WO2022147853A1 - 一种基于混合预测模型的复杂装备电源组故障预测方法 - Google Patents

一种基于混合预测模型的复杂装备电源组故障预测方法 Download PDF

Info

Publication number
WO2022147853A1
WO2022147853A1 PCT/CN2021/072780 CN2021072780W WO2022147853A1 WO 2022147853 A1 WO2022147853 A1 WO 2022147853A1 CN 2021072780 W CN2021072780 W CN 2021072780W WO 2022147853 A1 WO2022147853 A1 WO 2022147853A1
Authority
WO
WIPO (PCT)
Prior art keywords
model
prediction
time series
predict
core
Prior art date
Application number
PCT/CN2021/072780
Other languages
English (en)
French (fr)
Inventor
孙希明
王嫒娜
李英顺
仲崇权
Original Assignee
大连理工大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 大连理工大学 filed Critical 大连理工大学
Priority to US17/311,931 priority Critical patent/US20220341996A1/en
Publication of WO2022147853A1 publication Critical patent/WO2022147853A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/392—Determining battery ageing or deterioration, e.g. state of health
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Definitions

  • the invention belongs to the technical field of fault prediction of complex equipment, and relates to a method for predicting power group faults of complex equipment based on a hybrid prediction model, in particular to a hybrid prediction method based on ARIMA combined with ANN for operating data of complex equipment under image stabilization conditions Model-based power pack failure prediction method for complex equipment.
  • the present invention provides a fault prediction method for a power supply group of complex equipment based on a hybrid prediction model.
  • a high-precision hybrid prediction model based on ARIMA combined with ANN is proposed for monitoring the power supply of complex equipment.
  • the future development trend of the key parameters of the power group is based on the hybrid prediction model, and the core attribute parallel parameters are used to predict the operation state of the power group, which provides a basis for the comprehensive monitoring of the future operation state of the complex equipment.
  • the future state monitoring process of the power pack of complex equipment is established to guide the implementation of the prediction of the operation state of complex equipment.
  • a fault prediction method for a power group of complex equipment based on a hybrid prediction model the prediction method is aimed at operating data of complex equipment under image stabilization conditions;
  • the fault prediction method performs fault prediction based on a hybrid fault prediction model
  • the hybrid fault prediction model includes an ARIMA prediction model combined with an ANN prediction model;
  • the ARIMA prediction model is used to predict the time series of the power supply group with a linear variation law
  • the ANN prediction model is used to predict the time series of the power supply group with a nonlinear variation law
  • the hybrid prediction model is used to integrate the prediction results of the time series of the power supply group, and use the core attribute parallel parameters to perform state monitoring;
  • Complex equipment includes power pack, CPU board, KZB board, I/O board, ADA board, angular velocity sensor, cross wind sensor and tilt sensor;
  • the minimum disjunctive normal form is used to perform attribute reduction to obtain the core attribute set
  • ARIMA model identification draw the autocorrelation diagram and partial autocorrelation diagram of the stationary time series, and obtain the perceptual knowledge of the autoregressive order n and the moving average order m of the ARIMA model according to the autocorrelation diagram and the partial autocorrelation diagram ;
  • Akaike Information Criterion and the Bayesian Information Criterion to calculate the model order (n, m);
  • Evaluation indicators include: mean absolute error, mean square error, mean absolute percentage error.
  • the mean absolute error is the average of the absolute values of the deviations of all individual observations and the arithmetic mean, which can avoid the problem of mutual cancellation of errors, so it can accurately reflect the size of the actual prediction error;
  • the mean square error is the mathematical expectation of the square of the difference of an estimator of the overall parameter determined according to the sub-sample, which reflects a measure of the degree of difference between the estimator and the estimated quantity.
  • the standard error can also be obtained, which is also used to measure the observation. the deviation between the value and the true value;
  • the mean absolute percentage error refers to the percentage value that the predicted result deviates from the actual result on average. It is a percentage value, so it is easier to understand than other statistics;
  • S5 Use the state monitoring method of the core attribute parallel parameter monitoring combined with the upper and lower limit early warning to monitor the running state of the power supply group, and obtain the state monitoring result;
  • the above uses the ANN model and the ARIMA-ANN model to predict the core attribute parallel parameters, including:
  • ARIMA model is not added to compare the prediction results of the core attribute parallel parameters, because ARIMA predicts the linear change law of time series, and is only suitable for the prediction of single-parameter time series;
  • the beneficial effects of the invention are: by extracting the characteristic parameters of the power group of complex equipment, a core attribute set capable of expressing the attributes of the power group is obtained, and the time series of the core attribute set is regarded as two parts: linear change law and nonlinear change law
  • the time series is composed together; the ARIMA model is used to obtain the time series prediction results with a linear change law, and the residual e(t) including the nonlinear change law; the residual e(t) of the time series including the nonlinear change law is obtained by using the ANN model.
  • the linear part and the nonlinear part are synthesized to obtain the final prediction results, and based on the obtained hybrid fault prediction model, the core attribute parallel parameter monitoring is used to combine the upper and lower limits
  • the state monitoring method of early warning monitors the running state of the power pack and obtains the state monitoring results. It is confirmed that the use of the hybrid prediction model and the state monitoring method combining the upper and lower limit early warning with the core attribute parallel parameters can effectively reduce the false alarm rate of the power group.
  • Fig. 1 is the overall flow chart of the fault prediction method based on the hybrid prediction model of the present invention
  • Fig. 2 is the flow chart of ARIMA model fault prediction of the present invention
  • Fig. 3 is the flow chart of ANN model fault prediction of the present invention.
  • Fig. 4 is the overall process of the present invention based on the hybrid prediction model combined with the core attribute parallel parameter state monitoring;
  • the present invention is a method for predicting power supply group faults of complex equipment based on a hybrid prediction model.
  • the fault prediction method is aimed at the operation data of complex equipment under image stabilization conditions.
  • the hybrid prediction model is composed of a differential integrated moving average autoregressive model (ARIMA). ) and artificial neural network model (ANN) combination of fault prediction model;
  • the fault prediction method of the present invention includes the following steps:
  • S1 Analyze the typical failure mode of the power group of complex equipment, and use the difference matrix based on rough sets to extract the core attributes of the evaluation index of the power group of complex equipment, obtain the core attribute set, and divide the acquired time series X of the core attributes into Linear part Lt and nonlinear part Nt.
  • five typical fault state modes of typical complex equipment power supply groups can be obtained by analysis, including: normal power supply group status, ⁇ 15V power supply hidden danger status, power supply 26V01 hidden danger status, power supply 26V02 hidden danger status, and main power supply 26V status hidden state;
  • the present invention obtains the core attribute set of the power supply group, as shown below, the difference matrix M.
  • the present invention obtains the core attribute set of the power supply group, and the difference elements in the difference matrix M(T) are sets composed of conditional attributes. k i to represent.
  • the first case is, first, make x i and x j (i ⁇ j) obtain conditional attributes of different values, which constitute the difference element m ij , which means that in this conditional attribute set, any conditional attribute can be To distinguish x i from x j , just take one, and its relationship is called the disjunctive relationship " ⁇ ".
  • conditional attributes that distinguish x 1 from x 5 are c 5 , c 6 , c 7 , c 8 , and any conditional attribute can distinguish x 1 from x 5 , so just take one, which is called a disjunctive relation, denoted as: c 5 ⁇ c 6 ⁇ c 7 ⁇ c 8 ; Second, only c 5 can distinguish x 1 from x 6 .
  • the system is called a conjunction, denoted as: c 5 ⁇ (c 5 ⁇ c 6 ⁇ c 7 ⁇ c 8 ).
  • the difference matrix of the two situations in steps 3) and 4) is an empty set instead of 0.
  • the decision attribute D is the typical fault state mode of the power group of complex equipment, including five cases: the normal state of the power group is set to 1, the hidden state of the ⁇ 15V power supply is set to 2, the hidden state of the power supply 26V01 is set to 3, and the hidden state of the power supply 26V02 is set to 3.
  • conditional properties that can distinguish all individuals x i and x j in pairs should be satisfied, and the "conjunction" of the difference elements of all columns and the conjunction of all the difference elements also determine fM(T).
  • the linear time series prediction based on the ARIMA model includes the following steps:
  • step S1 the collected original time series of the power supply group is subjected to stabilization processing, specifically, the differential form is used to perform stabilization processing on the non-stationary time series data obtained due to the operation in the complex environment in the field, using the following method: is the first-order difference, and passed the ADF test and KPSS test.
  • Model identification of the ARIMA model draw the autocorrelation diagram and partial autocorrelation diagram of the stationary time series, and obtain the perceptual knowledge of the autoregressive order n and the moving average order m of the ARIMA model according to the autocorrelation diagram and the partial autocorrelation diagram ;
  • Akaike Information Criterion and the Bayesian Information Criterion to calculate the model order (n, m);
  • step S4 the purpose of checking the residual error is to ensure that the order of the model is appropriate, and the residual error is the residual amount after subtracting the time series fitted by the model from the original time series, including the following steps:
  • step S5 the determined ARIMA model is used to predict the linear time series, and the ratio of the data volume of the training data to the test data is 3:1;
  • the above a(t) is a random error.
  • the use of the ANN model to predict a time series with a nonlinear change law includes the following steps:
  • step S1 the ratio of the data amount of the training set to the test set is 3:1.
  • step S3 creating an ANN model includes the following steps:
  • step S4 the performance of the model is evaluated, and the evaluation indicators include: mean absolute error, mean square error, and mean absolute percentage error.
  • the mean absolute error is the average of the absolute values of the deviations of all individual observations and the arithmetic mean, which can avoid the problem of mutual cancellation of errors, so it can accurately reflect the size of the actual prediction error;
  • the mean square error is the mathematical expectation of the square of the difference of an estimator of the overall parameter determined according to the sub-sample, which reflects a measure of the degree of difference between the estimator and the estimated quantity.
  • the standard error can also be obtained, which is also used to measure the observation. the deviation between the value and the true value;
  • the mean absolute percentage error refers to the percentage value that the predicted result deviates from the actual result on average. It is a percentage value, so it is easier to understand than other statistics;
  • the state monitoring method of the core attribute parallel parameter monitoring combined with the upper and lower limit early warning is used to monitor the running state of the power supply group to obtain the state monitoring result, including the following steps:
  • step S1 the attribute reduction method based on rough set difference matrix includes the following steps:
  • the evaluation indicators of the prediction result error are: mean absolute error, mean square error, mean absolute percentage error.
  • S3 Use the ANN model to predict the time series of the parallel parameters of the core attributes of the power supply group, obtain the prediction results of the parallel parameters of the core attributes, compare with the upper and lower limit values, and evaluate the error of the prediction results;
  • the ANN model is used alone to predict the time series of the parallel parameters of the core attributes of the power group, and the prediction results of the parallel parameters of the core attributes are obtained, compared with the upper and lower limit values, and the error of the prediction results is evaluated, including: The following steps:
  • S4 Use the hybrid prediction model ARIMA-ANN to predict the time series of the parallel parameters of the core attributes of the power group, obtain the prediction results of the parallel parameters of the core attributes, compare with the upper and lower limit values, and evaluate the error of the prediction results. Contains the following steps:
  • step S3 use the hybrid prediction model ARIMA-ANN to predict the time series of the parallel parameters of the core attributes of the power group, obtain the prediction results of the parallel parameters of the core attributes, compare with the upper and lower limit values, and compare the prediction results.
  • the specific steps include:

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Geometry (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Business, Economics & Management (AREA)
  • Game Theory and Decision Science (AREA)
  • Computer Hardware Design (AREA)
  • Development Economics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

属于复杂装备的故障预测技术领域，提供了一种基于混合预测模型的复杂装备电源组故障预测方法。首先分析复杂装备电源组的典型故障，并提取其中的核心属性集，将电源模块的时间序列分为线性部分和非线性部分，利用差分整合移动平均自回归模型对线性部分进行预测，对得到的残差利用人工神经网络模型进行预测，并将非线性部分的预测结果与线性部分的预测结果进行加和得到电源组的预测结果。并通过混合模型对核心属性并行参数进行监测，结合上、下限预警的状态监测方式，得到电源组的运行状态信息。基于混合预测模型复杂装备电源组故障预测方法，能有效实现对复杂装备电源组核心属性时间序列的准确预测，并有效降低电源组的虚警率。

Description

一种基于混合预测模型的复杂装备电源组故障预测方法 技术领域
本发明属于复杂装备的故障预测技术领域，涉及一种基于混合预测模型的复杂装备电源组故障预测方法，尤其涉及一种针对复杂装备稳像工况下的运行数据，基于ARIMA结合ANN的混合预测模型进行的复杂装备电源组故障预测方法。
背景技术
大型装备由于其结构复杂，一旦发生故障则会造成巨大的损失。因此，迫切需要提高复杂装备系统的可靠性、可修复性和安全性。然而目前故障诊断的工作主要集中在“当前”的运行状态，对系统故障预测与健康管理的研究则较少。复杂装备领域越来越趋向于智能化、集成化、数字化，各组成部分的机理复杂且高度相关。发生故障时，设备无法有效、及时地判断故障位置和原因。目前在解决复杂装备预测问题的过程中，还主要存在以下两种问题(1)单一的预测模型本身往往存在一些缺陷，无法达到有效预测的目的；(2)单一运行参数存在反映信息不足的问题，难以进行准确的预测。
发明内容
本发明为解决上述局限性，本发明提供了一种基于混合预测模型的复杂装备电源组故障预测方法，一方面提出了一种基于ARIMA结合ANN的高精度混合预测模型，用于监测复杂装备电源组关键参数的未来发展趋势，在混合预测模型的基础上利用核心属性并行参数对电源组的运行状态进行预测，为复杂装备的未来运行状态的综合监控提供了依据。另一方面建立了复杂装备电源组未来状态监测流程，指导复杂装备的运行状态预测的实施。
为了实现上述目标，本发明采用如下的技术方案：
一种基于混合预测模型的复杂装备电源组故障预测方法，该预测方法针对复杂装备稳像工况下的运行数据；
所述故障预测方法基于混合故障预测模型进行故障预测；
所述混合故障预测模型包括ARIMA预测模型结合ANN预测模型；
所述ARIMA预测模型用于对电源组呈线性变化规律的时间序列进行预测；
所述ANN预测模型用于对电源组呈非线性变化规律的时间序列进行预测；
所述混合预测模型用于对电源组的时间序列的预测结果进行整合，并利用核心属性并行参数进行状态监测；
步骤如下：
将电源组的时间序列分为线性部分和非线性部分，利用ARIMA模型对线性部分进行预测，将电源组的原始时间序列与线性预测结果做差值，获取包含非线性变化规律的残差e(t)，对得到的残差利用ANN模型进行预测，并将非线性部分的预测结果与线性部分的预测结果进行加和得到电源组的预测结果；
S1：分析电源组故障，提取核心属性集；
S1.1：建立复杂装备中电源组包含属性集的评价指标系统；
复杂装备包含电源组、CPU板、KZB板、I/O板、ADA板、角速度传感器、横风传感器和倾斜传感器；
S1.2：分析属性之间的相关性，并利用基于粗糙集的差别矩阵进行属性约简；
S1.2.1：基于差别矩阵的定义计算差别矩阵M(T)；
S1.2.2：基于已得差别矩阵M(T)计算差别函数fM(T)；
S1.3：基于最小析取范式获取核心属性集；
根据所述的差别函数fM(T)，利用最小析取范式，进行属性约简，获取核心属性集；
计算上述经基于粗糙集差别矩阵的属性约简方法提取的核心属性集的上、下限值，上限值＝标准值+10％标准值、下限值＝标准值-10％标准值。
S2：利用ARIMA模型对呈线性变化规律的时间序列进行预测，获得包含非线性信息的残差；
S2.1：对采集到的电源组原始时间序列进行差分处理，得到平稳化处理的平稳时间序列；
S2.2：ARIMA模型识别：绘制平稳时间序列的自相关图与偏自相关图，依据自相关图与偏自相关图，获取ARIMA模型的自回归阶数n和移动平均阶数m的感性认识；利用赤池信息准则与贝叶斯信息准则，计算获取模型阶数(n，m)；
S2.3：参数估计：利用最小二乘法对ARIMA模型的参数进行参数估计；
S2.4：ARIMA模型的验证：对残差进行检验，判别残差是否是一段白噪声时间序列，即是否满足随机正态分布、不自相关；
S2.5：利用ARIMA模型对呈线性变化规律的时间序列进行预测；
S2.6：将电源组的原始时间序列与线性预测结果做差值，获取包含非线性变化规律的残差e(t)；
S3：利用ANN模型预测对呈非线性变化规律的时间序列进行预测，得到非线性预测结果；
S3.1：将核心属性集作为输入，将通过ARIMA模型获取包含非线性变化规律的残差e(t)作为输出，获取训练集和测试集；
S3.2：为防止数量级的影响做数据归一化处理；
S3.3：创建ANN模型、训练并测试；
S3.4：对ANN模型的性能进行评价；
S3.5：利用ANN模型获取呈非线性规律变化部分时间序列预测值e’(t)；
S4：利用ARIMA模型和ANN模型获得线性部分与非线性部分的预测结果，对两部分结果加和得到电源组的预测结果；
S4.1：单独利用ARIMA模型，对提取的核心属性集单参数进行预测，获得预测结果，并对预测结果的误差进行评价；
S4.2：单独利用ANN模型，对提取的核心属性集单参数进行预测，获得预测结果，并对预测结果的误差进行评价；
S4.3：基于混合预测模型，对提取的核心属性集单参数进行预测，获得电源组预测结果，并对预测结果的误差进行评价；
评价指标包含：平均绝对误差、均方误差、平均绝对百分比误差。
平均绝对误差是所有单个观测值与算术平均值的偏差的绝对值的平均，可以避免误差相互抵消的问题，因而可以准确反映实际预测误差的大小；
均方误差是根据子样确定的总体参数的一个估计量的差值平方的数学期望，反映了估计量与被估计量之间差异程度的一种度量，也可获得标准误差，同样来衡量观测值同真值之间的偏差；
平均绝对百分比误差是指在预测结果较真实结果平均偏离的百分比值，它是一个百分比值，因此比其他统计量更容易理解；
S4.4：通过对三种模型的预测结果的误差进行对比，将ARIMA模型结合ANN模型的混合预测模型预测结果作为最终结果；
S5：利用核心属性并行参数监测结合上、下限预警的状态监测方式，对电源组的运行状态进行监测，得到状态监测结果；
S5.1：计算提取的核心属性集的上、下限值；
S5.2：利用ANN模型，对电源组的核心属性并行参数时间序列进行预测，获得核心属性并行参数的预测结果，与上、下限值进行对比，并对预测结果的误差进行评价；
S5.3：利用混合预测模型，对电源组的核心属性并行参数时间序列进行预测，获得核心属性并行参数的预测结果，与上、下限值进行对比，并对预测结果的误差进行评价；
S5.4：获得对比结果，确认利用混合预测模型，并通过核心属性并行参数监测结合上、下限预警的状态监测方式，可有效降低电源组的虚警率。
上述采用ANN模型和ARIMA-ANN模型对核心属性并行参数进行预测，包含：
(1)并未加入ARIMA模型对核心属性并行参数的预测结果进行对比，原因在于ARIMA是对时间序列线性变化规律进行预测，只适用于单参数的时间序列的预测；
(2)将利用ARIMA-ANN模型与利用ARIMA模型和ANN模型对核心属性单参数预测结果进行对比，将利用ARIMA-ANN模型与利用ANN模型对核心属性并行参数预测结果进行对比，发现ARIMA-ANN模型较单模型的预测效果更精准。
(3)利用ARIMA-ANN模型对核心属性并行参数的运行状态的变化趋势进行预测，将其预测结果与利用ARIMA-ANN模型对核心属性单参数的运行状态的变化趋势的预测结果进行对比，发现利用核心属性并行参数的监测效果更精准。
本发明的有益效果：通过将复杂装备电源组的特征参数进行特征提取，得到能够表达电源组属性的核心属性集，并将核心属性集时间序列看作由线性变化规律与非线性变化规律两部分时间序列共同组成；利用ARIMA模型获取呈线性变化规律的时间序列预测结果，以及包含非线性变化规律的残差e(t)；利用ANN模型对包含非线性变化规律时间序列的残差e(t)进行预测，得到非线性变化规律时间序列的预测结果；将线性部分和非线性部分进行综合得到最终的预测结果，并基于已得到的混合故障预测模型，利用核心属性并行参数监测结合上、下限预警的状态监测方式，对电源组的运行状态进行监测，得到状态监测结果。确认利用混合预测模型，并通过核心属性并行参数结合上、下限预警的状态监测方式，可有效降低电源组的虚警率。
附图说明
图1为本发明基于混合预测模型的故障预测方法的整体流程图；
图2为本发明ARIMA模型故障预测的流程图；
图3为本发明ANN模型故障预测的流程图；
图4为本发明基于混合预测模型结合核心属性并行参数状态监测的整体流程；
具体实施方式
以下结合附图和技术方案，进一步说明本发明的具体实施方式。
本发明的一种基于混合预测模型的复杂装备电源组故障预测方法，所述故障预测方法针对复杂装备稳像工况下的运行数据，所述混合预测模型由差分整合移动平均自回归模型(ARIMA)和人工神经网络模型(ANN)组合构成的故障预测模型；
如图1所示，本发明的故障预测方法包括以下步骤：
S1：分析复杂装备电源组的典型故障模式，并利用基于粗糙集的差别矩阵对复杂装备电源组的评价指标进行核心属性提取，获取核心属性集，将获取后的核心属性的时间序列X分为线性部分Lt和非线性部分Nt。
S2：对其中线性部分Lt利用ARIMA模型进行预测，得到预测结果L’t和其与原始数据序列的残差e(t)，其隐含了非线性时间序列的信息。
S3：利用ANN模型对包含了非线性时间序列信息的残差e(t)进行预测，得到了非线性时间序列的预测结果e’(t)；
S4：将得到的线性时间序列和非线性时间序列的预测结果进行加和，得到电源组最终的预测结果X’＝e’(t)+L’t。
S5：基于混合预测模型，利用核心属性并行参数监测结合上下限预警的方式，对电源模块的运行状态监测，得到状态监测结果。
本实施例中，分析可得典型的复杂装备电源组的典型故障状态模式有五种情况，包含：电源组状态正常、±15V电源隐患状态、电源26V01隐患状态、电源26V02隐患状态、主电源26V隐患状态；
本实施例中，本发明获取电源组的核心属性集，如下所示，差别矩阵M。本实施例中，本发明获取电源组的核心属性集，差别矩阵M(T)中的差别元素是由条件属性构成的集合，由于该差别矩阵中的差别元素较多，为了方便体现，用字母k i来表示。
Figure PCTCN2021072780-appb-000001
决策属性值不同时：
1)第一种情况是，首先，使得x i与x j(i≠j)取得不同值的条件属性，其构成差别元素 m ij，含义是在此条件属性集合中，任意一个条件属性都可把x i与x j区分开，所以取一个即可，其关系称为析取关系“·”，以x 1与x 5为例，把x 1与x 5区分的条件属性是c 5，c 6，c 7，c 8,并且任意一个条件属性都可把x 1与x 5区分开，所以取一个即可，称为析取关系，记作：c 5·c 6·c 7·c 8；其次能把x 1与x 6区分开的只有c 5。
那么，能同时把x 1、x 5与x 6区分开的元素是c 5与(c 5·c 6·c 7·c 8)同时满足，这种逻辑关
系称为合取关系，记作：c 5∧(c 5·c 6·c 7·c 8)。
2)另一与之相反的情况是，无条件属性区分x i与x j取值，此时为空集。
决策属性相同时，有两种情况是可以不考虑的：
3)第一种是在差别矩阵的主对角线上的元素也就是U i＝U j；
4)另一种则是无论条件属性取值是否相同都不具有使决策属性区分的能力；
实施例，步骤3)、4)中两种情况的差别矩阵均为空集
Figure PCTCN2021072780-appb-000002
而不是0。
实施例中，所述条件属性C由电源组的13个评价指标构成，即C＝c i(i＝1,2,...,13)，所述x i与x j为采样的复杂装备电源组历史数据。决策属性D为复杂装备电源组的典型故障状态模式，包含五种情况：电源组状态正常设为1、±15V电源隐患状态设为2、电源26V01隐患状态设为3、电源26V02隐患状态设为3、主电源26V隐患状态设为5，即D＝(1,2,3,4,5)；
于是能把所有个体x i与x j两两区分开的条件属性应该满足，所有列的差别元素的“合取式”，全部的差别元素的合取式也确定fM(T)。
具体k i表示元素如下：
k 1＝k 23＝k 24～k 29＝k 32～k 35＝k 37＝k 51＝k 53＝k 57＝k 58＝k 59＝c 5,c 6,c 7,c 8；
k 2＝k 3＝k 4＝k 5＝k 6＝k 7＝k 9＝k 11＝k 12＝k 13＝k 15＝k 17＝k 19＝k 21＝k 22＝k 40＝k 42＝k 44＝k 46＝k 47＝k 61＝k 63＝k 65＝k 66＝k 68＝k 70＝k 72＝k 73＝k 75＝k 76＝k 77＝k 80＝k 81＝k 82＝k 83＝k 84＝k 93＝k 95＝k 96＝k 98＝k 99＝k 100＝k 101＝k 102＝k 103＝k 105＝k 106＝k 107＝k 112＝k 113＝k 114＝k 115＝k 120＝k 121＝c 5；
k 16＝k 20＝k 41＝k 45＝k 60＝k 64＝k 67＝k 71＝k 87＝k 88＝k 90＝k 94＝k 99＝k 116＝k 118＝c 8；
k 5＝k 10＝k 31＝k 74＝k 79＝k 86＝k 89＝k 91＝k 119＝c 5,c 8；
k 8＝k 78＝k 109＝k 111＝c 5,c 6,c 7；
k 43＝k 54＝k 62＝k 69＝k 92＝k 97＝k 102＝k 111＝c 6,c 7；
k 14＝k 30＝k 38,39＝k 48,50＝k 52＝k 55,56＝k 58＝k 85＝c 6,c 7,c 8；
如图2所示，所述基于ARIMA模型预测线性时间序列，包括以下步骤：
S1：对采集到的电源组原始时间序列进行平稳化处理；
实施例中，步骤S1所述对采集到的电源组原始时间序列进行平稳化处理，具体为采用差分形式，将由于运行在野外复杂环境中获取的不平稳时间序列数据进行平稳化处理，采用的是一阶差分，并通过了ADF检验和KPSS检验。
S2：ARIMA模型的模型识别：绘制平稳时间序列的自相关图与偏自相关图，依据自相关图与偏自相关图，获取ARIMA模型的自回归阶数n和移动平均阶数m的感性认识；利用赤池信息准则与贝叶斯信息准则，计算获取模型阶数(n，m)；
S3：参数估计：利用最小二乘法对ARIMA模型的参数进行参数估计；
S4：模型的验证：对残差进行检验，判别残差是否是一段白噪声时间序列，即是否满足随机正态分布、不自相关；
实施例中，步骤S4中，对残差进行检验目的是为确保模型的阶数合适，残差即为原始时间序列减掉模型拟合出的时间序列后的残余量，包括以下步骤：
1)在残差检验的结果图中。标准化残差是查看残差是否接近正态分布，理想的残差要接近正态分布；
2)依据自相关图(ACF)与偏自相关图(PACF)检验残差的自相关和偏自相关性，一般不存在超出边界的点；
3)检验残差是否接近正态分布，理想的情况下输入样本分位数与标准正太分位数应该靠近。
S5：利用ARIMA模型对呈线性变化规律的时间序列进行预测；
实施例中，步骤S5中，是利用已经确定的ARIMA模型对线性时间序列进行预测，采用的训练数据与测试数据的数据量的比值采用3：1；
S6：将电源组的原始时间序列与线性预测结果做差值，获取包含非线性变化规律的残差e(t)；
实施例中，步骤S6中，获取的包含非线性变化规律的残差e(t)＝f(e(t-1)，e(t-2)，...e(t-n))+a(t)；
上述的a(t)为随机误差。
如图3所示，所述利用ANN模型对呈非线性变化规律的时间序列进行预测，包含以下步骤：
S1：将核心属性集作为输入，将通过ARIMA模型获取包含非线性变化规律的残差e(t)作为输出，获取训练集和测试集；
实施例中，步骤S1中：训练集与测试集的数据量的比值采用的是3：1。
S2：为防止数量级的影响做数据归一化处理；
S3：创建ANN模型、训练并测试；
实施例中，步骤S3中：创建ANN模型包含以下步骤：
1)创建神经网络，三个输入、三个输出、4个隐含层；
2)设置模型的迭代次数为1000次、训练目标＝1e-6、以及学习率＝0.01；
3)训练网络，用训练好的ANN模型进行仿真测试，对预测数据进行反归一化；
S4：对ANN模型的性能进行评价：
实施例中，步骤S4中，对模型进行性能评价，评价指标包含：平均绝对误差、均方误差、平均绝对百分比误差。
平均绝对误差是所有单个观测值与算术平均值的偏差的绝对值的平均，可以避免误差相互抵消的问题，因而可以准确反映实际预测误差的大小；
均方误差是根据子样确定的总体参数的一个估计量的差值平方的数学期望，反映了估计量与被估计量之间差异程度的一种度量，也可获得标准误差，同样来衡量观测值同真值之间的偏差；
平均绝对百分比误差是指在预测结果较真实结果平均偏离的百分比值，它是一个百分比值，因此比其他统计量更容易理解；
S5：利用ANN模型获取呈非线性规律变化的时间序列预测值e’(t)。
如图4所示，利用核心属性并行参数监测结合上、下限预警的状态监测方式，对电源组的运行状态进行监测，得到状态监测结果，包括以下步骤：
S1：分析典型故障模式以及核心属性的提取，基于粗糙集差别矩阵的属性约简方法提取核心属性集
实施例中，步骤S1中，基于粗糙集差别矩阵的属性约简方法，包含以下步骤：
1)基于差别矩阵的定义计算差别矩阵M(T)；
2)基于已得差别矩阵M(T)计算差别函数fM(T)；
3)基于2)中的fM(T)，利用最小析取范式，获取核心属性集，进行属性约简。
S2：计算经基于粗糙集差别矩阵的属性约简方法提取的核心属性集的上、下限值，给出预测结果误差的评价指标：
1)核心属性集的上、下限值为：上限值＝标准值+10％标准值、下限值＝标准值-10％标准值；
2)预测结果误差的评价指标为：平均绝对误差、均方误差、平均绝对百分比误差。
S3：利用ANN模型，对电源组的核心属性并行参数时间序列进行预测，获得核心属性并行参数的预测结果，与上、下限值进行对比，对预测结果的误差进行评价；
实施例中，单独利用ANN模型，对电源组的核心属性并行参数时间序列进行预测，获得核心属性并行参数的预测结果，与上、下限值进行对比，并对预测结果的误差进行评价，包含以下步骤：
1)利用已经训练好的ANN模型，对已经提取的核心属性并行参数进行预测；
2)对核心属性的上、下限值进行计算，有效值＝标准值±10％标准值，超出有效值即为超出限值；
3)利用已经训练好的ANN模型，对已经提取的核心属性进行核心属性单参数的预测；
4)将核心属性单参数预测值与有效值进行对比，获取是否超出边界的预警提示；
5)发现在核心属性单参数状态监测的预警结果中，在核心属性并行参数监测结果中未发出预警信号；
6)对预测结果的误差进行评价。
S4：利用混合预测模型ARIMA-ANN，对电源组的核心属性并行参数时间序列进行预测，获得核心属性并行参数的预测结果，与上、下限值进行对比，并对预测结果的误差进行评价，包含以下步骤：
实施例中，步骤S3：利用混合预测模型ARIMA-ANN，对电源组的核心属性并行参数时间序列进行预测，获得核心属性并行参数的预测结果，与上、下限值进行对比，并对预测结果的误差进行评价，具体步骤包括：
1)利用已经训练好的混合预测模型ARIMA-ANN，对已经提取的核心属性并行参数进行预测；
2)对核心属性的上、下限值进行计算，有效值＝标准值±10％标准值，超出有效值即为超出限值；
3)利用已经训练好的混合预测模型ARIMA-ANN，对已经提取的核心属性进行核心属性单参数的预测；
4)将核心属性单参数预测值与有效值进行对比获取是否超出边界的预警提示；
5)发现在核心属性单参数状态监测的预警结果中，在核心属性并行参数监测结果中未发出预警信号；
6)对预测结果的误差进行评价。
S5：通过上述对比单独ANN模型与混合预测模型ARIMA-ANN的误差评价指标可得，混合预测模型的预测准确度高于单模型的预测准确度；通过对比核心属性单参数与核心属性并行参数的状态监测结果可获得，利用核心属性并行参数的状态监测方式可明显降低电源组的虚警率。

Claims (1)

  1. 一种基于混合预测模型的复杂装备电源组故障预测方法，该预测方法针对复杂装备稳像工况下的运行数据，其特征在于：
    所述混合预测模型由差分整合移动平均自回归模型ARIMA和人工神经网络模型ANN组合构成的故障预测模型；
    所述ARIMA模型用于对电源组呈线性变化规律的时间序列进行预测；
    所述ANN模型用于对电源组呈非线性变化规律的时间序列进行预测；
    所述混合预测模型用于对电源组呈线性变化规律的时间序列的预测结果和呈非线性变化规律的时间序列的预测结果进行整合，并利用核心属性并行参数进行状态监测；
    步骤如下：
    将电源组的原始时间序列分为线性部分和非线性部分，利用ARIMA模型对线性部分进行预测，得到线性部分预测结果，将电源组的原始时间序列与线性预测结果做差值，获取包含非线性变化规律的残差e(t)；利用ANN模型进行预测，并将非线性部分的预测结果与线性部分的预测结果进行加和得到电源组的预测结果：
    S1：分析电源组故障，提取核心属性集；
    S1.1：建立复杂装备中电源组所包含属性集的评价指标系统；
    所述复杂装备包含电源组、CPU板、KZB板、I/O板、ADA板、角速度传感器、横风传感器和倾斜传感器；
    S1.2：分析属性之间的相关性，并利用基于粗糙集的差别矩阵进行属性约简；
    S1.2.1：基于差别矩阵的定义计算差别矩阵M(T)；
    S1.2.2：基于已得差别矩阵M(T)计算差别函数fM(T)；
    S1.3：基于最小析取范式获取核心属性集；
    根据所述的的差别函数fM(T)，利用最小析取范式，进行属性约简，获取核 心属性集；
    S2：利用ARIMA模型对呈线性变化规律的时间序列进行预测，获得包含非线性信息的残差；
    S2.1：对采集到的电源组原始时间序列进行差分处理，得到平稳化处理的平稳时间序列；
    S2.2：ARIMA模型识别；
    绘制平稳时间序列的自相关图与偏自相关图，依据自相关图与偏自相关图，获取ARIMA模型的自回归阶数n和移动平均阶数m的感性认识；利用赤池信息准则与贝叶斯信息准则，计算获取模型阶数(n，m)；
    S2.3：参数估计：利用最小二乘法对ARIMA模型的参数进行参数估计；
    S2.4：ARIMA模型的验证：对残差进行检验，判别残差是否是一段白噪声时间序列，即是否满足随机正态分布、不自相关；
    S2.5：利用ARIMA模型对呈线性变化规律的时间序列进行预测；
    S2.6：将电源组的原始时间序列与线性预测结果做差值，获取包含非线性变化规律的残差e(t)；
    S3：利用ANN模型对呈非线性变化规律的时间序列进行预测，得到非线性预测结果；
    S3.1：将核心属性集作为输入，将通过ARIMA模型获取包含非线性变化规律的残差e(t)作为输出，获取训练集和测试集；
    S3.2：为防止数量级的影响做数据归一化处理；
    S3.3：创建ANN模型，并进行训练、测试；
    S3.4：对ANN模型的性能进行评价；
    S3.5：利用ANN模型获取呈非线性规律变化的时间序列预测值e’(t)；
    S4：利用ARIMA模型和ANN模型获得线性部分与非线性部分的预测结果，对两部分结果加和得到电源组的预测结果；
    S4.1：单独利用ARIMA模型，对提取的核心属性集单参数进行预测，获得预测结果，并对预测结果的误差进行评价；
    S4.2：单独利用ANN模型，对提取的核心属性集单参数进行预测，获得预测结果，并对预测结果的误差进行评价；
    S4.3：基于混合预测模型，对提取的核心属性集单参数进行预测，获得电源组预测结果，并对预测结果的误差进行评价；
    S4.4：通过对三种模型的预测结果的误差进行对比，将ARIMA模型结合ANN模型的混合预测模型预测结果作为最终结果；
    S5：利用核心属性并行参数监测结合上、下限预警的状态监测方式，对电源组的运行状态进行监测，得到状态监测结果；
    S5.1：计算提取的核心属性集的上、下限值；
    S5.2：利用ANN模型，对电源组的核心属性并行参数时间序列进行预测，获得核心属性并行参数的预测结果，与上、下限值进行对比，并对预测结果的误差进行评价；
    S5.3：利用混合预测模型，对电源组的核心属性并行参数时间序列进行预测，获得核心属性并行参数的预测结果，与上、下限值进行对比，并对预测结果的误差进行评价；
    S5.4：获得对比结果，确认利用混合预测模型，并通过核心属性并行参数监测结合上、下限预警的状态监测方式，可有效降低电源组的虚警率。
PCT/CN2021/072780 2021-01-11 2021-01-20 一种基于混合预测模型的复杂装备电源组故障预测方法 WO2022147853A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/311,931 US20220341996A1 (en) 2021-01-11 2021-01-20 Method for predicting faults in power pack of complex equipment based on a hybrid prediction model

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110029485.7A CN112685910B (zh) 2021-01-11 2021-01-11 一种基于混合预测模型的复杂装备电源组故障预测方法
CN202110029485.7 2021-01-11

Publications (1)

Publication Number Publication Date
WO2022147853A1 true WO2022147853A1 (zh) 2022-07-14

Family

ID=75457182

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/072780 WO2022147853A1 (zh) 2021-01-11 2021-01-20 一种基于混合预测模型的复杂装备电源组故障预测方法

Country Status (3)

Country Link
US (1) US20220341996A1 (zh)
CN (1) CN112685910B (zh)
WO (1) WO2022147853A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20230305551A1 (en) * 2022-03-25 2023-09-28 Yokogawa Electric Corporation Method and system for automated fault detection
CN115757561A (zh) * 2022-12-09 2023-03-07 国网湖北省电力有限公司信息通信公司 一种数据中心设备运行数据时间序列组合预测方法
CN116523120A (zh) * 2023-04-14 2023-08-01 成都飞机工业（集团）有限责任公司 一种作战系统健康状态预测方法
CN116167527B (zh) * 2023-04-21 2023-09-12 南方电网数字电网研究院有限公司 纯数据驱动的电力系统静态安全运行风险在线评估方法
CN117390465B (zh) * 2023-12-11 2024-02-20 天津南大通用数据技术股份有限公司 一种工作负载预测方法

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2390447A (en) * 2002-07-02 2004-01-07 Hewlett Packard Co Fault prediction in logical networks
US7324924B2 (en) * 2006-01-27 2008-01-29 Gm Global Technology Operations, Inc. Curve fitting for signal estimation, prediction, and parametrization
CN105376097A (zh) * 2015-11-30 2016-03-02 沈阳工业大学 网络流量的一种混合预测方法
CN107085750A (zh) * 2017-03-10 2017-08-22 广东工业大学 一种基于arma和ann的混合动态故障预测方法
CN108052528A (zh) * 2017-11-09 2018-05-18 华中科技大学 一种存储设备时序分类预警方法
CN109886316A (zh) * 2019-01-29 2019-06-14 西安交通大学 基于云系相似度权重分配的变压器状态参量组合预测方法
CN110221139A (zh) * 2019-05-05 2019-09-10 新奥数能科技有限公司 一种干式变压器的故障预测方法、装置及系统
CN111445006A (zh) * 2020-02-27 2020-07-24 上海交通大学 开源社区中开发者代码提交次数的预测方法及系统
US10769347B1 (en) * 2018-04-27 2020-09-08 Synopsys, Inc. Predicting no-defect-found physical failure analysis results using Bayesian inference and generalized linear models

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8055485B2 (en) * 2007-01-29 2011-11-08 International Business Machines Corporation Prediction of trends based on history values of an observable
US11022720B2 (en) * 2019-10-25 2021-06-01 The Florid International University Board of Trustees System for forecasting renewable energy generation
CN111985361A (zh) * 2020-08-05 2020-11-24 武汉大学 小波降噪和emd-arima的电力系统负荷预测方法及系统

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2390447A (en) * 2002-07-02 2004-01-07 Hewlett Packard Co Fault prediction in logical networks
US7324924B2 (en) * 2006-01-27 2008-01-29 Gm Global Technology Operations, Inc. Curve fitting for signal estimation, prediction, and parametrization
CN105376097A (zh) * 2015-11-30 2016-03-02 沈阳工业大学 网络流量的一种混合预测方法
CN107085750A (zh) * 2017-03-10 2017-08-22 广东工业大学 一种基于arma和ann的混合动态故障预测方法
CN108052528A (zh) * 2017-11-09 2018-05-18 华中科技大学 一种存储设备时序分类预警方法
US10769347B1 (en) * 2018-04-27 2020-09-08 Synopsys, Inc. Predicting no-defect-found physical failure analysis results using Bayesian inference and generalized linear models
CN109886316A (zh) * 2019-01-29 2019-06-14 西安交通大学 基于云系相似度权重分配的变压器状态参量组合预测方法
CN110221139A (zh) * 2019-05-05 2019-09-10 新奥数能科技有限公司 一种干式变压器的故障预测方法、装置及系统
CN111445006A (zh) * 2020-02-27 2020-07-24 上海交通大学 开源社区中开发者代码提交次数的预测方法及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WANG AINA, YING-SHUN LI, ZHE HE: "State Evaluation of Fire Control System Based on Fusion of D-S Evidence Theory and Rough Set", CONTROL ENGINEERING OF CHINA, vol. 27, no. 12, 31 December 2020 (2020-12-31), pages 2176 - 2184, XP055949487, ISSN: 1671-7848, DOI: 10.14107/j.cnki.kzgc.20180421 *

Also Published As

Publication number Publication date
CN112685910A (zh) 2021-04-20
US20220341996A1 (en) 2022-10-27
CN112685910B (zh) 2022-11-18

Similar Documents

Publication Publication Date Title
WO2022147853A1 (zh) 一种基于混合预测模型的复杂装备电源组故障预测方法
MIRI et al. Application of fuzzy fault tree analysis on oil and gas offshore pipelines
Liu et al. Risk evaluation approaches in failure mode and effects analysis: A literature review
CN106650797B (zh) 一种基于集成elm的配电网窃电嫌疑用户智能识别方法
CN111046564B (zh) 两阶段退化产品的剩余寿命预测方法
Neves et al. An approach to decision‐making analysis for implementation of structural health monitoring in bridges
CN112083244B (zh) 综合化航空电子设备故障智能诊断系统
CN112785091A (zh) 一种对油田电潜泵进行故障预测与健康管理的方法
CN103675525A (zh) 基于多svdd模型的dc-dc变换器健康监测与故障预测方法
Morais et al. Analysis and estimation of human errors from major accident investigation reports
CN112257914B (zh) 一种基于随机森林的航空安全因果预测方法
CN117032165A (zh) 一种工业设备故障诊断方法
Daher et al. New prognosis approach for preventive and predictive maintenance—Application to a distillation column
CN110220885B (zh) 一种机械设备磨损状态综合判断方法
Kabir A fuzzy data-driven reliability analysis for risk assessment and decision making using Temporal Fault Trees
CN113112123B (zh) 一种基于关联矩阵的飞机航电系统故障诊断评估方法
Seiti et al. R. Graph: A new risk-based causal reasoning and its application to COVID-19 risk analysis
CN112784277B (zh) 基于改进的d-s证据理论的软件可信性综合评估方法
Kouhestani et al. Data-driven prognosis of failure detection and prediction of lithium-ion batteries
CN117495094A (zh) 一种产业链安全风险综合评估预警方法与系统
CN115564247A (zh) 一种核电厂事故智能识别及决策的方法及系统
CN112686530B (zh) 一种继电保护运行可靠性评价方法
CN115936293A (zh) 一种基于pca的地铁施工安全事故风险评价方法
CN106650945A (zh) 一种基于证据合成理论的软件体系结构安全性评估方法
Su et al. Fuzzy FMECA risk evaluation and its applications in Chinese train control systems based on cloud model

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21916906

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21916906

Country of ref document: EP

Kind code of ref document: A1