CN106066252A - The health state evaluation method of equipment subsystem level destroyed by a kind of dangerous materials - Google Patents

Abstract

The health state evaluation method of equipment subsystem level destroyed by a kind of dangerous materials of the present invention, including following step: step one: health status classification；Step 2: the health degree of all parts level equipment that computing subsystem comprises；Step 3: the holistic health degree that the component-level cluster tool that computing subsystem comprises couples；Step 4: calculate the profust reliability characterized by subsystem irrespective of size status data；Step 5: the holistic health degree of subsystem irrespective of size calculates；Step 6: according to health degree difference numerical range, carry out health state evaluation.The method considers health degree, the connected mode of each parts that subsystem comprises parts, and the health degree of the status data of subsystem, and utilizes profust reliability preferably to merge the change impact of each state.Reflect health degree by profust reliability, the slight change of the holistic health state of subsystem can be described, contribute to realizing, overall monitor real-time to equipment health status and early warning, method highly versatile, it is simple to Project Realization.

Description

The health state evaluation method of equipment subsystem level destroyed by a kind of dangerous materials

[technical field]

The Combined health state that the present invention relates to the Fusion Strain data that equipment subsystem level destroyed by a kind of dangerous materials is commented Estimate method.The method is mainly based upon profust reliability, merged health degree that subsystem comprised each parts and subsystem from The health state evaluation that the profust reliability of body status data is carried out, solves data small sample, equipment inherent parameters and operation The health state evaluation problem of the Multi-information acquisition such as state parameter.Belong to complex equipment health state evaluation field.

[background technology]

All kinds of dangerous materials of technical finesse such as equipment utilization high temperature incineration and tail gas clean-up destroyed by dangerous materials, including: various have Poison chemicals, smoke agent, the sudden and violent Charge of ammunitions of control etc. containing stimulant.The structure destroying equipment is complicated, with high content of technology.Destroy and make Industry is in highly dangerous state, high to its serviceability rate, security requirement, comments in the urgent need to destruction equipment is carried out health status Estimate, contingent fault is judged in advance and reports to the police, it is to avoid accident occurs.

Complex device or system are divided by architecture feature, are commonly divided into component-level-subsystem irrespective of size-system-level three Level.Subsystem is alternatively referred to as subsystem, the equipment of multiple component-level of next stage combine.Destroying this class stream of equipment In formula system, subsystem, in addition to including the equipment of component-level of next stage, yet forms both with newly-increased specific function New " equipment ", this new " equipment ", namely the function of subsystem self also has some parameters representing self performance. These parameters are specific to the component-level equipment self that subsystem itself is specific rather than subsystem comprises, subsystem The measurable parameters of self distinctive what's new is referred to as the running state parameter of subsystem.

At present, the method for system health state estimation has a lot, and common method has: modelling, level and fuzzy evaluation Method, artificial neural network method, method based on Bayesian network etc..These methods are respectively arranged with its advantage and subject range.Typically comment When estimating subsystem health status, the most only consider each parts and the structure problem thereof of composition subsystem.Burning of destruction equipment Each unit construction that system is comprised forms the most again the New function having in addition to component function, composition complexity.For by The subsystem that multiple parts are formed creates new state parameter, and single appraisal procedure is difficult to preferably solve the strong of this kind of system Health state estimation.For the most preferably assessing the Multi-information acquisition such as part of appliance inherent parameters and subsystem job state parameter Health state evaluation, becomes the problem that need to solve.

It is theoretical that Beijing Institute of Aeronautics Cai Kaiyuan professor proposes profust reliability in the nineties in 20th century, is by classical reliability theory Probability assume and fringe in Fuzzy Reliability Theory assumes the theory that combines, be under the jurisdiction of Fuzzy Reliability Theory. This theory has been successfully applied to ACT proof machine, and the different system such as lithium battery and computer communication network can be preferably to equipment Various states analysis provide fundamental basis, meet the demand of equipment health state evaluation.Profust reliability index mainly has The profust reliability of system, average blur fault time, rate mould fault rate etc., reflect the characteristic of the performance and reliability of system, Therefore it is considered as and performance-relevant evaluation index.

We consider measured state value, the parts comprised and the phase that the newly-increased menu of reflection subsystem irrespective of size reveals Mutual relation, based on the multimode that profust reliability is theoretical, proposes to destroy the combination health state evaluation method of equipment subsystem, solves The health state evaluation problem of the Multi-information acquisition such as data of having determined small sample, equipment inherent parameters and job state parameter.

[summary of the invention]

1, purpose: it is an object of the invention to provide a kind of dangerous materials and destroy the health state evaluation side of equipment subsystem level Method, the method considers health status, the connected mode of each parts that subsystem comprises parts, and the distinctive shape of subsystem The profust reliability of state data, solves dangerous materials and destroys Multi-information acquisition strong of the complex equipment subsystem such as equipment subsystem Health state estimation problem, has preferably merged component-level equipment and the situation of the running status of subsystem own, can realize distinguishing son The slight change of the holistic health state of system, the health state evaluation for such complex equipment subsystem irrespective of size provides a kind of new Method.

2, technical scheme: the present invention designs a kind of dangerous materials and destroys the health state evaluation method of equipment subsystem level.Its Design philosophy is: first from system architecture angular divisions component-level and subsystem irrespective of size, then first each group of computing subsystem Become the health degree of parts；Then the health of the connecting mode calculating unit level set of all parts level equipment is comprised according to subsystem Degree；Secondly the profust reliability of the self-operating state of computing subsystem level is reflected as the health degree of subsystem state；Finally, Combined by the health degree of component-level set and the profust reliability of subsystem irrespective of size self-operating state, COMPREHENSIVE CALCULATING subsystem Overall profust reliability.The health status of subsystem is reflected by profust reliability.

Below in conjunction with the step in flow chart 2, specifically introduce the technical scheme of the method.

Step one: health status classification

To some degree, the health status of system refers to that system equipment keeps the ability of certain reliability level, is System reliability in a state of use is maintained at the confidence level of certain limit (on the premise of ensureing that system completes predetermined function). Health degree so can be used to carry out the health status of quantitative measurement system, equipment or parts.Assuming that R represents the health of system Degree, i.e. the probability of health status, span is [0,1].As R=0, represent that system is the most unhealthy, at equipment or parts In severe fault condition；As R=1, representing that system is the most healthy, equipment or parts are then in 100% health status.Typically The health degree using middle system falls between.

Health status grade determines according to health degree R, according to R difference numerical range, system is divided into n different healthy shape State grade.

By being calculated the health degree R of system, compare the scope that R is in, and then may determine that residing for current system Health status grade.

Step 2: the health degree of all parts level equipment that computing subsystem comprises

Complex device or system are divided by architecture feature, component-level-subsystem irrespective of size-system-level three grades can be divided into. Wherein the feature of component-level equipment is simple in construction, can set up its relatively accurate model by all means thus enter it Row health state evaluation based on model.Conventional health state evaluation method is: modelling, level and fuzzy mathematics method, artificial Neural network and method based on Bayesian network etc..In modelling, conventional health state evaluation method is: (1) physics Model, (2) knowledge based library model, (3) data mining model, (4) built-up pattern.

For the feature of parts, correlation method is selected to calculate the health degree H of each parts.

Step 3: the holistic health degree that the component-level cluster tool that computing subsystem comprises couples

Cluster tool S for component-level₀={ A₁,A₂,…,A_n, its health degree R is relevant with the connection form of equipment, its Typical case's connection form includes series, parallel, series-parallel connection, voting machine connecting mode etc..As shown in Figure 3.

1. (equipment of its n component-level constitutes subsystem, only n parts by series connection to cascade together The all normal working subsystem of the equipment of level the most normally works；The equipment failure of one of them component-level, then subsystem function loses Effect.)

If A₁,A₂,…,A_nSeries connection (shown in Fig. 3 (a)), then cluster tool S₀Health degree be:

$R_{S_0}=\underset{i=1}{\overset{n}{\mathrm{\Π}}}{R}_{A_i}---\left(1\right)$

2. parallel form (equipment of its n component-level by being linked together composition subsystem parallel, wherein any one The normal working subsystem of equipment of component-level just can normally work；Only n component-level equipment failure, then subsystem function just loses Effect.)

If A₁,A₂,…,A_n(shown in Fig. 3 (b)) in parallel, then cluster tool S₀Health degree be:

$R_{S_0}=1-\underset{i=1}{\overset{n}{\mathrm{\Π}}}(1-R_{A_i})---\left(2\right)$

3. series-parallel connection form (equipment of its component-level is formed by connecting to mix with parallel connection, forms mixing connection form)

If A₁,A₂,…,A_nSeries-parallel connection (shown in Fig. 3 (c)), then cluster tool S₀Health degree be:

$R_{S_0}=\underset{i=1}{\overset{k}{\mathrm{\Π}}}{R}_{A_i}\·\[1-\underset{i=k+1}{\overset{n}{\mathrm{\Π}}}(1-R_{A_i})\]---\left(3\right)$

4. voting machine form (subsystem of the equipment composition of n component-level, K the normal work of component-level equipment at least within Making, subsystem could normally work, and more than the equipment failure of (n-k) individual component-level, subsystem just lost efficacy.So system forms table Certainly device form)

If A₁,A₂,…,A_nFor n takes k model (shown in Fig. 4), then cluster tool S₀Health degree be:

$R_{S_0}=\underset{i=k}{\overset{n}{\mathrm{\Σ}}}\left(\begin{array}{c}n\\ i\end{array}\right)\underset{k}{\overset{i}{\mathrm{\Π}}}{R}_{A_k}\·\underset{k}{\overset{n-i}{\mathrm{\Π}}}(1-R_{A_k})---\left(4\right)$

5. (it is not above-mentioned four kinds of forms that the equipment of its component-level couples to complex form, and structure complexity is linked together composition Subsystem)

By canonical systems such as cascade, parallel form, voting machine forms, the calculating of the COMPLEX SYSTEM RELIABILITY of composition, Can by by the bottom upwards, in layer evaluate step by step, thus obtain the strong of the sophisticated subsystems that canonical system constitutes Kang Du result.

Step 4: calculate the profust reliability characterized by subsystem irrespective of size status data

Health degree based on status data is the profust reliability calculating measurable parameters data value, and then assesses this and be The health status of system.The method does not relies on system model, it is only necessary to the real-time measurement values of this subsystem state parameter, it is adaptable to The health state evaluation of unknown system model.Utilize the running state data of subsystem own to determine the system health shape of multiparameter State method is as follows: assume that subsystem S comprises the parameter alpha={ a characterizing subsystem running status₁,a₂,…,a_n, state parameter Set can be regarded as a new equipment, has

R_α=f (w, a) (5)

Wherein, w represents the weighing factor of subsystem state parameter sets sub-system health status, and f representation rate mould is reliable Degree algorithm.

Concrete calculation procedure is divided into four steps:

1. determine the status data that reflection subsystem runs

It is selected to reflect the key parameter of subsystem, or the parameter of subsystem performance change can be characterized as this son The health indicator of system, by the analysis to this state parameter, obtains the health status of reflection system.

2. gather data prediction

, in gatherer process, often there is frequency acquisition disunity in initial data, crosses to adopt or leak and adopts phenomenon.Therefore, need Data are carried out pretreatment.Phenomenon is adopted, the method often using resampling for crossing；Phenomenon is adopted for leakage or gathers data When amount is few, needs the data collected are carried out interpolation processing, cubic Hamiltonian symmetrical systems method can be selected to carry out interpolation.

The calculating of the profust reliability of the most each state parameter

Known by the theory of profust reliability relevant health state, define fuzzy event B, at initial time t₀At system In fuzzy success status.Here, system is at initial time t₀State be considered.

First, it is assumed that two Fuzzy Time are as follows:

A={T_SFAt time [t₀,t_o+ t] in do not occur

B={ system is at t₀Moment is in fuzzy success status }

So, it is also referred to as rate mould health degree in t profust reliability R (t), is defined as:

$R\left(t\right)=1-P\left(\stackrel{\‾}{A}\right|B)\·P\left(B\right)-P\left(\stackrel{\‾}{B}\right)$

Wherein:

$p_{ij}(t_0,t)=\frac{n_{ij}(t_0,t)}{N_i(t_0,t)}=\frac{n_{ij}(t_0,t)}{\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{n}_{ij}(t_0,t)}---\left(7\right)$

$\mathrm{\φ}\left(t\right)=\left({\mathrm{\φ}}_{S_1}\right(t),{\mathrm{\φ}}_{S_2}(t),\mathrm{...}{\mathrm{\φ}}_{S_n}(t\left)\right)\∈R^{n\×1}$

φ (t)=P (t₀,t)·φ(t₀) (8)

Parameter in above-mentioned formula, wherein P_ij(t₀, the system that t) represents is at time (t₀,t_o+ t) in, system is from S_iState turns Move on to S_jProbability size；Represent system degree of membership；The system that represents is at t₀Moment is in state S_iState Probability；μ_SAnd μ_FRepresent the degree of membership being in fuzzy success or fuzzy fault.n_ij(t₀, t) represent [t₀, t] and time interval is from S_iShape State transfers to S_jThe number of state；φ (t) represents this system mode probability vector of t time；(i=1,2 ..., n) represent In time t system mode S_iProbability.

4. multimode parameter health degree based on status data calculates

Utilize formula (5) computing subsystem profust reliability based on status data.

Step 5: the holistic health degree of subsystem irrespective of size calculates

Assume the equipment S that subsystem S comprises n component-level₀={ A₁,A₂,…,A_n, and characterize subsystem running status Parameter alpha={ a₁,a₂,…,a_n, then the equipment of the next stage component-level that subsystem health status is comprised by subsystem simultaneously The health status combination of health status and the running state parameter of subsystem own determines.Now, the structure of subsystem is S= {S₀, α }, the health degree that can obtain subsystem is:

$R_S=-w_1{R}_{S_0}+w_2{R}_{\mathrm{\α}}---\left(9\right)$

Wherein,Represent the health degree of the cluster tool of the component-level that subsystem comprises；R_aRepresent the state ginseng of subsystem The profust reliability of number；w₁,w₂Represent component-level cluster tool and the subsystem state parameter sets sub-system of subsystem respectively The weighing factor of health status, specifically can analyze according to safety and reliability etc. and draw, and need to be true in conjunction with practical engineering application Fixed.

Step 6: according to health degree difference numerical range, carry out health state evaluation

By the calculated R of previous step, compare the scope that R is in, and then may determine that being good for residing for current sub-system Health state grade state.I.e. according to the value of health degree, according to pre-determined health status grade (step one), by subsystem irrespective of size Holistic health degree value of calculation and health status interval contrast, draw the health status grade of subsystem irrespective of size.

Therefore, when the system is operated, system health state is showed by the value of health degree R, and incorporation engineering actual enter Row determines the level limit of classification, so that it is determined that system health state grade.

3, advantage and effect: the invention provides a kind of dangerous materials and destroy the Fusion Strain data combination of equipment subsystem level Formula health state evaluation method.Its advantage is to consider health degree, the connected mode of each parts that subsystem comprises parts, with And the health degree of the status data of subsystem, and profust reliability is utilized preferably to merge the change impact of each state.Percent of pass Mould reliability reflection health degree, can describe the slight change of the holistic health state of subsystem, contributes to realizing equipment health State in real time, overall monitor and early warning, method highly versatile, it is simple to Project Realization.

[accompanying drawing explanation]

Accompanying drawing described herein is used for providing being further appreciated by the embodiment of the present invention, constitutes of the application Point, it is not intended that the restriction to the embodiment of the present invention.In the accompanying drawings:

Fig. 1 is the combustion subsystem composition that equipment destroyed by dangerous materials.

Fig. 2 is the inventive method FB(flow block).

Fig. 3 a is equipment cascade.

Fig. 3 b is equipment parallel form.

Fig. 3 c is equipment series-parallel connection form.

Fig. 4 is that equipment voting machine couples form.

Fig. 5 is sensor measurement subsystem status data.

Fig. 6 is that a combustion room temperature changes over data after scattergram and interpolation processing.

Fig. 7 is the membership function of a combustion room temperature.

Fig. 8 is the profust reliability curve of a combustion room temperature.

The profust reliability curve of Tu9Shi dual firing chamber temperature.

Figure 10 is the profust reliability curve of furnace pressure.

Figure 11 be oil tank liquid level profust reliability curve.

Figure 12 is the rate mould health status curve based on state of combustion subsystem.

Figure 13 is combustion subsystem health evaluating result curve.

[detailed description of the invention]

The combination rate mould that the present invention provides a kind of dangerous materials theoretical based on profust reliability to destroy equipment subsystem is healthy State evaluating method, three contents of major design: design subsystem comprises the profust reliability meter that component-level cluster tool couples Calculate；Design utilization state data calculate the profust reliability of the subsystem characterized by status data；The overall rate of design subsystem Mould reliability calculating.

The combustion subsystem of destruction equipment is mainly by a burner, No. two burners, air supply device, oil feed pump and urgent Heat sink forms, and wherein, the ignition temperature of combustion subsystem is exactly state parameter specific to combustion subsystem, and it is by firing The peculiar parameter of subsystem that burner, air supply device are collectively forming.Subsystem primary structure is as shown in Figure 1.

The Data Source of the present embodiment destroys the sensor measurement data of equipment on July 9th, 2009, data mode such as figure 5.Data calculation process is to carry out under the Matlab2010a environment on CPU frequency 2.93GHz, internal memory 3.00GB computer.Real Execute step as follows:

Step one: calculate the health degree of each building block of combustion subsystem

Utilize sensor measurement data, health state evaluation method based on model, calculate combustion subsystem respectively each The health degree R of individual building block_Ai。

Step 2: computing subsystem comprises the health degree that component-level cluster tool couples

The connection form of each component-level equipment that combustion subsystem is comprised is series connection, then calculated by formula (1)

Step 3: utilization state data calculate the profust reliability of the subsystem characterized by status data

1. determine the state parameter that reflection system is run

Destruction equipment combustion subsystem is as the kernel subsystems of equipment, and its effect is to make material divide by high-temp combustion Solve, and the state parameter embodying system particular characteristic is a combustion room temperature, dual firing chamber's temperature, furnace pressure and oil tank liquid level.

2. data prediction

Calculate profust reliability and need to meet the law of large numbers, the online health state evaluation of system to be accomplished, need big data Amount.Accordingly, it would be desirable to initial data interpolation processing.As a example by a combustion room temperature, it is utilized respectively linear interpolation, cubic spline interpolation With three kinds of interpolation methods of cubic Hamiltonian symmetrical systems, initial data is carried out interpolation, after interpolation, between data point, be spaced apart 5s, as Shown in Fig. 6.

The most each parameter health degree of membership determines

As a example by a combustion room temperature, the normal range of a combustion room temperature is 700-900 DEG C, and can be higher than 900 DEG C, it is impossible to Less than 700 DEG C.Here select membership function as shown below.Arrange the lower threshold of normal range be 720 main If reason allows for being set to 700, then, when algorithm calculates health status decline, have resulted in serious consequence.Just arrange Often the lower limit of scope is 720 can to detect that temperature downward trend, i.e. system progressively disengage the trend of health status, Ke Yiti Before pinpoint the problems, prevent trouble before it happens.Its membership function is as shown in Figure 7.

The most each parameter profust reliability calculates

Calculating as a example by the profust reliability of a combustion room temperature, fetch data sampling period T=5s, from the data of Fig. 5 Curve chart takes a temperature every 5s, the temperature value t that will obtain₀,t₁,…t_nSubstitute in formula (7), i.e. available every 5s The state transition probability p of temperature_ij(t₀,t)；Again the temperature value obtained is substituted in a combustion room temperature membership function, to obtain final product To corresponding φ (t_i), then according to the available φ (t) of formula (8).According to the rate mould that can be calculated a combustion room temperature by formula (6) Reliability value, draws curve such as Fig. 8.In like manner can obtain the profust reliability value of other parameters, draw curve such as Fig. 9,10,11.

5. multiparameter health status based on status data calculates

According to formula (5), combustion subsystem profust reliability based on status data R is:

R=w₁·R_{One combustion room}+w₂·R_{Dual firing chamber}+w₃·R_{Furnace pressure}+w₄·R_{Oil tank liquid level}

Take w₁=0.3, w₂=0.3, w₃=0.2, w₄=0.2, then the health status curve of combustion subsystem, such as Figure 12 institute Show.Solid black lines represents the health status change of combustion subsystem, and oil tank liquid level is always 1, and other lines represent each status data Health status change.Be can be seen that health state evaluation based on status data can be from each state parameter of system by this example Middle acquisition systematic function change information, and then the health status change of system is given with the form of profust reliability, preferably merge Each Parameters variation information.

Step 4: the overall profust reliability of combustion subsystem calculates

According to formula (9), the health status of subsystem depends not only on state parameter, additionally depends on the parts under subsystem Level equipment, takes w₁=w₂=0.5, then the health evaluating result that combustion subsystem is final as shown in figure 13, contains the line of A mark in figure (uppermost) represents that the profust reliability after unit construction, middle dotted line represent the profust reliability of subsystem, and black is real Line (nethermost) represents the profust reliability of state parameter.Detailed data is shown in Table 1.

The profust reliability result of calculation of table 1 subsystem irrespective of size

Step 5: be classified to different health level according to profust reliability difference numerical range, carry out combustion subsystem Health state evaluation

The health state evaluation result of table 2 subsystem irrespective of size

Sum up design above and example, it can be deduced that conclusion: the rate mould that this method constructs a subsystem level is healthy The universal method of state estimation, has preferably merged component-level equipment and the situation of the running status of subsystem own, can realize district The slight change of the holistic health state of molecular system, it is possible to realize health status health evaluating and the monitoring of equipment well.

Claims (4)

1. the health state evaluation method of a dangerous materials destruction equipment subsystem level, it is characterised in that: specifically include following step Rapid:

Step one: health status classification

Assuming that R represents the probability of the health degree of system, i.e. health status, span is [0,1]；As R=0, represent system The most unhealthy, equipment or parts are in severe fault condition；As R=1, represent that system is the most healthy, equipment or parts Then it is in 100% health status；The health degree using middle system falls between；

Step 2: the health degree of all parts level equipment that computing subsystem comprises

For equipment or the feature of parts, correlation method is selected to calculate the health degree H of each parts；

Step 3: the holistic health degree that the component-level cluster tool that computing subsystem comprises couples

Cluster tool S for component-level₀={ A₁,A₂,…,A_n, its health degree R is relevant with the connection form of equipment, including such as Lower form:

3.1 cascade

The equipment that the equipment of n component-level constitutes subsystem, only n component-level by series connection together the most normally works Subsystem the most normally works；The equipment failure of one of them component-level, then subsystem function lost efficacy；

If A₁,A₂,…,A_nSeries connection, then cluster tool S₀Health degree be:

$R_{S_0}=\underset{i=1}{\overset{n}{\mathrm{\Π}}}{R}_{A_i}---\left(1\right)$

3.2 parallel form

The equipment of n component-level is by being linked together composition subsystem parallel, and wherein the equipment of any one component-level is normal Working subsystem just can normally work；Only n component-level equipment failure, then subsystem function just lost efficacy；

If A₁,A₂,…,A_nParallel connection, then cluster tool S₀Health degree be:

$R_{S_0}=1-\underset{i=1}{\overset{n}{\Π}}(1-R_{A_i})---\left(2\right)$

3.3 series-parallel connection forms

The equipment of component-level is formed by connecting to mix with parallel connection, forms mixing connection form；

If A₁,A₂,…,A_nSeries-parallel connection, then cluster tool S₀Health degree be:

$R_{S_0}=\underset{i=1}{\overset{k}{\Π}}{R}_{A_i}\·\[1-\underset{i=k+1}{\overset{n}{\Π}}(1-R_{A_i})\]---\left(3\right)$

3.4 voting machine forms

The subsystem of the equipment composition of n component-level, K component-level equipment normally works at least within, and subsystem could be normal Work, more than the equipment failure of (n-k) individual component-level, subsystem just lost efficacy；So system forms voting machine form；

If A₁,A₂,…,A_nFor n takes k model, then cluster tool S₀Health degree be:

$R_{S_0}=\underset{i=k}{\overset{n}{\Σ}}\left(\begin{array}{c}n\\ i\end{array}\right)\underset{k}{\overset{i}{\Π}}{R}_{A_k}\·\underset{k}{\overset{n-i}{\Π}}(1-R_{A_k})---\left(4\right)$

3.5 complex form

It is not above-mentioned four kinds of forms that the equipment of component-level couples, and structure complexity is linked together composition subsystem；

Step 4: calculate the profust reliability characterized by subsystem irrespective of size status data

Utilize the running state data of subsystem own as follows to the system health status method determining multiparameter: to assume subsystem S Comprise the parameter alpha={ a characterizing subsystem running status₁,a₂,…,a_n, a new equipment is regarded in the set of state parameter as, Have

R_α=f (w, a) (5)

Wherein, w represents the weighing factor of subsystem state parameter sets sub-system health status, and f represents that profust reliability is calculated Method；

Step 5: the holistic health degree of subsystem irrespective of size calculates

Assume the equipment S that subsystem S comprises n component-level₀={ A₁,A₂,…,A_n, and characterize the ginseng of subsystem running status Number α={ a₁,a₂,…,a_n, then the equipment of the next stage component-level that subsystem health status is comprised by subsystem simultaneously is healthy The health status combination of state and the running state parameter of subsystem own determines；Now, the structure of subsystem is S={S₀, α }, the health degree obtaining subsystem is:

$R_S=w_1{R}_{S_0}+w_2{R}_{\mathrm{\α}}---\left(9\right)$

Wherein,Represent the health degree of the cluster tool of the component-level that subsystem comprises；R_aRepresent the state parameter of subsystem Profust reliability；w₁,w₂Represent that component-level cluster tool and the subsystem state parameter sets sub-system of subsystem are healthy respectively The weighing factor of state；

Step 6: according to health degree difference numerical range, carry out health state evaluation

By calculated R, compare the scope that R is in, and then determine the health status level status residing for current sub-system； I.e. according to the value of health degree, according to pre-determined health status grade, by the holistic health degree value of calculation of subsystem irrespective of size with strong Health state interval contrasts, and draws the health status grade of subsystem irrespective of size.

A kind of dangerous materials the most according to claim 1 destroy the health state evaluation method of equipment subsystem level, its feature Being: in step one, health status grade determines according to health degree R, according to R difference numerical range, system is divided into n difference Health status grade；

By being calculated the health degree R of system, compare the scope that R is in, and then determine the health status residing for current system Grade.

A kind of dangerous materials the most according to claim 1 destroy the health state evaluation method of equipment subsystem level, its feature It is: in step 2, corresponding method includes modelling, level and fuzzy mathematics method, artificial neural network method and based on Bayes Network method.

A kind of dangerous materials the most according to claim 1 destroy the health state evaluation method of equipment subsystem level, its feature It is: step 4 also comprises the steps:

4.1 determine the status data that reflection subsystem runs

It is selected to reflect the key parameter of subsystem, or the parameter of subsystem performance change can be characterized as this subsystem Health indicator, by the analysis to this state parameter, obtain the health status of reflection system；

4.2 gather data prediction

For leakage adopt phenomenon or gather data volume few time, need the data collected are carried out interpolation processing, select three times Hermite interpolation method carries out interpolation；

The calculating of the profust reliability of 4.3 each state parameters

Known by the theory of profust reliability relevant health state, define fuzzy event B, at initial time t₀System is in mould Stick with paste success status；System is at initial time t₀State be considered；

First, it is assumed that two Fuzzy Time are as follows:

A={T_SFAt time [t₀,t_o+ t] in do not occur

B={ system is at t₀Moment is in fuzzy success status }

So, it is also referred to as rate mould health degree in t profust reliability R (t), is defined as:

$R\left(t\right)=1-P\left(\stackrel{\‾}{A}\right|B)\·P\left(B\right)-P\left(\stackrel{\‾}{B}\right)$

Wherein:

$p_{ij}(t_0,t)=\frac{n_{ij}(t_0,t)}{N_i(t_0,t)}=\frac{n_{ij}(t_0,t)}{\underset{j=1}{\overset{n}{\Σ}}{n}_{ij}(t_0,t)}---\left(7\right)$

$\mathrm{\φ}\left(t\right)=({\mathrm{\φ}}_{S_1}\left(t\right),{\mathrm{\φ}}_{S_2}\left(t\right),...{\mathrm{\φ}}_{S_n}\left(t\right))\∈R^{n\×1}$

φ (t)=P (t₀,t)·φ(t₀) (8)

Parameter in above-mentioned formula, wherein P_ij(t₀, the system that t) represents is at time (t₀,t_o+ t) in, system is from S_iState transfers to S_j Probability size；Represent system degree of membership；The system that represents is at t₀Moment is in state S_iState probability；μ_S And μ_FRepresent the degree of membership being in fuzzy success or fuzzy fault；n_ij(t₀, t) represent [t₀, t] and time interval is from S_iState shifts To S_jThe number of state；φ (t) represents this system mode probability vector of t time；Represent In time t system mode S_iProbability；

4.4 multimode parameter health degrees based on status data calculate

Utilize formula (5) computing subsystem profust reliability based on status data.