务专用算子，根据算子库中流程模块的顺序，对数据进行探索分析，灵活搭建出针对该数据的诊断或预测全流程。具体步骤如下：For certain types of data and faults that occur infrequently on the aircraft, according to the processing strategy of the airborne PHM algorithm proposed in the embodiment of the present invention, it is possible to use the rich data processing general operators and PHM task-specific operators in the operator library by itself. The sequence of the process modules in the sub-library is used to explore and analyze the data, and flexibly build the whole process of diagnosis or prediction for the data. Specific steps are as follows:
首先，在地面高性能平台中进行数据导入，数据处理以及模型训练。以windows平台采用python语言为例。在地面平台中进行的数据分析和模型训练均调用PHM算子库中的算子。将试验台采集到的数值型数据导出为txt、excel、csv等文件形式，并将采集到的文件数据导入至python环境，使用数据导入算子，将其转化为dataframe格式的数据。First, data import, data processing, and model training are performed in a ground-based high-performance platform. Take the python language as an example on the windows platform. Data analysis and model training in the ground platform call the operators in the PHM operator library. Export the numerical data collected by the test bench into txt, excel, csv and other file formats, import the collected file data into the python environment, and use the data import operator to convert it into dataframe format data.
对原始数据进行数据清洗。使用PHM算子库中数据预处理部分的数据清洗算子，检查是否存在缺失数据、重复数据、并且检查是否存在数据不一致的问题，并且对这些数据问题按照清洗算法进行处理，提升数据质量。使用PHM算子库中数据探索分析的算子，对质量提升后的数据进行初步探索，获知数据规模、分布特征、相关性等数据特性，为之后的步骤提供思路和指导。数据探索分析部分的算子包含：相关性热力图、折线图、箱线图、数据统计特征等。Data cleaning is performed on raw data. Use the data cleaning operator in the data preprocessing part of the PHM operator library to check for missing data, duplicate data, and data inconsistency, and process these data problems according to the cleaning algorithm to improve data quality. Use the data exploration and analysis operators in the PHM operator library to conduct preliminary exploration of the improved data, learn data characteristics such as data scale, distribution characteristics, and correlation, and provide ideas and guidance for subsequent steps. The operators in the data exploration and analysis part include: correlation heat map, line chart, boxplot, data statistical features, etc.
通过数据探索分析步骤之后，初步明确之后的分析建模思路。首先以任务为导向，确定建立怎样的模型。根据任务的实际情况，可分为回归问题、分类问题、聚类问题、预测问题等，每种问题需要建立不同的模型。回归与分类属于监督学习的范畴，是指训练集数据中待求特征(样本标签)已知，其中回归问题的待求特征是连续值，而分类问题的待求特征是离散值。聚类属于无监督学习的范畴，即在训练集数据中待求特征(样本标签)未知，聚类模型常用于异常检测等问题。而预测问题指给定一段时间的特征取值趋势，预测未来时刻该特征的取值，或根据一段时间内某些特征的取值，预测该设备的失效时刻。After the data exploration and analysis steps, the subsequent analysis and modeling ideas are initially clarified. First, be task-oriented and determine what kind of model to build. According to the actual situation of the task, it can be divided into regression problems, classification problems, clustering problems, prediction problems, etc. Each problem needs to establish a different model. Regression and classification belong to the category of supervised learning, which means that the features to be found (sample labels) in the training set data are known, and the features to be found in regression problems are continuous values, while the features to be found in classification problems are discrete values. Clustering belongs to the category of unsupervised learning, that is, the features (sample labels) to be found in the training set data are unknown, and the clustering model is often used in problems such as anomaly detection. The prediction problem refers to the trend of feature values for a given period of time, predicting the value of the feature in the future, or predicting the failure time of the device according to the value of some features in a period of time.
结合数据探索分析得出的结论，对数据进行预处理和特征工程操作。如若在实际情况中模型工况未知，则在预处理步骤使用算子库中的数据合并算子，将所有输入的工况数据文件进行合并，并且工况不作为模型输入。若选择的算法模型对不同输入特征的范围差别敏感，则使用预处理算子库中的数据标准化算子，将数据特征标准化为服从标准正态分布的数据。在特征工程步骤，需要从原始数据中选择或构建出模型的输入特征。如若在探索分析步骤中发现某个特征的方差很小，几乎没有变化趋势，则删除该特征。Combined with the conclusions drawn from the data exploration analysis, data preprocessing and feature engineering operations are performed. If the model conditions are unknown in the actual situation, the data merging operator in the operator library is used in the preprocessing step to merge all the input conditions data files, and the operation conditions are not used as the model input. If the selected algorithm model is sensitive to the range difference of different input features, the data standardization operator in the preprocessing operator library is used to standardize the data features into data that obey the standard normal distribution. In the feature engineering step, the input features of the model need to be selected or constructed from the raw data. If it is found in the exploratory analysis step that the variance of a feature is small and there is almost no trend of change, the feature is deleted.
在模型训练阶段，选择合适的算法进行模型的训练，该部分的算子库中包含：线性回归、决策树、支持向量机、梯度提升树、随机森林、人工神经网络等算法，支持构建各类回归、分类、预测模型。根据数据探索分析中提供的思路，选择几种合适的算法构建模型；将全部样本按照合适比例划分训练集和测试集，并将训练集样本送入构建出的几种模型进行模型训练，分别得到几个特定的训练好的模型。In the model training stage, select an appropriate algorithm for model training. The operator library in this part includes: linear regression, decision tree, support vector machine, gradient boosting tree, random forest, artificial neural network and other algorithms, supporting the construction of various Regression, classification, prediction models. According to the ideas provided in the data exploration and analysis, select several suitable algorithms to build the model; divide all the samples into the training set and the test set according to the appropriate ratio, and send the training set samples into the constructed models for model training, and get Several specific trained models.
至此，在地面平台使用高性能服务器进行的数据分析和模型训练阶段结束。由于机载计算资源有限，因此考虑仅在机上进行模型的预测部分：将在地面训练好的模型使用C语言进行轻量化封装，并部署在机上的嵌入式平台中，进行机上的实时诊断及预测。为模拟机载嵌入式环境，选择合适的嵌入式开发板或自己构建合适的开发板作为原理样机，首先将训练好的几种模型使用C语言进行轻量化的封装，并将得到的几种训练好的模型分别烧入原理样机，输入测试集数据，在板中进行模型的预测，该步骤作为机上预测的环境模拟，以实现算法的验证环节。At this point, the data analysis and model training phase using high-performance servers on the ground platform is over. Due to the limited computing resources onboard, only the prediction part of the model is considered to be carried out on the machine: the model trained on the ground will be lightweight packaged in C language, and deployed in the embedded platform on the machine for real-time diagnosis and prediction on the machine . In order to simulate the airborne embedded environment, choose a suitable embedded development board or build a suitable development board by yourself as the principle prototype. First, use C language to carry out lightweight packaging of the trained models, and combine the obtained training models. The good models are burned into the principle prototype respectively, the test set data is input, and the model prediction is carried out in the board. This step is used as the environment simulation of the prediction on the machine to realize the verification of the algorithm.
对板中几种模型预测得到的结果使用各类评估指标进行评估。常用的回归模型准确度评估指标有：RMSE,MAE,R2；常用的分类模型准确度评估指标有：查准率、查全率、正确率、F1值等。除模型准确度评估指标外，还需要模型训练时间、模型测试时间等指标，用以评估模型的预测实时性。将测试集数据分别输入原理样机中的训练模型中，在板中进行预测，并计算几种模型评估指标的数值，得到图2所示的模型评估表。该评估表作为最佳模型的选择和机载部署的参考，根据实际情况的要求，选出最符合实际要求的最优模型，即可将该模型部署于机上。The results predicted by several models in the board are evaluated using various evaluation indicators. Commonly used regression model accuracy evaluation indicators are: RMSE, MAE, R2; commonly used classification model accuracy evaluation indicators are: precision rate, recall rate, correct rate, F1 value, etc. In addition to the model accuracy evaluation indicators, indicators such as model training time and model testing time are also required to evaluate the real-time prediction of the model. Input the test set data into the training model in the principle prototype respectively, make predictions in the board, and calculate the values of several model evaluation indicators to obtain the model evaluation table shown in Figure 2. The evaluation table is used as a reference for the selection of the best model and the deployment of the aircraft. According to the requirements of the actual situation, the optimal model that best meets the actual requirements can be selected, and then the model can be deployed on the aircraft.
虽然本发明所揭露的实施方式如上，但所述的内容仅为便于理解本发明而采用的实施方式，并非用以限定本发明。任何本发明所属领域内的技术人员，在不脱离本发明所揭露的精神和范围的前提下，可以在实施的形式及细节上进行任何的修改与变化，但本发明的专利保护范围，仍须以所附的权利要求书所界定的范围为准。Although the embodiments disclosed in the present invention are as above, the described contents are only the embodiments adopted to facilitate the understanding of the present invention, and are not intended to limit the present invention. Any person skilled in the art to which the present invention belongs, without departing from the spirit and scope disclosed by the present invention, can make any modifications and changes in the form and details of the implementation, but the scope of the patent protection of the present invention still needs to be The scope defined by the appended claims shall prevail.
Claims (9)
Translated fromChinesePriority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111663607.4ACN114417501B (en) | 2021-12-30 | 2021-12-30 | A predictive modeling approach to health management for airborne deployment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111663607.4ACN114417501B (en) | 2021-12-30 | 2021-12-30 | A predictive modeling approach to health management for airborne deployment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114417501Atrue CN114417501A (en) | 2022-04-29 |
| CN114417501B CN114417501B (en) | 2025-06-17 |
Family
ID=81271043
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111663607.4AActiveCN114417501B (en) | 2021-12-30 | 2021-12-30 | A predictive modeling approach to health management for airborne deployment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114417501B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115599002A (en)* | 2022-12-16 | 2023-01-13 | 中国航空工业集团公司西安飞机设计研究所（Cn） | Method and device for monitoring state life of airborne PHM system |
| CN117454232A (en)* | 2023-12-22 | 2024-01-26 | 山东未来集团有限公司 | Production network construction fault diagnosis, prediction and health management system and method |
| CN119129230A (en)* | 2024-08-27 | 2024-12-13 | 北京航空航天大学 | PHM system architecture design method, electronic device and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030120402A1 (en)* | 2001-09-08 | 2003-06-26 | Jaw Link C. | Intelligent condition-based engine/equipment management system |
| CN106598033A (en)* | 2016-12-31 | 2017-04-26 | 清华大学深圳研究生院 | PHM (Prognostic and Health Management) system for unmanned aerial vehicle based on middleware |
| CN111896244A (en)* | 2020-07-29 | 2020-11-06 | 北京天地龙跃科技有限公司 | Fully mechanized mining equipment PHM system |
| CN112464569A (en)* | 2020-12-10 | 2021-03-09 | 北京明略软件系统有限公司 | Machine learning method and system |
- 2021
- 2021-12-30CNCN202111663607.4Apatent/CN114417501B/enactiveActive
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030120402A1 (en)* | 2001-09-08 | 2003-06-26 | Jaw Link C. | Intelligent condition-based engine/equipment management system |
| CN106598033A (en)* | 2016-12-31 | 2017-04-26 | 清华大学深圳研究生院 | PHM (Prognostic and Health Management) system for unmanned aerial vehicle based on middleware |
| CN111896244A (en)* | 2020-07-29 | 2020-11-06 | 北京天地龙跃科技有限公司 | Fully mechanized mining equipment PHM system |
| CN112464569A (en)* | 2020-12-10 | 2021-03-09 | 北京明略软件系统有限公司 | Machine learning method and system |
Non-Patent Citations (1)
| Title |
|---|
| 张莉;袁海文;: "飞机电源多Agent预测与健康管理系统研究", 计算机工程与应用, no. 19, 1 July 2010 (2010-07-01)* |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115599002A (en)* | 2022-12-16 | 2023-01-13 | 中国航空工业集团公司西安飞机设计研究所（Cn） | Method and device for monitoring state life of airborne PHM system |
| CN117454232A (en)* | 2023-12-22 | 2024-01-26 | 山东未来集团有限公司 | Production network construction fault diagnosis, prediction and health management system and method |
| CN119129230A (en)* | 2024-08-27 | 2024-12-13 | 北京航空航天大学 | PHM system architecture design method, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114417501B (en) | 2025-06-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220027762A1 (en) | Predictive maintenance model design system | |
| Rose et al. | Application of structural topic modeling to aviation safety data | |
| Helgo | Deep learning and machine learning algorithms for enhanced aircraft maintenance and flight data analysis | |
| CN114417501A (en) | Airborne deployment-oriented health management predictive modeling method | |
| CN114282579A (en) | A Fault Diagnosis Method for Aerospace Bearings Based on Variational Mode Decomposition and Residual Networks | |
| Bisanti et al. | Digital twins for aircraft maintenance and operation: A systematic literature review and an IoT-enabled modular architecture | |
| Karaoğlu et al. | Applications of machine learning in aircraft maintenance | |
| De Luca et al. | A deep attention based approach for predictive maintenance applications in IoT scenarios | |
| US11807389B2 (en) | Systems and methods for aircraft management | |
| US20250292104A1 (en) | Automated feature generation for sensor subset selection | |
| Luo et al. | Big data analytics–enabled cyber-physical system: model and applications | |
| CN111324635A (en) | Industrial big data cloud platform data processing method and system | |
| JP2022042989A (en) | Automatic prediction of repairs based on sensor data | |
| Azyus et al. | Prediction of remaining useful life using the CNN-GRU network: A study on maintenance management | |
| Li et al. | Civil aircraft health management research based on big data and deep learning technologies | |
| Jose et al. | Solving time alignment issue of multimodal data for accurate prognostics with CNN-Transformer-LSTM network | |
| Giannoulidis et al. | A context-aware unsupervised predictive maintenance solution for fleet management | |
| CN114312930B (en) | Train operation abnormality diagnosis method and device based on log data | |
| Wang et al. | Software defect fault intelligent location and identification method based on data mining | |
| Bond et al. | A hybrid learning approach to prognostics and health management applied to military ground vehicles using time-series and maintenance event data | |
| Stanton et al. | Data augmentation for predictive maintenance: Synthesising aircraft landing gear datasets | |
| Lin et al. | Design and implementation of a CPS‐based predictive maintenance and automated management platform | |
| CN111965995B (en) | Three-in-one aircraft avionics system health assessment and prediction system and method thereof | |
| Soni et al. | Predictive maintenance of gas turbine using prognosis approach | |
| Lacaille | A maturation environment to develop and manage health monitoring algorithms |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |