CN117611128A - Health management system and method considering single spare part guarantee - Google Patents
Health management system and method considering single spare part guarantee Download PDFInfo
- Publication number
- CN117611128A CN117611128A CN202311309433.0A CN202311309433A CN117611128A CN 117611128 A CN117611128 A CN 117611128A CN 202311309433 A CN202311309433 A CN 202311309433A CN 117611128 A CN117611128 A CN 117611128A
- Authority
- CN
- China
- Prior art keywords
- spare part
- time
- representing
- replacement
- acquisition
- 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
- 230000036541 health Effects 0.000 title claims abstract description 26
- 238000000034 method Methods 0.000 title abstract description 28
- 238000012544 monitoring process Methods 0.000 claims abstract description 55
- 230000001186 cumulative effect Effects 0.000 claims abstract description 38
- 238000005315 distribution function Methods 0.000 claims abstract description 32
- 230000000694 effects Effects 0.000 claims abstract description 25
- 230000015556 catabolic process Effects 0.000 claims abstract description 24
- 238000006731 degradation reaction Methods 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 22
- 230000007774 longterm Effects 0.000 claims description 37
- 238000012423 maintenance Methods 0.000 claims description 32
- 238000007726 management method Methods 0.000 claims description 24
- 230000003449 preventive effect Effects 0.000 claims description 23
- 238000009826 distribution Methods 0.000 claims description 17
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 9
- 238000005070 sampling Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 3
- 238000002360 preparation method Methods 0.000 claims 1
- 230000006870 function Effects 0.000 description 38
- 238000010586 diagram Methods 0.000 description 7
- 238000011160 research Methods 0.000 description 6
- 230000009286 beneficial effect Effects 0.000 description 5
- 238000013136 deep learning model Methods 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000036449 good health Effects 0.000 description 2
- 230000002035 prolonged effect Effects 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000013486 operation strategy Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- 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
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0283—Price estimation or determination
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Human Resources & Organizations (AREA)
- Operations Research (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a health management system and a method taking single spare part guarantee into consideration, belonging to the technical field of reliability engineering, wherein the system comprises a first processing subsystem, a second processing subsystem and a third processing subsystem, wherein the first processing subsystem is used for calculating PDF of the residual life of a part according to monitored single spare part degradation data, and calculating a cumulative distribution function and a reliability function of the residual life of the part; the second processing subsystem is used for determining optimal spare part acquisition and part replacement time through the multi-objective joint decision model based on the cumulative distribution function, the reliability function and the prediction result; and the third processing subsystem is used for judging whether the optimal spare part acquisition time is before the next state monitoring time, if so, installing and maintaining activities according to the predicted optimal spare part acquisition and part replacement time, otherwise, acquiring state monitoring data of the next monitoring time, and returning to the first processing subsystem. The invention solves the problems that the engineering equipment is inevitably subjected to health management and has overlong downtime and the usability of the system cannot be ensured.
Description
Technical Field
The invention belongs to the technical field of reliability engineering, and particularly relates to a health management system and method considering single spare part guarantee.
Background
In the 4.0 era of the industrial revolution, in parallel with this new revolution, concepts such as predictive maintenance (Predictive Maintenance, PM) have emerged, playing a key role in sustainable manufacturing and production systems by way of introducing digitized models of machine maintenance. Data extracted from the production process grows exponentially due to the development of sensing technology. PM can minimize machine downtime and associated costs, maximize machine lifecycle, and improve production quality and cadence. This approach is typically characterized by a very accurate workflow, starting with project understanding and data collection, and ending with a decision stage.
The first two parts surround a large amount of monitoring data generated in the running process of the complex system, and a plurality of prediction information can be obtained through analysis by a missing data generation and RUL prediction method. The prediction information includes probability density functions (Probability Density Function, PDF), cumulative distribution functions (Cumulative Distribution Function, CDF), reliability functions (Reliability Function, RF), and the like of the RUL. Based on RUL prediction information, maintenance staff can perform Condition-based maintenance (CBM) on a complex system, and reasonable health management activities are formulated in advance for problems possibly occurring in the running process so as to implement early prevention and remedy measures.
The existing research mainly aims at the condition of stock-free spare part guarantee, takes a plurality of conditions such as operation cost, availability and the like as constraint targets, and develops a combined decision of maintenance and spare part acquisition according to conditions. However, in actual situations, the complex system operation components have unexpected failures to increase the shutdown risk and maintenance cost, so that the system is separated from the health state, in order to reduce the replacement time of the components to the maximum extent, reduce the shutdown risk, improve the availability of the components, a spare part inventory system operation strategy is needed, and the operation service life of the system is prolonged.
Disclosure of Invention
Aiming at the defects in the prior art, the sensitive text characterization method based on contrast learning solves the problems that original semantics are distorted due to existing data enhancement and contrast learning training is low-efficient due to the existing method by utilizing a computer vision contrast frame, so that text characterization quality is affected.
In order to achieve the above purpose, the invention adopts the following technical scheme:
the scheme provides a health management system considering single spare part guarantee, which comprises:
the first processing subsystem is used for calculating PDF of the residual service life of the component according to the monitored degradation data of the single spare part, and calculating a cumulative distribution function and a reliability function of the residual service life of the component based on the PDF;
the second processing subsystem is used for determining optimal spare part acquisition and part replacement time through a multi-objective joint decision model according to the requirements of the actual working environment on the availability of the parts based on the cumulative distribution function, the reliability function and the prediction result;
and the third processing subsystem is used for judging whether the optimal spare part acquisition time is before the next state monitoring moment according to the determination result, if so, installing and maintaining activities according to the predicted optimal spare part acquisition and part replacement time, otherwise, acquiring state monitoring data of the next monitoring moment, and returning to the first processing subsystem.
The beneficial effects of the invention are as follows: based on the predicted remaining life information, the invention provides a single-component operation system of single-stock spare parts, and simultaneously considers a high availability and low cost spare part acquisition and spare part replacement joint decision method. The cost and the availability are simultaneously used as decision targets, spare part acquisition and component replacement time are used as decision variables, and the cost is minimized while the availability requirement is met through the trade-off problem between the cost and the availability of the constructed decision boundary, so that the optimal spare part acquisition and component replacement time of the multi-element degradation component is obtained. Finally, the effectiveness of the method is verified through the aeroengine data set, and the obtained decision result can help operation maintenance personnel to scientifically and reasonably arrange spare parts and replacement activities, so that the method has certain engineering application value.
Further, the expression of PDF of the remaining life of the component is as follows:
wherein,PDF, which indicates the remaining life of the component,/>Representing the total number of samples, +.>Represents the number of samples, +.>Probability density function PDF representing one-time sampling result based on optimal variation distribution, < >>Representing predicted component RUL results, +.>Representing input single spare part degradation monitoring data, < +.>Representing training set input data, +.>RUL tags representing the corresponding training set, +.>Representing a result of one sampling based on the optimal variation distribution.
The beneficial effects of the above-mentioned further scheme are: the PDF requesting to obtain the residual life is a precondition for health management maintenance activities, and the result of the residual life predicted by the deep learning method can be unified with the statistical method through the expression.
Still further, the cumulative distribution function is expressed as follows:
the expression of the reliability function is as follows:
wherein,representing cumulative distribution function->Representing a reliability function +.>Representation->Time-of-day predicted RUL results, < >>New single spare part degradation monitoring data representing inputs, < +.>Representing cumulative distribution density function, +.>Time of presentation->Representing the differential sign.
The beneficial effects of the above-mentioned further scheme are: the PDF based on the residual life can better serve the following multi-objective joint decision model by obtaining the information such as the cumulative distribution function, the reliability function and the like.
Still further, the second processing subsystem includes:
the first calculation module is used for respectively constructing a long-term average cost model and a long-term average availability model based on the cumulative distribution function and the reliability function;
the second calculation module is used for constructing a multi-objective joint decision model based on the long-term average cost model and the long-term average availability model;
the third calculation module is used for obtaining a prediction result of the residual service life of the single spare part;
and the fourth calculation module is used for determining optimal spare part acquisition and part replacement time through a multi-target joint decision model according to the requirement of the actual working environment on the availability of the parts based on the prediction result of the residual service life of the single spare part.
The beneficial effects of the above-mentioned further scheme are: the invention effectively determines the optimal spare part acquisition and part replacement time by constructing the multi-target combined decision model.
Still further, the expression of the long-term average cost model is as follows:
wherein,representing a long-term average cost model, +.>And->Representing the spare part acquisition and the part replacement time, respectively, that need to be optimized,/->Representing the cost of the desired cycle, +.>Indicating the desired period length, +.>Representing the total cost of acquisition of spare parts,/->Indicate->Time of day (I)>Indicating the length of time between spare part acquisition and spare part arrival,/->Representing the inventory-keeping costs per unit time,representing a reliability function +.>Representation->Time-of-day predicted RUL results, < >>Time to complete the failed replacement activity, +.>Indicates the time to complete the preventive replacement activity, +.>A cumulative distribution function representing the component replacement time,representing accumulationDistribution function (F)>Representing preventive replacement costs,/->Representing cumulative distribution density function, +.>Representing the cost of replacement ineffectively,/->Representing the cost of single monitoring,/->Representing spare part acquisition costs,/->Representing the total maintenance costs of inventory parts,/->Representing inventory costs->Representing replacement costs->Indicating the monitoring time point,/->Representing the cost of preventive replacement.
Still further, the expression of the long-term average availability model is as follows:
wherein,representing a long-term average availability model, +.>Indicating the expected run time of the component +.>A distribution function representing the acquisition time.
Still further, the expression of the multi-objective joint decision model is as follows:
wherein,representing an availability threshold.
Still further, the expression of the optimal spare part acquisition and part replacement time is as follows:
wherein,indicating optimal spare part acquisition and part replacement time, respectively,/->The lowest point and the highest point of the acquisition time and the replacement time are respectively represented.
The invention provides a health management method considering single spare part guarantee, which comprises the following steps:
s1, calculating PDF of the residual service life of the part according to monitored single spare part degradation data, and calculating a cumulative distribution function and a reliability function of the residual service life of the part;
s2, determining optimal spare part acquisition and part replacement time through a multi-objective joint decision model according to the requirements of the actual working environment on the availability of the parts based on the cumulative distribution function, the reliability function and the prediction result;
and S3, judging whether the optimal spare part acquisition time is before the next state monitoring time according to the determination result, if so, installing maintenance activities according to the predicted optimal spare part acquisition and part replacement time, otherwise, acquiring state monitoring data of the next monitoring time, and returning to the step S1.
The beneficial effects of the invention are as follows: the invention provides a spare part acquisition and replacement combined decision research method under single spare part guarantee, which aims at the actual situation that the health management of engineering equipment is inevitably faced with overlong downtime and the availability of a system cannot be guaranteed, and based on the predicted residual life distribution of an aeroengine, a multi-objective combined decision model which simultaneously considers cost and availability is established, the availability of components is maximized, the maintenance cost is reduced as much as possible, and the optimal spare part acquisition and replacement time is further determined, so that the problems that the health management of the engineering equipment is inevitably faced with overlong downtime and the availability of the system cannot be guaranteed are solved, and a good health management effect is obtained.
Drawings
Fig. 1 is a schematic diagram of a system structure according to the present invention.
Fig. 2 is a schematic diagram of a single spare part assurance system in the present embodiment.
Fig. 3 is a schematic diagram of two possible situations in the whole life cycle of the present embodiment.
Fig. 4 is a PDF schematic of the remaining life of different monitoring cycles in this embodiment.
Fig. 5 is a schematic diagram of the constraint condition in the present embodiment.
FIG. 6 is a single cost decision diagram of the present embodiment.
FIG. 7 is a diagram of the joint decision result in the present embodiment.
FIG. 8 is a schematic flow chart of the method of the present invention.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and all the inventions which make use of the inventive concept are protected by the spirit and scope of the present invention as defined and defined in the appended claims to those skilled in the art.
Before describing the present invention, the following parameters will be described:
PDF: a probability density function;
RUL: and the service life is prolonged.
Example 1
As shown in fig. 1, the present invention provides a health management system considering single spare part guarantee, comprising:
the first processing subsystem is used for calculating PDF of the residual service life of the component according to the monitored degradation data of the single spare part, and calculating a cumulative distribution function and a reliability function of the residual service life of the component based on the PDF;
the second processing subsystem is used for determining optimal spare part acquisition and part replacement time through a multi-objective joint decision model according to the requirements of the actual working environment on the availability of the parts based on the cumulative distribution function, the reliability function and the prediction result;
the second processing subsystem includes:
the first calculation module is used for respectively constructing a long-term average cost model and a long-term average availability model based on the cumulative distribution function and the reliability function;
the second calculation module is used for constructing a multi-objective joint decision model based on the long-term average cost model and the long-term average availability model;
the third calculation module is used for obtaining a prediction result of the residual service life of the single spare part;
the fourth calculation module is used for determining optimal spare part acquisition and part replacement time through a multi-target joint decision model according to the requirements of the actual working environment on the availability of the parts based on the prediction result of the residual service life of the single spare part;
and the third processing subsystem is used for judging whether the optimal spare part acquisition time is before the next state monitoring moment according to the determination result, if so, installing and maintaining activities according to the predicted optimal spare part acquisition and part replacement time, otherwise, acquiring state monitoring data of the next monitoring moment, and returning to the first processing subsystem.
In the embodiment, based on the RUL prediction model, the invention considers a single-stock spare part single-component operation system model, and in order to ensure the high availability operation of the whole system, the invention researches a spare part acquisition and replacement joint decision method considering a single spare part guarantee system based on RUL residual service life prediction information. And introducing a single spare part guarantee system of the single-stock spare part single-part operation system respectively, and optionally replacing and acquiring multiple conditions of joint decision of spare parts.
In this embodiment, the single spare part guarantee system, that is, the single spare part operation system, aims at a single large, valuable and single-inventory-capable replaceable service part, so as to reduce the system downtime and improve the operation efficiency of the whole system, and the stock quantity of the spare parts should be not less than one. As shown in fig. 2, in the actual operation period, two parts of the same model are firstly purchased at the same time, one part is used as an operation part, the other part is used as an inventory part, the inventory part is stored in a warehouse, and after the original service part fails suddenly, the replacement can be immediately carried out, so that the normal operation of the system is restored. The single component is in service, in an operating state, redundant in inventory, and in a storage state. And then, taking the spare parts of the stock parts in advance, namely, after the original parts are failed and replaced by one stock spare part, the stock can still be used for redundancy of one spare part, so that the spare part stock is free due to the fact that the new parts just replaced are failed suddenly. In particular, during the operation life cycle of one component, the cost of acquiring spare parts is comprehensively considered, and the minimum single spare part inventory system only needs to acquire one component in the later period except for acquiring two components for the first time. For convenience of distinction, the operation of two parts acquired first is referred to as original acquisition, and the operation of one part acquired later is referred to as spare part acquisition.
In this embodiment, under the single spare part guarantee system, the optional replacement is mainly divided into ineffective replacement and preventive replacement, and specifically includes the following steps:
ineffectiveness replacement: if inTime-of-day single spare part degradation monitoring data +.>Exceeding a preset failure threshold +.>Consider the component to be defective, i.e. at +.>Performing ineffective replacement at the moment, wherein the replacement cost is +.>。
Preventive replacement: if inTime-of-day single spare part degradation monitoring data +.>Does not exceed a predetermined failure threshold +.>Calculating to obtain->The remaining lifetime of the moment is +.>. Let the remaining lifetime threshold of the preventive replacement be +.>If->In order to prevent a component from failing in the next test interval, a preventive replacement of the component is performed at a replacement cost of +.>And->。
In this embodiment, both cost and availability are considered to obtain optimal spare part acquisition and part replacement times. Order theFor the current monitoring time, < >>The next monitoring time is the next monitoring time; />,/>Respectively->Spare part acquisition time and part replacement time predicted at the moment; />For a long-term average cost function->As a function of long-term average availability; decision variable is +.>And->An objective function of [ ]>,/>]I.e. a multi-objective joint decision model taking into account both cost and availability, provided that +.>. Since spare part acquisition is something that will happen in the future, +.>Is a basic requirement, < >>Because the corresponding component replacement activities can only be performed when spare parts are present. According to the update-reward theory, one life cycle of a component is represented by the time span between two replacements, including both a failed replacement and a preventive replacement. Based on this, the multi-objective joint decision model is expressed as:
(1)
in this strategy, the spare part acquisition and part replacement joint decision problem that considers both cost and availability based on predictive information includes cost functions, determination of availability functions, and how to optimize spare part acquisition and part replacement time. Before constructing the decision model, firstly, the whole life cycle of a single spare part inventory system part needs spare part acquisition for both failure replacement and preventive replacement, and the method is as follows:
based on these two acquisition schemes, two possible situations of the component throughout its life cycle are analyzed, as shown in particular in fig. 3.
Case 1: pre-fetching a failed replacement: the component has been acquired at the predicted acquisition time and has arrived, but fails before the predicted replacement time, the existing spare part is replaced immediately, i.e. the component is inIn between, the stock maintenance cost of stock spare parts and the stock cost of acquiring spare parts exist;
case 2: pre-acquisition of preventive replacement: the component has been acquired according to the predicted acquisition time and has arrived, the original component has not failed before the predicted replacement time, the existing spare part is replaced according to the predicted replacement time,i.e. the component is inIn between, there is a stock maintenance cost of stock spare parts and a stock cost of acquiring spare parts.
In this embodiment, in the PHM technical framework, a maintenance decision scheme of the degradation component is to be formulated, provided that a probability density function of the degradation component RUL is obtained. The existing research has proved that the traditional deep learning model using dropout is equivalent to a corresponding Bayesian deep learning model based on variation inference, and can be used for describing the uncertainty of a prediction result through the randomness of the weight, giving out the interval estimation of the prediction RUL, and obtaining the PDF of the RUL prediction result under the deep learning model as follows:
(2)
wherein,PDF, which indicates the remaining life of the component,/>Representing the total number of samples, +.>Represents the number of samples, +.>Probability density function PDF representing one-time sampling result based on optimal variation distribution, < >>Representing predicted component RUL results, +.>Representing input single spare part degradation monitoring data, < +.>Representing training set input data, +.>RUL tags representing the corresponding training set, +.>Representing a result of one sampling based on the optimal variation distribution.
After training the RUL prediction network through the training set, new single-component degradation monitoring data are input, and corresponding RUL prediction results can be obtained. For the convenience of the subsequent representation, the probability density function of the predicted RUL represented by equation (2) is reduced toWherein->Is->RUL, & gt for time prediction>Is the input monitoring data. From the probability density function of RUL, the cumulative distribution function of RUL can be further obtained>And reliability function->The method comprises the following steps:
(3)
(4)
wherein,representing cumulative distribution function->Representing a reliability function +.>Representation->Time-of-day predicted RUL results, < >>New single spare part degradation monitoring data representing inputs, < +.>Representing cumulative distribution density function, +.>Time of presentation->Representing the differential sign.
The combined decision model provided by the invention is analyzed based on RUL prediction information of the multi-element degradation component, so that a corresponding cumulative distribution function and reliability function can be further deduced as long as PDF of the residual service life is obtained, and a reasonable maintenance strategy is formulated on the basis, so that safe and stable operation of the component is ensured.
In this embodiment, according to the update-reward theory, a long-term average cost model and a long-term average availability model of the single spare part assurance system are respectively established.
In this embodiment, for four situations and engineering reality that may occur in the whole life cycle of the above component, the following assumptions are made for the multi-objective joint decision model provided by the present invention:
(1) Component condition monitoring is periodic, and the single condition monitoring cost is recorded asThe cost of preventive replacement is recorded asFailure is replaced byThe root is->And->;
(2) The condition monitoring behavior of the original service spare parts does not affect the remaining life of the component, and the degradation of the storage of redundant spare parts in the warehouse is negligible, and the component is restored as originally after preventive replacement or failed replacement. The time to complete the failed replacement activity isThe time for completing the preventive replacement activity is +.>;
(3) A period of time exists between the acquisition of the spare part and the arrival of the spare part, and the length of the period of time is generally considered to be fixed and is marked as L;
(4) Spare part acquisition cost isStock keeping cost per unit time is recorded as +.>;
(5) Inventory costs can be noted asThe stock time is recorded as->Replacement cost is marked as->Including the cost of failed replacement and the cost of preventive replacement.
In this embodiment, based on the model assumption described above, according to the update-reward theory, the long-term average cost model can be expressed as:
(5)
wherein,representing the cost of the desired cycle, +.>Indicating the desired period length, +.>And->The spare part acquisition and part replacement times are optimized respectively.
First step, deriving desired cycle costsIt can be expressed as:
(6)
spare part acquisition total cost: since the number of spare parts to be acquired is 1 in one life cycle, the total cost of acquiring spare parts is +.>。
Total maintenance cost of inventory components: in the case 1 and the case 2, long-time inventory component maintenance cost exists in the early stage, and the total inventory time is divided into two parts, wherein the inventory component inventory maintenance time is from the starting operation to the current monitoring time +.>Likewise, spare parts are acquiredInventory maintenance time of->The calculation is as follows:
(7)
to get the total cost of inventoryFirst of all the corresponding stock time needs to be deduced +.>Both case 1 and case 2 have inventory time.
(8)
Thus, the inventory cost is. Replacement cost->Including preventative replacement costs and failed replacement costs, may be expressed as:
(9)
to sum up, the expected cycle cost is obtainedThe method comprises the following steps:
(10)
second step, deriving desired period length. Two conditions that may occur throughout the life cycle of a componentIn case 1, the spare part has arrived, but the part fails before the predicted preventive replacement time, thus immediately making a failed replacement; case 2 spare parts have also arrived and the component has not failed within the predicted replacement time, then a preventive replacement is made at the predicted replacement time. Based on the above analysis, the desired period length +.>Can be expressed as:
(11)
combining equation (10) and equation (11) yields:
(12)
in this embodiment, the availability is an important index for measuring the actual running performance of the component, and is used to describe the probability that the component can normally run or the expected value of the time occupancy within a certain investigation time, and the construction form is as follows:
(13)
wherein,indicating the expected run time of the component +.>The desired cycle length is expressed as the sum of the component run time and the downtime. Among the four cases that may occur throughout the life cycle of a component, case 1 and case 2 have downtime due to out-of-stock and replacement activities, since spare parts have not arrived at the time of component failure. Case 1 has downtime due to failed replacement and case 2 has downtime due to preventive replacement.
First step, deriving the expected runtimeIt can be expressed as:
(14)
by combining equation (11) and equation (10), a long-term average availability model can be obtained as:
(15)
wherein,representing a long-term average cost model, +.>And->Representing the spare part acquisition and the part replacement time, respectively, that need to be optimized,/->Representing the cost of the desired cycle, +.>Indicating the desired period length, +.>Representing the total cost of acquisition of spare parts,/->Indicate->Time of day (I)>Indicating the length of time between spare part acquisition and spare part arrival,/->Representing the inventory-keeping costs per unit time,representing a reliability function +.>Representation->Time-of-day predicted RUL results, < >>Time to complete the failed replacement activity, +.>Indicates the time to complete the preventive replacement activity, +.>A cumulative distribution function representing the component replacement time,representing cumulative distribution function->Representing preventive replacement costs,/->Representing cumulative distribution density function, +.>Representing the cost of replacement ineffectively,/->Representing the cost of single monitoring,/->Representing spare part acquisition costs,/->Representing the total maintenance costs of inventory parts,/->Representing inventory costs->Representing replacement costs->Indicating the monitoring time point,/->Representing the cost of the preventive replacement,representing a long-term average availability model, +.>Indicating the expected run time of the component +.>A distribution function representing the acquisition time.
In this embodiment, the long-term average cost model and the long-term average availability model are respectively constructed, and a multi-objective optimization model is constructed on the basis of the long-term average cost model and the long-term average availability model, so that the cost is minimized while the availability requirement is met, and the optimal spare part acquisition and part replacement time is determined.
(16)
At any monitoring time pointThe built multi-objective joint decision model is shown in formula (16). In order to solve the multi-objective optimization problem, the invention constructs a decision boundary. Specifically, let long-term average availability +.>Meet the use efficiency of the component in engineering practice>On the basis of which the long-term average cost +.>The multi-objective joint decision model may be rewritten as shown in equation (17).
(17)
In the formula (17) of the present invention,for the availability threshold, it is usually set according to the actual operating requirements of the project. In order to ensure that the component meets the usability requirement, the spare part acquisition and component replacement times are within a certain range of values, i.e. +.>. In view of this, optimal spare part acquisition and part replacement time +.>The optimal spare part acquisition and part replacement time can be determined by calculation through the formula (18), namely, the cost is minimized while the usability requirement is met.
(18)
Wherein,indicating optimal spare part acquisition and part replacement time, respectively,/->The lowest point and the highest point of the acquisition time and the replacement time are respectively represented.
The present invention will be further described below.
Aiming at RUL prediction information of an aeroengine, a combined decision research method for acquiring and replacing spare parts under the guarantee of single spare parts is researched, and the effectiveness of the system provided by the invention is verified.
Based on the prediction result of a certain deep learning model, FIG. 4 shows the residual life PDF of the No. 76 engine in the test set, and the RUL prediction method has good prediction effect and certain actual reference value.
Based on the prediction result, setting parameters in the multi-objective joint decision model as follows respectivelyThe number of the elements is the number of the elements,yuan (Yuan) and (Fu)>Yuan (Yuan) and (Fu)>Yuan (Yuan) and (Fu)>Yuan (Yuan) and (Fu)>Meta, availability threshold->. Engine 76 has 176 monitoring cycles of CM data, assuming a 3 hour interval for each monitoring cycle, with a 72 hour pre-staging of the spare. The optimal spare part acquisition and part replacement time obtained by the single cost decision model at different monitoring time points and the combined decision model provided by the invention are shown in the following table 1, and the specific results are shown in the following table 1, wherein the decision results of two methods at different monitoring time points are shown in the table 1:
TABLE 1
As the component runs over time, the degradation characteristics continue to increase, and as the component approaches failure, its availability gradually decreases. In table 1, the optimal spare part acquisition and replacement times obtained by the multi-objective joint decision model are in advance of the single cost decision model, consistent with the actual operation and maintenance of the component.
To further illustrate the feasibility of the method of the present invention, the long term average costs are plotted in fig. 5 (a) and 5 (b), respectively, on the 135 th monitoring cycle, i.e., 405 hours, at the current monitoring time pointAnd long-term average availabilityThe functional images as a function of time are acquired and replaced by spare parts, and fig. 6 (a) and 6 (b) give two-dimensional schematic diagrams of single cost model decision results. As can be seen from fig. 6, at the 405 th hour of the current monitoring time point, if only the lowest cost is taken as a decision goal, the obtained optimal spare part acquisition and part replacement time are 442 th and 480 th hours, respectively.
The model provided by the invention takes cost and availability as decision targets simultaneously, and a multi-objective decision function based on prediction information is constructed, so that optimal spare part acquisition and part replacement time is obtained. In order to make the decision result more visual, taking the 405 th hour of the current monitoring time as an example, assume that the threshold requirement of the long-term average availability isAs described above, the value ranges of the spare part acquisition time and the replacement time can be obtained, respectively, and the minimum value of the cost can be obtained in this section. Fig. 7 shows the joint decision results, and it can be seen from fig. 7 that the optimal spare part acquisition and replacement times for the multi-objective joint decision model are 439 hours and 477 hours, respectively, because the degradation of the components will increase with the accumulation of running time, and the availability will gradually decrease. Spare part acquisition and replacement times may be advanced in order to meet availability threshold requirements. Compared with a decision model aiming at single cost at the lowest, the optimal spare part acquisition time and replacement obtained by the model provided by the inventionThe replacement time is all in advance, so that operation maintenance personnel can be helped to arrange spare part acquisition and replacement activities in time, and the efficient and stable completion of tasks of the parts is ensured.
Aiming at the actual demands of the components under the guarantee of single spare parts on the maintenance cost and the actual use efficiency in the actual operation and maintenance process, the invention provides a maintenance decision method based on prediction information, which mainly works as follows:
based on the predicted remaining life information, the invention provides a single-component operation system of single-stock spare parts, and simultaneously considers a high availability and low cost spare part acquisition and spare part replacement joint decision method. The cost and the availability are simultaneously used as decision targets, spare part acquisition and component replacement time are used as decision variables, and the cost is minimized while the availability requirement is met through the trade-off problem between the cost and the availability of the constructed decision boundary, so that the optimal spare part acquisition and component replacement time of the multi-element degradation component is obtained. Finally, the effectiveness of the method is verified through the aeroengine data set, and the obtained decision result can help operation maintenance personnel to scientifically and reasonably arrange spare parts and replacement activities, so that the method has certain engineering application value.
Example 2
As shown in fig. 8, the present invention provides a health management method considering single spare part guarantee, comprising the steps of:
s1, calculating PDF of the residual service life of the part according to monitored single spare part degradation data, and calculating a cumulative distribution function and a reliability function of the residual service life of the part;
s2, determining optimal spare part acquisition and part replacement time through a multi-objective joint decision model according to the requirements of the actual working environment on the availability of the parts based on the cumulative distribution function, the reliability function and the prediction result;
and S3, judging whether the optimal spare part acquisition time is before the next state monitoring time according to the determination result, if so, installing maintenance activities according to the predicted optimal spare part acquisition and part replacement time, otherwise, acquiring state monitoring data of the next monitoring time, and returning to the step S1.
The invention provides a spare part acquisition and replacement combined decision research method under single spare part guarantee, which aims at the actual situation that the health management of engineering equipment is inevitably faced with overlong downtime and the availability of a system cannot be guaranteed, and based on the predicted residual life distribution of an aeroengine, a multi-objective combined decision model which simultaneously considers cost and availability is established, the availability of components is maximized, the maintenance cost is reduced as much as possible, and the optimal spare part acquisition and replacement time is further determined, so that the problems that the health management of the engineering equipment is inevitably faced with overlong downtime and the availability of the system cannot be guaranteed are solved, and a good health management effect is obtained.
Claims (9)
1. A health management system considering single spare part guarantee, comprising:
the first processing subsystem is used for calculating PDF of the residual service life of the component according to the monitored degradation data of the single spare part, and calculating a cumulative distribution function and a reliability function of the residual service life of the component based on the PDF;
the second processing subsystem is used for determining optimal spare part acquisition and part replacement time through a multi-objective joint decision model according to the requirements of the actual working environment on the availability of the parts based on the cumulative distribution function, the reliability function and the prediction result;
and the third processing subsystem is used for judging whether the optimal spare part acquisition time is before the next state monitoring moment according to the determination result, if so, installing and maintaining activities according to the predicted optimal spare part acquisition and part replacement time, otherwise, acquiring state monitoring data of the next monitoring moment, and returning to the first processing subsystem.
2. The health management system considering single spare part assurance according to claim 1, wherein the expression of PDF of the remaining life of the part is as follows:
wherein,PDF, which represents the remaining life of the component, N represents the total number of samples, N represents the number of samples,probability density function PDF representing one-time sampling result based on optimal variation distribution, < >>Representing predicted component RUL results, +.>Representing the input single spare part degradation monitoring data, X represents the training set input data, Y represents the corresponding RUL label of the training set, and +.>Representing a result of one sampling based on the optimal variation distribution.
3. The health management system considering single spare part security as claimed in claim 1, wherein the expression of the cumulative distribution function is as follows:
the expression of the reliability function is as follows:
R(l k |X 1:k )=1-F(l k |X 1:k )
wherein F (l) k |X 1:k ) Represents a cumulative distribution function, R (l k |X 1:k ) Representing a reliability function, l k Representing t k Time-of-day prediction RUL results, X 1:k New preparation for representing inputPart degradation monitoring data, f (τ|X 1:k ) Represents a cumulative distribution density function, τ represents time, and d represents a differential sign.
4. The health management system under consideration of single spare parts assurance of claim 1, wherein the second processing subsystem comprises:
the first calculation module is used for respectively constructing a long-term average cost model and a long-term average availability model based on the cumulative distribution function and the reliability function;
the second calculation module is used for constructing a multi-objective joint decision model based on the long-term average cost model and the long-term average availability model;
the third calculation module is used for obtaining a prediction result of the residual service life of the single spare part;
and the fourth calculation module is used for determining optimal spare part acquisition and part replacement time through a multi-target joint decision model according to the requirement of the actual working environment on the availability of the parts based on the prediction result of the residual service life of the single spare part.
5. The health management system considering single spare part support according to claim 4, wherein the expression of the long-term average cost model is as follows:
wherein C is k (t o ,t p ) Representing a long-term average cost model, t o And t p Respectively are provided withRepresenting spare part acquisition and part replacement time requiring optimization, EU representing desired cycle cost, EV representing desired cycle length, C o Representing the total cost of acquisition of spare parts, t k Represents the t k The time, L, represents the length of time between spare part acquisition and spare part arrival, C h Representing inventory maintenance cost per unit time, R (l) k |X 1:k ) Representing a reliability function, l k Representing t k Time-of-day prediction RUL results, T f Indicating the time to complete the failed replacement activity, T p Indicating the time to complete the preventive replacement activity, F (t p |X 1:k ) Cumulative distribution function representing component replacement time, F (l k |X 1:k ) Representing cumulative distribution function, C p Represents the cost of preventive replacement, f (l) k |X 1:k ) Representing cumulative distribution density function, C f Representing cost of replacement by failure, C m Represents the cost of single monitoring, C o Representing spare part acquisition cost, C c Representing total maintenance costs of inventory components, C H Representing inventory costs, C R Represents replacement cost, k represents monitoring time point, C p Representing the cost of preventive replacement.
6. The health management system considering single spare part support according to claim 5, wherein the expression of the long-term average availability model is as follows:
wherein A is k (t o ,t p ) Represents a long-term average availability model, EO represents a component expected run time, F (t) o +L|X 1:k ) A distribution function representing the acquisition time.
7. The health management system considering single spare part security of claim 6, wherein the expression of the multi-objective joint decision model is as follows:
maximizing C k (t o ,t p )
Satisfy A k (t o ,t p )≥ζ
t k ≤t o ≤t p
Where ζ represents the availability threshold.
8. The health management system considering single spare part assurance according to claim 7, wherein the expression of the optimal spare part acquisition and part replacement time is as follows:
t o * ,t p * =argminC k (t o ,t p )
t o ,t p ∈[t min ,t max ]
wherein t is o * ,t p * Respectively representing optimal spare part acquisition and part replacement time, t min ,t max The lowest point and the highest point of the acquisition time and the replacement time are respectively represented.
9. A health management method for executing the health management system under consideration of single spare part assurance according to any one of claims 1 to 8, characterized by comprising the steps of:
s1, calculating PDF of the residual service life of the part according to monitored single spare part degradation data, and calculating a cumulative distribution function and a reliability function of the residual service life of the part;
s2, determining optimal spare part acquisition and part replacement time through a multi-objective joint decision model according to the requirements of the actual working environment on the availability of the parts based on the cumulative distribution function, the reliability function and the prediction result;
and S3, judging whether the optimal spare part acquisition time is before the next state monitoring time according to the determination result, if so, installing maintenance activities according to the predicted optimal spare part acquisition and part replacement time, otherwise, acquiring state monitoring data of the next monitoring time, and returning to the step S1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311309433.0A CN117611128A (en) | 2023-10-10 | 2023-10-10 | Health management system and method considering single spare part guarantee |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311309433.0A CN117611128A (en) | 2023-10-10 | 2023-10-10 | Health management system and method considering single spare part guarantee |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117611128A true CN117611128A (en) | 2024-02-27 |
Family
ID=89944880
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311309433.0A Pending CN117611128A (en) | 2023-10-10 | 2023-10-10 | Health management system and method considering single spare part guarantee |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117611128A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118735307A (en) * | 2024-09-02 | 2024-10-01 | 山东金精智能制造有限公司 | Multi-component equipment string co-production line opportunity maintenance method considering multiple correlations |
-
2023
- 2023-10-10 CN CN202311309433.0A patent/CN117611128A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118735307A (en) * | 2024-09-02 | 2024-10-01 | 山东金精智能制造有限公司 | Multi-component equipment string co-production line opportunity maintenance method considering multiple correlations |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20210397501A1 (en) | System and method for unsupervised prediction of machine failures | |
Lu et al. | Predictive condition‐based maintenance for continuously deteriorating systems | |
US11669083B2 (en) | System and method for proactive repair of sub optimal operation of a machine | |
CN109828869B (en) | Method, device and storage medium for predicting hard disk fault occurrence time | |
US10732618B2 (en) | Machine health monitoring, failure detection and prediction using non-parametric data | |
US11055450B2 (en) | Industrial asset health model update | |
US20180348747A1 (en) | System and method for unsupervised root cause analysis of machine failures | |
JP6865189B2 (en) | Failure probability evaluation system and method | |
US20150205657A1 (en) | Predicting failure of a storage device | |
Siraskar et al. | Reinforcement learning for predictive maintenance: A systematic technical review | |
CN113196311A (en) | System and method for identifying and predicting abnormal sensing behavior patterns of a machine | |
Pandey et al. | A methodology for simultaneous optimisation of design parameters for the preventive maintenance and quality policy incorporating Taguchi loss function | |
US11150631B2 (en) | Predictive maintenance utilizing supervised sequence rule mining | |
Golmakani | Optimal age-based inspection scheme for condition-based maintenance using A* search algorithm | |
Levitin et al. | Optimal multiple replacement and maintenance scheduling in two-unit systems | |
CN117611128A (en) | Health management system and method considering single spare part guarantee | |
Zhebka et al. | Methodology for Predicting Failures in a Smart Home based on Machine Learning Methods | |
EP3586233A1 (en) | Methods and systems for problem-alert aggregation | |
CN112534371A (en) | Optimizing accuracy of machine learning algorithms for monitoring operation of industrial machines | |
US20140361978A1 (en) | Portable computer monitoring | |
Xu et al. | Integrated system health management-oriented maintenance decision-making for multi-state system based on data mining | |
Xiao et al. | Prognostics and Health Management for Maintenance-Dependent Processes | |
CN109933890A (en) | A kind of product comprehensive maintenance method and apparatus | |
Ahmed et al. | Application of predictive maintenance in industry 4.0: A use-case study for datacenters | |
Beduschi et al. | Optimizing rotating equipment maintenance through machine learning algorithm |
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 |