CN106600072A - Maintenance decision optimization algorithm based on failure prediction - Google Patents

Abstract

The invention relates to the field of equipment maintainability and testability, and specifically relates to a maintenance decision optimization algorithm based on failure prediction in order to at least solve the problem that the existing maintenance decision optimization method is of low precision. The optimization algorithm comprises the following steps: getting the maintenance time and maintenance cost corresponding to a target unit; generating a life sample of the target unit and a life sample of a canary unit; recording the number of times of preventative replacement and maintenance and the number of times of reparative replacement and maintenance; calculating the total available time of the system, the total operation time of a target system, the total downtime of the system, and the total operation cost of the system; calculating three maintenance utility indexes of cost per unit time, average use availability and average cost-effectiveness; and calculating an optimal utility index and the damage parameters of the corresponding canary unit. By using the optimization algorithm, the health status of a chip-level or board-level electronic component or system can be monitored effectively, a warning can be issued when there is a fault, and a reasonable means of maintenance can be decided to maximize the efficiency of maintenance. The method is of higher precision.

Description

A kind of maintenance decision optimized algorithm based on failure predication

Technical field

The present invention relates to equipment Maintainability and testability field, more particularly to a kind of maintenance decision based on failure predication is excellent Change algorithm.

Background technology

With the development of fault diagnosis and fault prediction technology, equipment Maintenance Policy breaches the restriction of technology, is no longer limited to Simple correction maintenance and periodic maintenance, the maintenance of low-cost high-efficiency state base and prospective maintenance progress into engineer applied neck Domain.Based on failure predication information carry out maintenance decision can effectively improve equipment use availability, reduce LCC.

When some current maintenance Strategy Models are keeped in repair using the overall static reliability feature of application apparatus come decision-making Machine, and only account for the situation of predicting unit and object element with distribution.Such as in the pre- megacell of some radar model applications To predict radar system key component, the maintenance of important parts change part opportunity, pre- megacell using with manufacturer production obedience equally therefore The single-piece of barrier probability density function, its failure mechanism, fault mode are consistent as far as possible with measured piece, and in quality, encapsulation Decline compared with unit under test in terms of with material etc. manufacturing process；In voltage, electric current, operating temperature, humidity, Oscillating Coefficients Etc. working stress and environmental stress aspect, also there is no drop volume during omen unit load uses, and the design of pre- megacell is predicting Distance no more than 200h be target component, according to actual needs at full capacity or over loading with system under test (SUT) run.It is external typical In document, it puies forward the characteristic of maintenance policy only from the angle analysis of long-run cost rate for the research of Pecht, does not consider system The time of shutdown, for military hardware is also very important index and constraint using availability, the optimization for needing multiparameter is tieed up Repair strategy.

In the past, due to technological means and the limitation of cognition, it is generally recognized that the life-span of electronic product obeys exponential, There is no the loss phase, thus without life requirements, and increasing engineering practice proves that the product such as electronic devices and components there is also Loss problem, its life-span not necessarily obeys exponential, and with products such as machinery, structures life consumption problem is equally there is also.Pass The static prediction model of system cannot meet the needs of engineering practice, and these are used for the static reliability feature that opportunity is keeped in repair in decision-making Parameter is not affected by actual degenerative process, also cannot be distinguished by the different degenerative characters between equipment individuality, is not suitable for predictability dimension The needs of decision optimization are repaiied, causes the degree of accuracy of maintenance key point decision-making relatively low.

Modern field maintenance system for current constantly popularization and application and health management system arranged, needs pin in radar engineering The maintenance for having system-level preventative maintenance to drive with the pre- megacell without system-level preventative maintenance is set up to electronic system LRM level Model.The use of availability is very important binding target for device systems.For the direct employing correction maintenance of radar system Strategy, age change the decision method that the maintenance policy of part is not economically optimum, and exponential distribution, normal distribution The life characteristic and practical situation for portraying LRM does not meet again, it is therefore desirable to which the failure predication design based on pre- megacell is adapted to thunder Up to the maintenance decision algorithm of system.

The content of the invention

It is an object of the invention to provide a kind of maintenance decision optimized algorithm based on failure predication, at least to solve at present The relatively low problem of the degree of accuracy of maintenance key point decision-making.

The technical scheme is that：

A kind of maintenance decision optimized algorithm based on failure predication, including：

Step one, the prosthetic alternate maintenance time and preventative alternate maintenance time that obtain object element, and calculating are repaiied The installation and operating cost of renaturation alternate maintenance cost, preventative alternate maintenance cost with pre- megacell；

Step 2, the life-span according to the object element and the feature in the life-span of pre- megacell are generated Wherein,For the life-span sample of object element described in system operation,The longevity of pre- megacell described in system operation Life sample, Z is the life-span of the object element, and X is the life-span of the pre- megacell；

Step 3, whenWhen, preventative alternate maintenance is carried out to system, the single update cycle is Alternate maintenance cost is installation and operating cost (i.e. C of the preventative alternate maintenance cost plus the pre- megacell_p+C_m), The alternate maintenance time of needs is preventative alternate maintenance time (the i.e. T_p), record the frequency n of preventative alternate maintenance_p(i) =1；

WhenWhen, prosthetic alternate maintenance is carried out to system, the single update cycle isAlternate maintenance Cost is installation and operating cost (i.e. C of the prosthetic alternate maintenance cost plus the pre- megacell_c+C_m), the part that changes of needs is tieed up The time is repaiied for prosthetic alternate maintenance time (i.e. T_c), record the frequency n of prosthetic alternate maintenance_c(i)=1；

Step 4, according to following relational expression (1), (2), (3), (4) the total pot life of accumulation computing system, target system respectively The totle drilling cost of the total time, system total down-time and system operation of system operation：

Wherein, T_cFor the prosthetic alternate maintenance time；T_pFor preventative alternate maintenance time, C_cFor prosthetic alternate maintenance Cost；C_pFor preventative alternate maintenance cost；C_mInstallation and operating cost for pre- megacell；

Step 5, according to following relational expression (5), (6), (7) respectively unit of account time cost, averagely using availability with And three maintenance utility indexs of average efficiency-cost ratio：

Step 6, the impairment parameter for adjusting the pre- megacell, three maintenance utility indexs in generation step five (i.e. EC, EA, EB) sequence, and according to equation below (8), (9), (10) therefrom solve optimum utility index and its corresponding omen The impairment parameter of unit：

Optionally, in the step 2, also include：

When the life-span independence of the pre- megacell and object element, according to life-span and the pre- megacell of the object element Life-span feature generate respectively the object element life-span sample and the pre- megacell life-span sample (i.e. system fortune Respective life-span sample in rowWith)；

When the pre- megacell is related to the life-span of object element, the object element of monitored electronic system is firstly generated Life-span sample, then according to the life-span dependence of pre- megacell and goal systems, generate the life-span sample of the pre- megacell This is (i.e.)。

Invention effect：

The present invention the maintenance decision optimized algorithm based on failure predication, with long-run cost rate, averagely using availability, Average efficiency-cost ratio obtains pre- megacell prediction optimization parameter, based on omen monotechnics as maintenance key point target by algorithm Predictive maintenance decision optimization can effectively monitor the health status of chip-scale or plate level electronic unit or system, and early warning failure With the rational maintenance mode of decision, the maximization of maintenance efficiency is reached, degree of accuracy is higher.

Description of the drawings

Fig. 1 is omen cell failure prediction of the present invention based on different distributions in the maintenance decision optimized algorithm of failure predication Model；

Fig. 2 is general flow chart of the present invention based on the maintenance decision optimized algorithm of failure predication.

Specific embodiment

To make purpose, technical scheme and the advantage of present invention enforcement clearer, below in conjunction with the embodiment of the present invention Accompanying drawing, the technical scheme in the embodiment of the present invention is further described in more detail.In the accompanying drawings, identical from start to finish or class As label represent same or similar element or the element with same or like function.Described embodiment is the present invention A part of embodiment, rather than the embodiment of whole.It is exemplary below with reference to the embodiment of Description of Drawings, it is intended to use It is of the invention in explaining, and be not considered as limiting the invention.Based on the embodiment in the present invention, ordinary skill people The every other embodiment that member is obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.Under Face combines accompanying drawing and embodiments of the invention is described in detail.

In describing the invention, it is to be understood that term " " center ", " longitudinal direction ", " horizontal ", "front", "rear", The orientation or position relationship of the instruction such as "left", "right", " vertical ", " level ", " top ", " bottom ", " interior ", " outward " is based on accompanying drawing institute The orientation for showing or position relationship, are for only for ease of the description present invention and simplify description, rather than indicate or imply the dress of indication Put or element with specific orientation, with specific azimuth configuration and operation, therefore it is not intended that must be protected to the present invention The restriction of scope.

1 and Fig. 2 the present invention is done further in detail based on the maintenance decision optimized algorithm of failure predication below in conjunction with the accompanying drawings Explanation.

The invention provides a kind of maintenance decision optimized algorithm based on failure predication, including：

Step one, object element determine after, obtain object element prosthetic alternate maintenance time T_cChange with preventative Part maintenance time T_p, and calculate (generally estimate, be that the expense used by emulation experiment simulation assemblage etc. is obtained) prosthetic Alternate maintenance cost C_c, preventative alternate maintenance cost C_pWith pre- megacell (canary units, for predict object element what When it is bad, belong to prior art) installation and operating cost C_m；Wherein, object element can be monitored electronic system or portion Part.

Step 2, the life-span according to object element and the feature in the life-span of pre- megacell are generatedWherein,For the life-span sample of object element in system operation,The life-span sample of pre- megacell in system operation, Z is target The life-span of unit, X is the life-span of pre- megacell.

Further, in this step, in addition it is also necessary to judge the relation between the life-span of pre- megacell and the life-span of object element, Difference is as follows：

1), when the life-span independence of pre- megacell and object element, according to the longevity of life-span of object element and pre- megacell The feature of life generates respectively life-span sample (respective longevity i.e. in system operation of the life-span sample of object element and pre- megacell Life sampleWith)。

If it should be noted that pre- megacell and object element (electronic system or part) are using different material or adopt Manufactured with different technique, then it is assumed that pre- megacell is separate with the life-span of object element, and its joint probability density function can To be expressed as f (x, z)=f_X(x)f_Z(z).Define Prediction distance ε=E (X)-E (Z) to weigh the accuracy of pre- megacell, be used in combination The standard deviation sigma of X_sThe degree of accuracy of pre- megacell is weighed, wherein E (X) and E (Z) is respectively the expectation of pre- megacell and goal systems Life-span.

When life expectancy E (X) of pre- megacell is less than life expectancy E (Z) of goal systems, i.e. during ε ＜ 0, pre- megacell Can break down earlier than goal systems and realize fault pre-alarming.Due to omen cell life and the randomness sheet in goal systems life-span Matter, still suffers from a small amount of prosthetic alternate maintenance situation and occurs.When the degree of accuracy in pre- megacell life-span itself increases, can be corresponding Ground reduces the probability of prosthetic alternate maintenance, and increases the probability of preventative alternate maintenance, to improve the availability of system.The limit Situation is exactly E (X) ＜ E (Z) and σ_s→ 0, the now life-span distribution of pre- megacell is changed into a delta-function, and maintenance policy is equivalent to Age changes part strategy, and wherein E (X) is equivalent to and changes the part cycle；As E (X) >=E (Z) and σ_sWhen → 0, maintenance policy is equivalent to afterwards Maintenance policy.

For long-run cost rate, averagely using different system maintenance optimization aims such as availability, average efficiency-cost ratios, point Not Cun an optimum pre- megacell Prediction distance and precision of prediction, be now capable of achieving optimum predictive maintenance.

2), when pre- megacell is related to the life-span of object element, the object element of monitored electronic system is firstly generated Life-span sample, then according to the life-span dependence of pre- megacell and goal systems, generate the life-span sample of pre- megacell (i.e.)。

It should be noted that for pre- megacell and object element are manufactured using identical material and using identical technique Situation, it is believed that the life-span of pre- megacell is related to the life-span of object element；There are close feature, pre- megacell energy between the two Enough running statuses for more accurately tracking object element.In this case, both joint probability density functions can be expressed as f (x, z)=f_X|Z(x|z)f_Z(z), f_X|Z(x | z) determined with the intrinsic valuation characteristic of electronic system (or part) by pre- megacell It is fixed.Define the accumulated damage factor of pre- megacell0 average normal random variable γ independently of Z is defined simultaneously, i.e.,To weigh the degree of accuracy of pre- megacell.As ω ＜ 1, life expectancy E (X) of pre- megacell is less than monitored Life expectancy E (Z) of system, therefore pre- megacell can send warning before physical fault occurs, the less explanation omen of the value The unit warning time in advance is longer, but the waste in part effecting surplus life-span is also bigger.Prediction standard difference σ_sFor describing The noise for predicting the outcome, the value is less, illustrates that the warning of pre- megacell is more accurate.In σ_sIt is determined that in the case of, there is a ω ＜ 1 so that maintenance effect can realize optimization；As ω → 1 and σ_sWhen → 0, pre- megacell being capable of the monitored system of accurate tracking The state of system and alarm signal is sent earlier than monitored thrashing, realize the predictive maintenance of deactivation system.

If step 3, changing piece number for N, the simulation times for setting Monte Carlo are keeped in repair certainly as N, now Plan is as follows：

1), whenWhen, preventative alternate maintenance is carried out to system, the single update cycle isChange part Maintenance cost is installation and operating cost (i.e. C of the preventative alternate maintenance cost plus pre- megacell_p+C_m), the part that changes of needs is tieed up The time is repaiied for preventative alternate maintenance time (i.e. T_p), record the frequency n of preventative alternate maintenance_p(i)=1；

2), whenWhen, prosthetic alternate maintenance is carried out to system, the single update cycle isChange part Maintenance cost is installation and operating cost (i.e. C of the prosthetic alternate maintenance cost plus pre- megacell_c+C_m), the part that changes of needs is tieed up The time is repaiied for prosthetic alternate maintenance time (i.e. T_c), record the frequency n of prosthetic alternate maintenance_c(i)=1.

Step 4, according to following relational expression (1), (2), (3), (4) the total pot life T of accumulation computing system respectively_u, target Total time T of system operation, system total down-time T_dAnd totle drilling cost C (T) of system operation：

Wherein, T_cFor the prosthetic alternate maintenance time；T_pFor preventative alternate maintenance time, C_cFor prosthetic alternate maintenance Cost；C_pFor preventative alternate maintenance cost；C_mInstallation and operating cost for pre- megacell.

Step 5, according to following relational expression (5), (6), (7) respectively unit of account time cost E (C), average using available Three maintenance utility indexs of degree E (A) and average efficiency-cost ratio E (B)：

Step 6, the impairment parameter for adjusting pre- megacell, three maintenances utility index (i.e. E (C), E in generation step five (A), E (B)) sequence, and therefrom solve optimum (i.e. minE (C), maxE (A), maxE according to equation below (8), (9), (10) (B) utility index) and its impairment parameter of corresponding pre- megacell：

(ε_C,γ_C)=arg min E (C)；

(ε_B, γ_B)=arg max E (B)；

(ε_A, γ_A)=arg max E (A).

Finally, under different applied environments, according to different use demand selection unit time costs, average use can One or several parameters in expenditure, average efficiency-cost ratio are obtained as the target of maintenance decision according to aggregation type in step 6 The parameter value (accuracy, degree of accuracy, the accumulated damage factor) of correspondence optimum maintenance decision target, with reference to the ginseng of actual pre- megacell Numerical value is modified and adjusts, and makes the maintenance decision for optimizing to greatest extent.

The maintenance decision optimized algorithm based on failure predication of the present invention, it is not simple in the MTTR and to keep in repair into This two dimension discussion and correction maintenance strategy, age change the efficiency of the maintenance policy of part, and with long-run cost rate, averagely make With availability, average efficiency-cost ratio as maintenance key point target, pre- megacell prediction optimization parameter is obtained by algorithm, based on omen The predictive maintenance decision optimization of monotechnics can effectively monitor the health status of chip-scale or plate level electronic unit or system, and Early warning failure and the rational maintenance mode of decision, reach the maximization of maintenance efficiency, and degree of accuracy is higher.

In sum, the maintenance decision optimized algorithm based on failure predication of the invention, mainly by changing prediction part dimension Repair the analysis of strategy and omen element characteristics (accuracy and precision) to predictive maintenance influential effect and devise maintenance decision The flow process of optimization, use and maintenance decision making algorithm calculates optimum prediction parameter.For long-run cost rate, averagely using availability, The different system maintenance optimization aims such as average efficiency-cost ratio, are respectively present the pre- megacell Prediction distance and prediction essence of an optimum Degree, is now capable of achieving optimum predictive maintenance.By the way that it is calculated object element and omen cell life sample extraction update week Phase and renewal cost, obtain total time and the system operation of goal systems operation on the premise of system total down-time is considered Totle drilling cost, obtains being predicted factor adjustment using Monte Carlo method iteration after system maintenance optimization aim, final to obtain Better than the maintenance decision parameters optimization based on failure predication that the age changes part strategy.

The above, the only specific embodiment of the present invention, but protection scope of the present invention is not limited thereto, any Those familiar with the art the invention discloses technical scope in, the change or replacement that can be readily occurred in, all should It is included within the scope of the present invention.Therefore, protection scope of the present invention should be with the scope of the claims It is accurate.

Claims (2)

1. a kind of maintenance decision optimized algorithm based on failure predication, it is characterised in that include：

Step one, the prosthetic alternate maintenance time of acquisition object element and preventative alternate maintenance time, and calculate prosthetic The installation and operating cost of alternate maintenance cost, preventative alternate maintenance cost with pre- megacell；

Step 2, the life-span according to the object element and the feature in the life-span of pre- megacell are generatedWherein,For the life-span sample of object element described in system operation,The life-span sample of pre- megacell described in system operation This, Z is the life-span of the object element, and X is the life-span of the pre- megacell；

Step 3, whenWhen, preventative alternate maintenance is carried out to system, the single update cycle isChange part dimension Accomplish installation and the operating cost for originally adding the pre- megacell for the preventative alternate maintenance cost, the alternate maintenance of needs Time is the preventative alternate maintenance time, records the frequency n of preventative alternate maintenance_p(i)=1；

WhenWhen, prosthetic alternate maintenance is carried out to system, the single update cycle isAlternate maintenance cost It is installation and operating cost of the prosthetic alternate maintenance cost plus the pre- megacell, the alternate maintenance time of needs is reparation Property the alternate maintenance time, record prosthetic alternate maintenance frequency n_c(i)=1；

Step 4, according to following relational expression (1), (2), (3), (4) respectively the total pot life of accumulation computing system, goal systems fortune The totle drilling cost of capable total time, system total down-time and system operation：

$T_u=\underset{i=1}{\overset{N}{\Σ}}min\left(t_X^f\right(i),t_Z^f(i\left)\right)\mathrm{......}\left(1\right);$

$T_d=T_p\underset{i=1}{\overset{N}{\Σ}}{n}_p\left(i\right)+T_c\underset{i=1}{\overset{N}{\Σ}}{n}_c\left(i\right)\mathrm{...}\left(2\right);$

$T=T_u+T_d=\underset{i=1}{\overset{N}{\Σ}}\min \left(t_X^f\right(i),t_Z^f(i\left)\right)+T_p\underset{i=1}{\overset{N}{\Σ}}{n}_p\left(i\right)+T_c\underset{i=1}{\overset{N}{\Σ}}{n}_c\left(i\right)\mathrm{......}\left(3\right);$

$C\left(T\right)=(C_p+C_m)\underset{i=1}{\overset{N}{\Σ}}{n}_p\left(i\right)+(C_c+C_m)\underset{i=1}{\overset{N}{\Σ}}{n}_c\left(i\right)\mathrm{......}\left(4\right);$

Wherein, T_cFor the prosthetic alternate maintenance time；T_pFor preventative alternate maintenance time, C_cFor prosthetic alternate maintenance cost； C_pFor preventative alternate maintenance cost；C_mInstallation and operating cost for pre- megacell；

Step 5, according to following relational expression (5), (6), (7) respectively unit of account time cost, averagely using availability and flat Three of efficiency-cost ratio keep in repair utility indexs：

$E\left(C\right)=\frac{C\left(T\right)}{T}\mathrm{......}\left(5\right);$

$E\left(A\right)=\frac{T_u}{T}\mathrm{......}\left(6\right);$

$E\left(B\right)=\frac{E\left(A\right)}{E\left(C\right)}=\frac{T_u}{C\left(T\right)}\mathrm{......}\left(7\right);$

Step 6, the impairment parameter for adjusting the pre- megacell, the sequence of three maintenance utility indexs in generation step five, and root The utility index of optimum and its impairment parameter of corresponding pre- megacell are therefrom solved according to equation below (8), (9), (10)：

$({\mathrm{\ω}}_C,{\mathrm{\σ}}_C)={\arg }_{\mathrm{\ω},{\mathrm{\σ}}_s}\min E\left(C\right)\mathrm{......}\left(8\right);$

$({\mathrm{\ω}}_B,{\mathrm{\σ}}_B)={\arg }_{\mathrm{\ω},{\mathrm{\σ}}_s}\max E\left(B\right)\mathrm{......}\left(9\right);$

$({\mathrm{\ω}}_A,{\mathrm{\σ}}_A)={\arg }_{\mathrm{\ω},{\mathrm{\σ}}_s}\max E\left(A\right)\mathrm{......}\left(10\right).$

2. the maintenance decision optimized algorithm based on failure predication according to claim 1, it is characterised in that in the step In two, also include：

When the life-span independence of the pre- megacell and object element, according to the longevity of life-span of the object element and pre- megacell The feature of life generates respectively the life-span sample of the life-span sample of the object element and the pre- megacell；

When the pre- megacell is related to the life-span of object element, the longevity of the object element of monitored electronic system is firstly generated Life sample, then according to pre- megacell and the life-span dependence of goal systems, generates the life-span sample of the pre- megacell.