CN112711828A - Maintenance and spare part supply joint optimization method under partially observable information - Google Patents
Maintenance and spare part supply joint optimization method under partially observable information Download PDFInfo
- Publication number
- CN112711828A CN112711828A CN202010973772.9A CN202010973772A CN112711828A CN 112711828 A CN112711828 A CN 112711828A CN 202010973772 A CN202010973772 A CN 202010973772A CN 112711828 A CN112711828 A CN 112711828A
- Authority
- CN
- China
- Prior art keywords
- time
- spare part
- state
- expected
- data
- 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.)
- Pending
Links
- 238000012423 maintenance Methods 0.000 title claims abstract description 23
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000005457 optimization Methods 0.000 title claims abstract description 17
- 238000005070 sampling Methods 0.000 claims abstract description 30
- 238000012544 monitoring process Methods 0.000 claims abstract description 17
- 238000006731 degradation reaction Methods 0.000 claims abstract description 7
- 230000008439 repair process Effects 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 5
- 239000000126 substance Substances 0.000 claims description 4
- 230000036541 health Effects 0.000 claims description 3
- 230000015556 catabolic process Effects 0.000 claims description 2
- 238000005315 distribution function Methods 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 238000012546 transfer Methods 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Images
Classifications
-
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a maintenance and spare part supply combined optimization method under partially observable information, which comprises the steps of firstly, establishing a vector autoregressive model based on real-time multidimensional state monitoring data of a system, and calculating residual errors of an integral data set; secondly, establishing a hidden Markov model capable of reflecting the system degradation process based on residual data, and estimating unknown state parameters and observation parameters in the hidden Markov model by using an expectation-maximization algorithm; then, updating the probability density function of the remaining service life of the system at each sampling moment in real time based on Bayesian theorem; finally, the delivery time of the spare parts is considered as a random variable rather than a traditional constant, and the optimal part replacement time and the spare part ordering time are dynamically updated with the aim of minimizing the average cost rate. The invention can fully utilize the multi-dimensional state monitoring information and the prediction information of the system, guide a maintenance engineer to flexibly adjust the distribution of the corresponding spare part delivery time according to the actual requirement of the spare part and determine the optimal maintenance decision.
Description
Technical Field
The invention belongs to the field of maintenance engineering of mechanical systems, and particularly relates to a maintenance and spare part supply joint optimization method under partially observable information.
Background
The problem of maintenance and spare part optimization for non-serviceable mechanical systems is a critical link in the implementation of maintenance activities. In the existing maintenance model, reliability distribution based on historical failure statistics is widely used for a replacement model and a spare part inventory model, solving such problems as optimal replacement time of parts and spare part order time in the case of satisfying economic indicators. In addition, there is a need to optimize the quantity of orders and inventory of spare parts to minimize storage costs while maximizing spare part availability.
The traditional approach is to use a reliability function that determines lifetime based on historical failure time of the components, which reflects the statistical properties of the component population, without taking into account the potential physical failure process between components. And performing sequential optimization decision of part replacement and spare part ordering on the basis of statistical life distribution. However, this method does not take into account the variability of individual components, nor does it describe the lifetime distribution function of a single individual. For newly developed equipment and large or expensive critical equipment, historical failure data is mostly lacked, so that the distribution function of the service life is difficult to obtain. In recent years, advanced state monitoring technology enables prediction of residual life of an individual, and lays a foundation for maintenance decision based on prediction information.
Conventional repair and spare part decision models often have some assumptions that are not reasonable enough, such as assuming that the life distribution function of the component is known, that spare parts are always available or that spare part lead times are not considered, etc. Although the literature researches the part replacement and spare part ordering sequential optimization decision based on the service life prediction information, the decision is not a global optimal solution and is difficult to implement in engineering practice.
Disclosure of Invention
The purpose of the invention is as follows: the invention provides a maintenance and spare part supply combined optimization method under partially observable information, which can make full use of the state monitoring information of a system to formulate a maintenance plan and guide a maintenance engineer to flexibly adjust the distribution of spare part delivery time according to actual needs.
The technical scheme is as follows: the invention relates to a maintenance and spare part supply combined optimization method under partially observable information, which comprises the following steps:
(1) selecting a healthy data part to establish a vector autoregressive model based on real-time multidimensional state monitoring data of a monitored mechanical system, and calculating a residual error of an integral data set to enable preprocessed data to meet normality and independence;
(2) establishing a hidden Markov model reflecting the system degradation process based on residual data, and estimating unknown state parameters in the hidden Markov model by using an expectation-maximization algorithmAnd observed parameters
(3) Real-time updating system at each sampling moment t based on Bayesian theoremkProbability density function f (t | Π) of remaining useful life ofk);
(4) The optimal part replacement time and the spare part ordering time are dynamically updated by considering the delivery time of the spare part as a random variable rather than a traditional constant with the aim of minimizing the average cost rate.
Further, the step (1) is realized as follows:
for multidimensional state monitoring data collected by the sensor, it is expressed asd is the dimension of the data; the health data part is assumed to follow a stationary vector autoregressive process:
wherein epsilonnSubject to N for independent and same distributiond(0,∑);Is the order of the model;is an autocorrelation matrix;is an average value;is a covariance;
further, the step (2) is realized as follows:
establishing 3 states capable of reflecting the degradation of the monitored mechanical system, including a hidden Markov model of a state 0, a state 1 and a state 2, wherein the state 0 represents a healthy state, the state 1 represents an unhealthy state, and the state 2 represents a failed state; estimating state parameters in hidden Markov models using expectation-maximization algorithmsAnd observed parametersBy ΠkRepresents the sampling time t at the k-th timekGiven observation data yΔ,y2Δ,...,Posterior probability of system in state 1 under the conditions:
Πk=Pr(Xk=1|ξ>kΔ,yΔ,y2Δ,...,ykΔ) (3)
further, the step (3) is realized as follows:
the posterior probability pi of each sampling point is determined by Bayes' theoremkThe update can be iteratively updated by:
at sampling time tkThe conditional reliability function for the remaining life of the system is:
the probability density function is:
at sampling time tk(i.e., k-th sample, total time k Δ), the joint optimization function for repair and spare parts ordering under part of observable information is:
wherein E (CC) is the expected total cost of a life cycle, and E (CL) is the expected time length of a life cycle, so thatThe minimum value of the average cost rate of one period is the corresponding optimal replacement timeAnd spare part order timeConstraint conditionsIndicating that the spare part ordering time should be before the current sampling time and the replacement time; if the optimized spare part ordering time is after the next sampling time, the next sampling is continued without adopting a spare part ordering strategy, and the optimal replacement time and the optimal spare part ordering time are updated; and if the spare part ordering time is before the next sampling, taking maintenance measures according to the optimized replacement time and the spare part ordering time.
Further, the step (4) is realized as follows:
at sampling time tkThere are five possible scenarios between the spare part order time, the spare part arrival time, the replacement time, and the time to failure: a component failure before the spare part order time point; a component failure occurs between the point in time when the spare part has issued an order and the spare part has not been delivered; failure of the component between the time the spare part has been delivered and the optimal replacement time; the spare part has issued an order request but not delivered before the optimal replacement time point; spare parts have been delivered, and component failure occurs after optimal replacement time;
the expected spare part shortage times in the five cases are respectively represented by ES1, ES2, ES3, ES4 and ES5, and the expected spare part shortage times in each condition are calculated as follows:
ES3=0 (14)
ES5=0 (16)
the expected spare part shortage times in the five cases are respectively represented by EH1, EH2, EH3, EH4 and EH5, and the expected spare part holding time under each condition is calculated as follows:
EH1=0 (17)
EH2=0 (18)
EH4=0 (20)
the expected spare part shortage time ES for a life cycle is:
ES=ES1+ES2+ES3+ES4+ES5 (22)
the expected spare part hold time EH for a life cycle is:
EH=EH1+EH2+EH3+EH4+EH5 (23)
the expected total cost e (cc) for a life cycle is then:
the expected length of time in a life cycle, e (cl), is:
has the advantages that: compared with the prior art, the invention has the beneficial effects that: the invention can make a maintenance plan by fully utilizing the state monitoring information of the system, guides a maintenance engineer to flexibly adjust the distribution of spare part delivery time according to actual needs and further makes an optimal maintenance decision according with actual engineering application.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
As shown in fig. 1, the present invention provides a joint optimization method for maintenance and spare part supply under partially observable information, which specifically comprises the following steps:
step 1: based on real-time multidimensional state monitoring data of a monitored mechanical system, a healthy data part is selected to establish a vector autoregressive model, and the residual error of the whole data set is calculated, so that preprocessed data meet normality and independence.
For multidimensional state monitoring data collected by the sensor, it is expressed asK is the dimension of the data. The health data portion is assumed to follow a stationary Vector Autoregressive (VAR) process:
wherein epsilonnSubject to N for independent and same distributionK(0,∑);Is the order of the model;is an autocorrelation matrix;is an average value;is the covariance. The multidimensional state monitoring data refers to monitoring data collected by a plurality of sensors, the data collected by one sensor is one-dimensional, and generally a system can be provided with a plurality of sensors at different positions.
Using estimated VAR model parametersResidual error Y capable of calculating overall monitoring datan:
step 2: establishing a hidden Markov model reflecting the system degradation process based on residual data, and estimating unknown state parameters in the hidden Markov model by using an expectation-maximization algorithmAnd observed parametersReal-time updating system at each sampling moment t based on Bayesian theoremkProbability density function f (t | Π) of remaining useful life ofk)。
Based on the overall residual data, the degradation process of the system is assumed to conform to a Hidden Markov Model (HMM) of 3 states (state space S ═ {0, 1, 2 }): a healthy state (state 0), a warning state or unhealthy state (state 1), and a failed state (state 2). State 0 and state 1 are non-observable states, i.e., the states are hidden. Only state 2 can be observed directly. Suppose the system always starts in a healthy state, i.e., P (X)00) 1. The transient transfer rate matrix is:
wherein λ is01,λ02,λ12E (0, + ∞) is an unknown state parameter in the model, representing the state transition rate.
When xi ═ inf { t ≧ 0: xt2 represents the failure time of the system, and Δ is the sampling interval. Observed valueIndependent given the system state. By yΔ,y2Δ,...,Representing d-dimensional observations, then y in state xkΔObey Nd(μx,∑x) And x is a d-element normal distribution of 0 and 1, and the probability density function is as follows:
wherein the content of the first and second substances,the unknown observation parameters in the model respectively represent the mean value and the covariance in each state.
Assuming N sets of collected condition monitoring failure data, use F1,...,FNAnd (4) showing. Failure data FiBy usingDenotes, TiThe total number of samples of the ith group of data is xii=tiWherein T isiΔ<ti≤(Ti+1) Δ. Assuming that M sets of truncated data are collected, S is used1,...,SjAnd (4) showing. Likewise, truncated data SjBy usingIndicating time of failure ξi>TiAnd delta. With O ═ F1,...,FN,S1,...,SMDenotes the observed data, k ═ (a, Ψ | O) as the corresponding likelihood function, where a ═ λ (λ ═ O)01,λ02,λ12),Ψ=(μ0,μ1,∑0,∑1) Is the parameter to be estimated. Sample path (X) due to hidden Markov model state processt: t ≧ 0) is not observable, so the analytical expression for the maximum likelihood function is difficult to solve. The expectation-maximization (EM) algorithm may be solved by iteratively maximizing the pseudo-likelihood function. Order toAndfor the initial value of the parameter to be estimated, the EM algorithm comprises the following steps:
e-step: calculating a pseudo-likelihood function:
whereinFor a complete data set, i.e. an observation data set o each set of failure data FiAnd truncated data SjNon-observable sample path information for the state process is augmented.
M-step: selecting Λ*,Ψ*So that
Parameter Λ updated per step*,Ψ*Then used as the first generationE-step, so that E-step and M-step iterate operation until Euclidean normWhere ε is an arbitrarily small positive number.
By maximizing the expected value of each update step, the estimated values of the state parameters and the observation parameters of each update step, i.e. the points at which the derivative of each parameter is 0, can be obtained. Updated at each stepThe explicit expression of (c) is as follows:
wherein:
parameter in completion statusAnd observed parametersAfter the estimation. By ΠkIndicating that the observation data y is given at the time k ΔΔ,y2Δ,...,Posterior probability of system in state 1 under the conditions:
Πk=Pr(Xk=1|ξ>kΔ,yΔ,y2Δ,...,ykΔ) (12)
the posterior probability pi of each sampling point is determined by Bayes' theoremkThe update can be iteratively updated by:
wherein the initial value pi0=0,Pij(t)=P(Xt=j|X0I) is the transition probability matrix. Transition probability matrix Pij(t) torque available for transfer rateThe array is solved by Kolmogorov backward differential equations.
Thus, at the sampling time tkThe conditional reliability function for the remaining life of the system is:
the probability density function is:
and step 3: the optimal part replacement time and the spare part ordering time are dynamically updated by considering the delivery time of the spare part as a random variable rather than a traditional constant with the aim of minimizing the average cost rate.
At sampling time tk(i.e., kth sample, total time k Δ), there are five possible scenarios between spare part order time, spare part arrival time, replacement time, and failure time:
1) a component failure before the spare part order time point;
2) a component failure occurs between the point in time when the spare part has issued an order and the spare part has not been delivered;
3) failure of the component occurs between the time when the spare part has been delivered and the optimal replacement time.
4) The order request has been issued by the spare part, but the spare part is not delivered until the optimal replacement time point.
5) Spare parts have been delivered and failure of the part occurs after the optimal replacement time.
The expected spare part shortage times in the five cases are respectively represented by ES1, ES2, ES3, ES4 and ES5, and the expected spare part shortage times in each condition are calculated as follows:
ES3=0 (18)
ES5=0 (20)
wherein l is lead time; (l) is a probability density function of delivery time;is tkThe time of the order of the spare part at the moment,is tkTime of day component replacement.
The expected spare part shortage times in the five cases are respectively represented by EH1, EH2, EH3, EH4 and EH5, and the expected spare part holding time under each condition is calculated as follows:
EH1=0 (21)
EH2=0 (22)
EH4=0 (24)
from the above analysis, the expected spare part shortage time ES in a life cycle is:
ES=ES1+ES2+ES3+ES4+ES5 (26)
the expected spare part hold time EH for a life cycle is:
EH=EH1+EH2+EH3+EH4+EH5 (27)
the expected total cost e (cc) for a life cycle is then:
the expected length of time in a life cycle, e (cl), is:
combining the above analysis, the replacement and spare part ordering joint optimization function based on the prediction information is:
in the above equation, the value that minimizes the average cost rate of one cycle is the optimal replacement timeAnd spare part order timeThe constraint indicates that the spare part order time should be between the current sampling time and the replacement time.
And if the optimized spare part ordering time is larger than the next sampling time, the next sampling is continued without adopting a spare part ordering strategy. And if the spare part ordering time is before the next sampling, taking maintenance measures according to the optimized replacement time and the spare part ordering time.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to be limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, the protection scope of the present invention should be determined by the appended claims.
Claims (5)
1. A method for jointly optimizing maintenance and spare part supply under partially observable information, comprising the steps of:
(1) selecting a healthy data part to establish a vector autoregressive model based on real-time multidimensional state monitoring data of a monitored mechanical system, and calculating a residual error of an integral data set to enable preprocessed data to meet normality and independence;
(2) establishing a hidden Markov model reflecting the system degradation process based on residual data, and estimating unknown state parameters in the hidden Markov model by using an expectation-maximization algorithmAnd observed parameters
(3) Real-time updating system at each sampling moment t based on Bayesian theoremkProbability density function f (t | Π) of remaining useful life ofk);
(4) The optimal part replacement time and the spare part ordering time are dynamically updated by considering the delivery time of the spare part as a random variable rather than a traditional constant with the aim of minimizing the average cost rate.
2. The joint optimization method for repair and spare part supply under partially observable information according to claim 1, wherein the step (1) is implemented as follows:
for multidimensional state monitoring data collected by the sensor, it is expressed asd is the dimension of the data; the health data part is assumed to follow a stationary vector autoregressive process:
wherein epsilonnAre independently and identically distributed, obeyNd(0, Σ); p belongs to the order of the model; phir∈d×dIs an autocorrelation matrix; delta0∈dIs an average value; Σ ∈d×dIs a covariance;
3. the joint optimization method for repair and spare part supply under partially observable information as claimed in claim 1, wherein the step (2) is implemented as follows:
establishing 3 states capable of reflecting the degradation of the monitored mechanical system, including a hidden Markov model of a state 0, a state 1 and a state 2, wherein the state 0 represents a healthy state, the state 1 represents an unhealthy state, and the state 2 represents a failed state; estimating state parameters in hidden Markov models using expectation-maximization algorithmsAnd observed parametersBy ΠkRepresents the sampling time t at the k-th timekGiven observation data yΔ,y2Δ,...,ykΔ∈dPosterior probability of system in state 1 under the conditions:
Πk=Pr(Xk=1|ξ>kΔ,yΔ,y2Δ,...,ykΔ)。 (3)
4. the joint optimization method for repair and spare part supply under partially observable information as claimed in claim 1, wherein the step (3) is implemented as follows:
the posterior probability pi of each sampling point is determined by Bayes' theoremkThe update can be iteratively updated by:
at sampling time tkThe conditional reliability function for the remaining life of the system is:
the probability density function is:
at sampling time tk(i.e., k-th sample, total time k Δ), the joint optimization function for repair and spare parts ordering under part of observable information is:
wherein E (CC) is the expected total cost in a life cycle, E (CL) is the expected time length of a life cycle, and the value of minimizing the average cost rate of a cycle is the corresponding optimal replacement timeAnd spare part order timeConstraint conditionsIndicating that the spare part ordering time should be before the current sampling time and the replacement time; if the optimized spare part ordering time is after the next sampling time, the next sampling is continued without adopting a spare part ordering strategy, and the optimal replacement time and the optimal spare part ordering time are updated; and if the spare part ordering time is before the next sampling, taking maintenance measures according to the optimized replacement time and the spare part ordering time.
5. The joint optimization method for repair and spare part supply under partially observable information according to claim 1, wherein the step (4) is implemented as follows:
at sampling time tkThere are five possible scenarios between the spare part order time, the spare part arrival time, the replacement time, and the time to failure: a component failure before the spare part order time point; a component failure occurs between the point in time when the spare part has issued an order and the spare part has not been delivered; failure of the component between the time the spare part has been delivered and the optimal replacement time; the spare part has issued an order request but not delivered before the optimal replacement time point; spare parts have been delivered, and component failure occurs after optimal replacement time;
the expected spare part shortage times in the five cases are respectively represented by ES1, ES2, ES3, ES4 and ES5, and the expected spare part shortage times in each condition are calculated as follows:
ES3=0 (14)
ES5=0 (16)
the expected spare part shortage times in the five cases are respectively represented by EH1, EH2, EH3, EH4 and EH5, and the expected spare part holding time under each condition is calculated as follows:
EH1=0 (17)
EH2=0 (18)
EH4=0 (20)
the expected spare part shortage time ES for a life cycle is:
ES=ES1+ES2+ES3+ES4+ES5 (22)
the expected spare part hold time EH for a life cycle is:
EH=EH1+EH2+EH3+EH4+EH5 (23)
the expected total cost e (cc) for a life cycle is then:
the expected length of time in a life cycle, e (cl), is:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010973772.9A CN112711828A (en) | 2020-09-16 | 2020-09-16 | Maintenance and spare part supply joint optimization method under partially observable information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010973772.9A CN112711828A (en) | 2020-09-16 | 2020-09-16 | Maintenance and spare part supply joint optimization method under partially observable information |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112711828A true CN112711828A (en) | 2021-04-27 |
Family
ID=75542377
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010973772.9A Pending CN112711828A (en) | 2020-09-16 | 2020-09-16 | Maintenance and spare part supply joint optimization method under partially observable information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112711828A (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103632214A (en) * | 2013-12-23 | 2014-03-12 | 清华大学 | Maintenance and spare part supply combined optimization method for use in absence of inventory degradation data |
CN109117566A (en) * | 2018-08-24 | 2019-01-01 | 中国电子科技集团公司第三十六研究所 | A kind of Combined maintenance planing method based on Survey of product life prediction model |
-
2020
- 2020-09-16 CN CN202010973772.9A patent/CN112711828A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103632214A (en) * | 2013-12-23 | 2014-03-12 | 清华大学 | Maintenance and spare part supply combined optimization method for use in absence of inventory degradation data |
CN109117566A (en) * | 2018-08-24 | 2019-01-01 | 中国电子科技集团公司第三十六研究所 | A kind of Combined maintenance planing method based on Survey of product life prediction model |
Non-Patent Citations (1)
Title |
---|
李鑫: "《基于监测数据的飞机齿轮箱健康预测及维修优化方法研究》", 《《中国博士学位论文全文数据库 工程科技II辑》》, no. 1, pages 031 - 58 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7769561B2 (en) | Robust sensor correlation analysis for machine condition monitoring | |
CN111142501B (en) | Fault detection method based on semi-supervised autoregressive dynamic hidden variable model | |
US6937966B1 (en) | System and method for on-line adaptive prediction using dynamic management of multiple sub-models | |
CN109917777B (en) | Fault detection method based on mixed multi-sampling rate probability principal component analysis model | |
CN106067034B (en) | Power distribution network load curve clustering method based on high-dimensional matrix characteristic root | |
CN113221282B (en) | Aero-engine service life prediction method based on Bayesian residual convolutional network | |
CN110580488B (en) | Multi-working-condition industrial monitoring method, device, equipment and medium based on dictionary learning | |
CN107977728A (en) | It is a kind of medium-term and long-term by hour Temperature prediction method based on BP artificial neural networks | |
CN110188015A (en) | A kind of host access relation abnormal behaviour self-adapting detecting device and its monitoring method | |
CN117273489A (en) | Photovoltaic state evaluation method and device | |
CN113487086B (en) | Method, device, computer equipment and medium for predicting residual service life of equipment | |
CN112711828A (en) | Maintenance and spare part supply joint optimization method under partially observable information | |
WO2020230422A1 (en) | Abnormality diagnosis device and method | |
US20230350402A1 (en) | Multi-task learning based rul predication method under sensor fault condition | |
CN115423005B (en) | Big data reconstruction method and device for combine harvester | |
CN115994617A (en) | Residual life prediction method and system combining cyclic neural network and filtering algorithm | |
CN107862394A (en) | A kind of equipment active maintenance support Synergistic method | |
CN116562120A (en) | RVE-based turbine engine system health condition assessment method and RVE-based turbine engine system health condition assessment device | |
CN113449914B (en) | Power system monitoring method and system | |
CN114818116A (en) | Aircraft engine failure mode identification and service life prediction method based on joint learning | |
CN114861759A (en) | Distributed training method of linear dynamic system model | |
CN112631890A (en) | Method for predicting cloud server resource performance based on LSTM-ACO model | |
EP3417169A1 (en) | A prognostics and health management model for predicting wind turbine oil filter wear level | |
Tinawi | Machine learning for time series anomaly detection | |
CN114298487B (en) | Reliability evaluation method and system for ship equipment system |
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 |