CN104536292A - Method for conducting fault diagnosis on heat exchanger of aircraft environmental control system based on STF (Strong Tracking Filter) and MB - Google Patents

Abstract

The invention provides a method for conducting fault diagnosis on a heat exchanger of an aircraft environmental control system based on an STF (Strong Tracking Filter) and an MB. The method includes the steps that first, the fault feature parameter of the heat exchanger is estimated based on the STF; second, fault detection and diagnosis are carried out based on the MB algorithm; third, fault amplitude estimation is conducted. On the basis of the fault analysis of the heat exchanger, in accordance with the characteristic that the fault feature parameter cannot be measured directly, the fault feature parameter of the heat exchanger is estimated in real time through the STF in the method, and the fault feature parameter of the heat exchanger is input to the MB algorithm to judge the working condition and the fault type of the heat exchanger. The analysis of experimental results shows that the method can be effectively applied to fault diagnosis of the heat exchanger of the aircraft environmental control system.

Description

A kind of plane environmental control system heat interchanger method for diagnosing faults based on STF and MB

Technical field

The present invention relates to the technical field of plane environmental control system heat interchanger fault diagnosis, be specifically related to a kind of based on STF (Strong Tracking Filter) and the plane environmental control system heat interchanger method for diagnosing faults revising Bayes (MB) sorting algorithm (Modified Bayes ' classificationalgorithm) model.

Background technology

Plane environmental control system is the critical system in aircraft, the temperature of primary responsibility aircraft cockpit and equipment compartment controls, for passenger and driver supply air, suitable temperature and pressure, take away electronic product in time and run the amount of heat produced, prevent the fault caused because of apparatus overheat from occurring.Sharply increasing progressively in recent years along with air electronics quantity, the operation of electronic equipment produces a large amount of heats, if take cooling measure not in time, will cause the generation of electronic failure, and the major accident of fatal crass even can be caused to occur.Therefore, guarantee that the reliability service of plane environmental control system seems particularly important.Fault diagnosis technology is by the tracking to equipment failure characteristic parameter, and can make diagnosis to the health problem of equipment, be the effective measures guaranteeing plane environmental control system reliability service.

Heat interchanger is the main and common building block of plane environmental control system, and its reliability directly determines operational efficiency and the reliability of environmental control system.It is guarantee the important guarantee of the efficient and safe operation of environmental control system that heat exchanger fault implements fault diagnosis.Therefore, select heat interchanger as the object of fault diagnosis herein.

At present, the research for heat interchanger mainly concentrates on design of heat exchanger, structure optimization and emulation aspect, and the research of heat interchanger fault diagnosis and health evaluating aspect is less.Current existing heat interchanger method for diagnosing faults mainly based on EKF (EKF) and dual model filtering (DMF) algorithm (based on EKF algorithm, utilize the EKF model that two different, follow the tracks of soft phase and mutation status respectively), these two kinds of methods are all the fault diagnosises being realized equipment by the estimation of heat exchanger fault phase related parameter, but all exist and the problems such as more responsive are chosen for modeling error and initial value, and EKF model is open loop models, its gain matrix is the dynamic conditioning with the change of the situation of tracking not, therefore tracking accuracy and effect are not high.Strong tracking filter can be good at overcoming above problem, and STF algorithm is a kind of closed loop model, its gain matrix can along with the online dynamic conditioning of change of the situation of tracking, and therefore STF more effectively can realize parameter estimation, and estimates for the Fault characteristic parameters of heat interchanger.

Summary of the invention

The object of the present invention is to provide a kind of plane environmental control system heat interchanger method for diagnosing faults based on STF and MB, the method can be effectively applied to the fault diagnosis of plane environmental control system heat interchanger.

The technical solution used in the present invention is: a kind of plane environmental control system heat interchanger method for diagnosing faults based on STF and MB, and the method step is as follows:

The first step: the heat interchanger Fault characteristic parameters based on STF is estimated

On the basis of heat-exchanger model, by the analysis of heat exchanger fault mode, draw heat interchanger Fault characteristic parameters; STF algorithm is with the inlet port temperature of the cold limit of heat interchanger and Re Bian for input, and real-time online estimates heat interchanger Fault characteristic parameters;

Second step: based on the fault detection and diagnosis of MB algorithm

Select to revise Bayesian Classification Arithmetic as heat interchanger fault detection and diagnosis algorithm; Estimate the input that the Fault characteristic parameters drawn is MB algorithm with STF algorithm, draw the fault amplitude of each Fault characteristic parameters, when the fault amplitude of Fault characteristic parameters is greater than the threshold value preset, be judged to be that the fault that this Fault characteristic parameters is corresponding occurs; Concrete fault detection and diagnosis strategy is as follows:

STF algorithm is estimated that all Fault characteristic parameters drawn carry out following operation respectively:

If ● certain Fault characteristic parameters γ _imB arithmetic result be greater than predetermined threshold duration report to the police, then show that the fault corresponding with certain Fault characteristic parameters occurs; Jump to the 3rd step;

● if above-mentioned condition does not meet, then jump to the first step;

3rd step: fault Amplitude Estimation

The Fault characteristic parameters γ reported to the police _icorresponding equivalent fault amplitude EEFA (estimated equivalent fault amplitude) through type (1) calculates,

$\mathrm{EEFA}:{\widehat{\mathrm{\γ}}}_i\left(k\right)-{\mathrm{\γ}}_i^0---\left(1\right)$

Wherein: EEFA is the equivalent fault amplitude of heat interchanger when breaking down, for heat interchanger normal run time fault characteristic parameter γ _iestimated result, for heat interchanger K moment run time fault characteristic parameter γ _iestimated result;

Jump to the first step.

Further, described plane environmental control system heat interchanger is the air-air convection recuperator of plate-fin, intersect the pipe passage of 90 degree and heat exchange flat board by cold limit air duct and Re Bian air duct two to form, hot-air and cold air carry out heat exchange by heat exchange manifold and heat exchange flat board; In the actual use of heat interchanger, the entrance on cold limit and hot limit, outlet temperature are measured.Heat interchanger is often revealed, block and fouling fault, and this three classes fault is " parameter error " type fault, and fault signature shows as the change of the immeasurability parameters such as MAF, air valid circulation area, heat transfer coefficient of heat exchanger; Therefore, be difficult to by the entrance on cold limit and hot limit, outlet temperature the fault detection and diagnosis directly carrying out heat interchanger.

Further, the first step of described plane environmental control system heat interchanger method for diagnosing faults, with the entrance on the cold limit of heat interchanger measurement parameter and hot limit, outlet temperature for input, adopt a kind of strong tracking filfer based on heat-exchanger model correction (STF), realize heat interchanger state and parametric joint is estimated, real-time online estimates the heat interchanger fault signature that can not directly measure by measurable parameter; Uncertain and Unmarried pregnancy in strong tracking filfer heat exchanger model has stronger robustness, and the gradual or mutation failure characteristic parameter after heat exchanger reaches stable state has very strong tracking power.

Further, the second step of described plane environmental control system heat interchanger method for diagnosing faults, the heat interchanger Fault characteristic parameters estimated with strong tracking filfer is input, MB method is adopted to estimate the fault amplitude showing that each Fault characteristic parameters is corresponding in real time, when the fault amplitude of Fault characteristic parameters is greater than the threshold value preset, be judged to be that the fault that this Fault characteristic parameters is corresponding occurs, thus realize effective aircraft environmental control system heat interchanger fault diagnosis.

The present invention's advantage is compared with prior art:

(1) the fault detection and diagnosis correlative study for aircraft environmental control system heat interchanger is deficient, and the present situation that achievement in research practical application effect has much room for improvement, proposes the effective ways of the plane environmental control system heat interchanger fault diagnosis of complete set;

(2) utilize Strong tracking filter method, with heat interchanger measurable parameter for immeasurablel heat interchanger Fault characteristic parameters is estimated in input, compensate for the problem effectively can not being carried out heat interchanger fault diagnosis by measurable parameter.

(3) based on the strong tracking filfer that heat-exchanger model builds, uncertain and Unmarried pregnancy in heat exchanger model has stronger robustness, gradual or mutation failure characteristic parameter after heat exchanger reaches stable state has very strong tracking power, there is the feature that real-time online estimates heat interchanger Fault characteristic parameters simultaneously, achieve effective Fault characteristic parameters in real time to estimate, for the raising of the accuracy rate of fault diagnosis, and the reduction of false alarm rate provides effective support.

(4) based on the heat interchanger method for diagnosing faults of STF and MB model, breach most of method for diagnosing faults in the past and only can provide fault diagnosis result and less to the problem of the amplitude that is out of order, described method can be estimated on basis at effective Fault characteristic parameters, provide the fault amplitude of each fault, and carry out effective fault diagnosis according to the change of fault amplitude.

Accompanying drawing explanation

Fig. 1 is the plane environmental control system heat interchanger method for diagnosing faults flow process based on STF and MB;

Fig. 2 is the heat interchanger Fault characteristic parameters estimation model Establishing process based on STF;

Fig. 3 is STF algorithm t under normal condition _c.out, t _h.out, γ ₁, γ ₂, γ ₃and γ ₄estimated value;

Fig. 4 is MB algorithm γ under normal condition ₁, γ ₂, γ ₃and γ ₄fault Amplitude Estimation value;

Fig. 5 is STF algorithm t under fault F1 state _c.out, t _h.out, γ ₁, γ ₂, γ ₃and γ ₄estimated value;

Fig. 6 is MB algorithm γ under fault F1 state ₁, γ ₂, γ ₃and γ ₄fault Amplitude Estimation value;

Fig. 7 is STF algorithm t under fault F8 state _c.out, t _h.out, γ ₁, γ ₂, γ ₃and γ ₄estimated value;

Fig. 8 is MB algorithm γ under fault F8 state ₁, γ ₂, γ ₃and γ ₄fault Amplitude Estimation value;

Fig. 9 is γ under fault F9 and fault F1 state ₁the estimated value contrast of fault amplitude;

Figure 10 is STF (a) algorithm DMF (b) algorithm γ under fault F9 state ₁, estimated value contrast.

Embodiment

The present invention is further illustrated below in conjunction with accompanying drawing and specific embodiment.

The present invention is based on the plane environmental control system heat interchanger method for diagnosing faults flow process of STF and MB as shown in Figure 1.Idiographic flow can be summarized as following three steps:

The first step: the heat interchanger Fault characteristic parameters based on STF is estimated

On the basis of heat-exchanger model, by the analysis of heat exchanger fault mode, draw heat interchanger Fault characteristic parameters.STF algorithm is with the inlet port temperature of the cold limit of heat interchanger and Re Bian for input, and real-time online estimates heat interchanger Fault characteristic parameters.

Second step: based on the fault detection and diagnosis of MB algorithm

Select to revise Bayesian Classification Arithmetic (Modified bayes ' classification algorithm) as heat interchanger fault detection and diagnosis algorithm.Estimate the input that the Fault characteristic parameters drawn is MB algorithm with STF algorithm, draw the fault amplitude of each Fault characteristic parameters, when the fault amplitude of Fault characteristic parameters is greater than the threshold value preset, be judged to be that the fault that this Fault characteristic parameters is corresponding occurs.Concrete fault detection and diagnosis strategy is as follows:

STF algorithm is estimated that all Fault characteristic parameters drawn carry out following operation respectively:

If ● certain Fault characteristic parameters γ _imB arithmetic result be greater than predetermined threshold duration report to the police, then show that the fault corresponding with certain Fault characteristic parameters occurs; Jump to the 3rd step.

● if above-mentioned condition does not meet, then jump to the first step.

3rd step: fault Amplitude Estimation

The Fault characteristic parameters γ reported to the police _icorresponding equivalent fault amplitude EEFA (estimated equivalent fault amplitude) through type (1) calculates.

$\mathrm{EEFA}:{\widehat{\mathrm{\γ}}}_i\left(k\right)-{\mathrm{\γ}}_i^0---\left(1\right)$

Wherein: for heat interchanger normal run time fault characteristic parameter γ _iestimated result, for heat interchanger K moment run time fault characteristic parameter γ _iestimated result.

Jump to the first step.

The first step is based in the aircraft heat interchanger Fault characteristic parameters estimation of STF, concrete:

1), about STF algorithm

Become stochastic system when considering a quasi-nonlinear, available following separate manufacturing firms model describes.

x(k+1)＝f(k,u(k),x(k))+Γ(k)v(k) (2)

y(k+1)＝g(k+1,x(k+1))+e(k+1) (3)

Wherein, x ∈ IR ⁿfor system state, u ∈ IR ^pfor system input; Y ∈ IR ^mfor system exports; F:IR ⁿ× IR ^p→ IR ⁿfor nonlinear equation; G:IR ⁿ→ IR ^mfor observation equation, v ∈ IR ^qfor the process noise of zero mean Gaussian white noise, Q is the covariance of v; Measurement noise e ∈ IR ^malso be one for zero mean Gaussian white noise, its covariance is R; Γ is a known matrix of coefficients; V and e is statistically independent.

The step that employing STF carries out state estimation is as follows:

(1) initialization

Setting initial value p (0|0).

(2) predict

The a step of forecasting value of state is:

$\hat{x}(k+1|k)=f(k,u\left(k\right),\hat{x}\left(k\right|k))---\left(4\right)$

$F(k,u\left(k\right),\hat{x}\left(k\right|k))=\frac{\mathrm{\δf}(k,u\left(k\right),x\left(k\right))}{\mathrm{\δx}}{|}_{x=\hat{x}\left(k\right|k)}---\left(5\right)$

$G(k+1,\hat{x}(k+1|k))=\frac{\mathrm{\δg}(k+1,x(k+1))}{\mathrm{\δx}}{|}_{x=\hat{x}(k+1|k)}---\left(6\right)$

(3) residual sequence is asked

$\mathrm{\γ}(k+1)=y(k+1)-g(k+1,\hat{x}(k+1|k))---\left(7\right)$

(4) multiple suboptimum fading factor L (k+1) is asked

L(k+1)＝diag{λ ₁(k+1),λ ₂(k+1),...,λ _n(k+1)} (8)

${\mathrm{\λ}}_1=\left\{\begin{array}{cc}{\mathrm{\ζ}}_i\mathrm{\η}(k+1);& {\mathrm{\ζ}}_i\mathrm{\η}(k+1)>1\\ 1;& {\mathrm{\ζ}}_i\mathrm{\η}(k+1)\≤1\end{array}\right.---\left(9\right)$

i＝1,2,…,n

Wherein:

$\mathrm{\η}(k+1)=\frac{\mathrm{tr}\left[N(k+1)\right]}{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\mathrm{\ζ}}_i{d}_{\mathrm{ii}}(k+1)}---\left(10\right)$

$N(k+1)=V_0(k+1)-\mathrm{\β}\·R(k+1)-G(k+1,\hat{x}(k+1|k))\mathrm{\Γ}\left(k\right)Q\left(k\right){\mathrm{\Γ}}^T\left(k\right)G^T(k+1,\hat{x}(k+1|k))---\left(11\right)$

$\begin{array}{c}D(k+1)=F(k,u\left(k\right),\hat{x}\left(k\right|k))P\left(k\right|k)F^T(k,u\left(k\right),\hat{x}\left(k\right|k))G^T(k+1,\hat{x}(k+1|k))G(k+1,\hat{x}(k+1|k))\\ =\left(d_{\mathrm{ij}}\right)\end{array}$

$\left(12\right)$

Residual sequence covariance matrix V ₀(k+1) can be calculated by following formula:

V ₀(k+1)＝E[γ(k+1)γ ^T(k+1)] (13)

$\≈\left\{\begin{array}{cc}\mathrm{\γ}\left(1\right){\mathrm{\γ}}^T\left(1\right);& k=0\\ \frac{\mathrm{\ρ}{V}_0\left(k\right)+\mathrm{\γ}(k+1){\mathrm{\γ}}^T(k+1)}{1+\mathrm{\ρ}};& k\≥1\end{array}\right.---\left(14\right)$

ζ in formula (9) _i>=1, i=1,2 ... n is the coefficient of predefined numerical value.These numerical value can be determined by the priori of system.If state x _ifaster than the change of other states, priori should be utilized to select the ζ of bigger numerical _i.If there is no priori, all coefficient ζ _ican be decided to be " 1 ".

In formula (11), β >=1 is a selected reduction factor, and the object of introducing makes state estimation more level and smooth.This value can be selected by experiment, and its criterion is:

$\mathrm{\β}:\min \left(\underset{k=0}{\overset{L}{\mathrm{\Σ}}}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\right|x_i\left(k\right)-{\hat{x}}_i\left(k\right|k)\left|\right)---\left(15\right)$

Wherein L is emulation step number, and this criterion reflects the cumulative errors of wave filter.

In formula (14), ρ is forgetting factor, generally gets ρ=0.95.

(5) prediction error conariance battle array is asked

$P(k+1|k)=L(k+1)F(k,u\left(k\right),\hat{x}\left(k\right|k))P\left(k\right|k)F^T(k,u\left(k\right),\hat{x}\left(k\right|k))+\mathrm{\Γ}\left(k\right)Q\left(k\right){\mathrm{\Γ}}^T\left(k\right)---\left(16\right)$

(6) gain battle array is asked

Force orthogonalization can calculate gain matrix K (k+1) based on residual error, its computing formula is as follows:

$K(k+1)=P(k+1|k)G^T(k+1,\hat{x}(k+1|k)){[G(k+1,\hat{x}(k+1|k))P(k+1|k)G^T(k+1,\hat{x}(k+1|k))+R\left(k\right)]}^{-1}---\left(17\right)$

(7) state estimation error covariance matrix is asked

$P(k+1|k+1)=[I_n-K(k+1)G(k+1,\hat{x}(k+1|k))]P(k+1|k)---\left(18\right)$

(8) upgrade

$\hat{x}(k+1|k+1)=\hat{x}(k+1|k)+K(k+1)\mathrm{\γ}(k+1)---\left(19\right)$

Finally, make k=k+1, repeat previous step.

STF can be considered to a kind of wave filter of closed loop, because filter gain matrix K (k+1) calculates online according to orthogonalization principle self-adaptation.And the matrix of coefficients calculated off-line that the gain matrix of Kalman filtering is foundation dynamic process equation draws, therefore compare the expanded Kalman filtration algorithm of open loop, STF has better tracking performance.

2) the heat interchanger Fault characteristic parameters estimation model, based on STF is set up

Based on STF heat interchanger Fault characteristic parameters estimation model Establishing process as shown in Figure 2.

1) heat-exchanger model is set up: to according to heat interchanger operation logic, set up the mathematical model of heat interchanger.

2) heat interchanger fault analysis and Fault characteristic parameters analysis: the fault mode analyzing heat interchanger, and on the basis of the heat interchanger mathematical model of above-mentioned foundation, analyze the Fault characteristic parameters that each fault mode of heat interchanger is corresponding.

3) heat-exchanger model correction: according to the feature of STF algorithm, and the potential demand of heat interchanger Fault characteristic parameters, revise the heat-exchanger model set up, with the STF algorithm model facilitating follow-up foundation can estimate above-mentioned Fault characteristic parameters.

4) based on the heat interchanger Fault characteristic parameters estimation model of STF algorithm: based on the correction result of heat-exchanger model, in conjunction with STF algorithm, the heat interchanger Fault characteristic parameters estimation model based on STF algorithm is drawn.

Second step is based in the heat interchanger fault detection and diagnosis of MB algorithm, concrete,

When supposing the system normally runs, parameter is had to meet normal distribution, namely can diagnose with following formula when occurring abnormal.

Order represent θ _ithe estimated value of (k), and

$\widehat{\mathrm{\θ}}\left(k\right|k)=\left[\begin{array}{c}{\widehat{\mathrm{\θ}}}_1\left(k\right|k)\\ {\widehat{\mathrm{\θ}}}_2\left(k\right|k)\\ \\ {\widehat{\mathrm{\θ}}}_l\left(k\right|k)\end{array}\right]~N(\widehat{{\mathrm{\θ}}^0},{\mathrm{\σ}}_{\widehat{{\mathrm{\θ}}^0}}^2).---\left(20\right)$

${\mathrm{\μ}}_{{\mathrm{\θ}}_i}\left(k\right)=\frac{1}{N_1}\underset{j=1}{\overset{N_1}{\mathrm{\Σ}}}{\widehat{\mathrm{\θ}}}_i(k-j|k-j)---\left(21\right)$

${\mathrm{\σ}}_{{\mathrm{\θ}}_{i}1}^2\left(k\right)=\frac{1}{N_1-1}\underset{j=1}{\overset{N_1}{\mathrm{\Σ}}}{[{\widehat{\mathrm{\θ}}}_i(k-j|k-j)-{\widehat{\mathrm{\θ}}}_i^0]}^2---\left(22\right)$

${\mathrm{\σ}}_{{\mathrm{\θ}}_{i}2}^2\left(k\right)=\frac{1}{N_1-1}\underset{j=1}{\overset{N_1}{\mathrm{\Σ}}}{[{\widehat{\mathrm{\θ}}}_i(k-j|k-j)-{\mathrm{\μ}}_{{\mathrm{\θ}}_i}\left(k\right)]}^2---\left(23\right)$

$M{B}_i\left(k\right)=\frac{{\mathrm{\σ}}_{{\mathrm{\θ}}_{i}1}^2\left(k\right)}{{\mathrm{\σ}}_{\widehat{{\mathrm{\θ}}^0}}^2}-\ln \frac{{\mathrm{\σ}}_{{\mathrm{\θ}}_{i}2}^2\left(k\right)}{{\mathrm{\σ}}_{\widehat{{\mathrm{\θ}}^0}}^2}-1---\left(24\right)$

Wherein N ₁for data window length.Definition threshold value beta _i>0, therefore has following two propositions:

H ₀:MB _i(k)≤β _i(25)

H ₁:MB _i(k)＞β _i(26)

On the basis estimating each Fault characteristic parameters drawn at STF, by calculating the MB value of each Fault characteristic parameters in real time, and by MB value and the threshold value beta preset _icompare and work as H ₀when proposition is set up, namely judge to show that the fault that this Fault characteristic parameters is corresponding occurs.Wherein, threshold value beta _ican be selected by Computer Simulation.If select less β _i, the fault that fault degree is less can be detected, but the false alarm rate of fault detection system can be made to raise.In contrast, if select larger β _i, the fault only having fault degree more serious just can be detected, and the rate of failing to report of such fault detection system raises.

Practical application example is as follows:

1, plane environmental control system heat-exchanger model is set up

For the plate fin type heat exchanger that aviation field is conventional, illustrate how the environmental control system dynamic fault diagnosis method based on Strong tracking filter (STF) algorithm realizes.Lumped-parameter method is adopted to set up Heat Exchanger Dynamic Model.In the process of Modling model, the introduction of parameter and footnote is as subordinate list.

Cold limit population temperature is ram air temperature:

t _c.in(τ)＝t _ram(τ) (1)

Cold edge journey temperature variation:

$\left({\stackrel{\·}{m}}_c{c}_{\mathrm{pc}}\right)\mathrm{dxd}{t}_c(x,\mathrm{\τ})/\mathrm{dx}+\left({\mathrm{\η}}_c{\mathrm{\α}}_c{A}_{\mathrm{cw}}\right)\·[t_c(x,\mathrm{\τ})-t_w]=0---\left(2\right)$

Laplace conversion and inverse transformation is utilized to obtain cold limit outlet temperature:

$t_{c.\mathrm{out}}\left(\mathrm{\τ}\right)=t_w\left(\mathrm{\τ}\right)+[t_{c.\mathrm{in}}(x,\mathrm{\τ})-t_w\left(\mathrm{\τ}\right)]\·e^{-{\mathrm{\γ}}_1/{\mathrm{\γ}}_2}---\left(3\right)$

In formula: γ ₁=η _cα _ca _cw, γ ₂=ρ _ca _cv _cc _pc; η _cfor cold limit heat exchange surface efficiency; α _cfor cold limit convection transfer rate (W/m ²k); A _cwfor cold limit effective heat exchange area (m ²); for cold limit mass rate (kg.s), ρ _c, A _c, v _c, be respectively cold limit atmospheric density (kg/m ³), valid circulation area (m ²), air velocity (m/s); c _pcfor cold limit pressurization by compressed air specific heat capacity (J/ (kgK)).

Cold limit medial temperature:

$t_{\mathrm{pc}}\left(\mathrm{\τ}\right)=t_w\left(\mathrm{\τ}\right)+[t_{c.\mathrm{in}}\left(\mathrm{\τ}\right)-t_w\left(\mathrm{\τ}\right)]\·(1-e^{-{\mathrm{\γ}}_1/{\mathrm{\γ}}_2})/({\mathrm{\γ}}_1/{\mathrm{\γ}}_2)---\left(4\right)$

Hot limit temperature in is engine bleed temperature:

t _h.in(τ)＝t _bledd(τ) (5)

Hot edge journey temperature variation:

$\left({\stackrel{\·}{m}}_h{c}_{\mathrm{ph}}\right)\mathrm{dxd}{t}_h(x,\mathrm{\τ})/\mathrm{dx}+\left({\mathrm{\η}}_h{\mathrm{\α}}_h{A}_{\mathrm{hw}}\right)\·[t_h(x,\mathrm{\τ})-t_w]=0---\left(6\right)$

Laplace conversion and inverse transformation is utilized to obtain hot limit outlet temperature:

$t_{h.\mathrm{out}}\left(\mathrm{\τ}\right)=t_w\left(\mathrm{\τ}\right)+[t_{h.\mathrm{in}}(x,\mathrm{\τ})-t_w\left(\mathrm{\τ}\right)]\·e^{-{\mathrm{\γ}}_3/{\mathrm{\γ}}_4}---\left(7\right)$

In formula: γ ₃=η _hα _ha _hw, γ ₄=ρ _ha _hv _hc _ph; η _hfor cold limit heat exchange surface efficiency; α _hfor cold limit convection transfer rate (W/m ²k); A _hwfor cold limit effective heat exchange area (m ²); for cold limit mass rate (kg/s), ρ _h, A _h, v _h, be respectively cold limit atmospheric density (kg/m ³), valid circulation area (m ²), air velocity (m/s); c _phfor cold limit pressurization by compressed air specific heat capacity (J/ (kgK)).

Hot limit medial temperature:

$t_{\mathrm{ph}}\left(\mathrm{\τ}\right)=t_w\left(\mathrm{\τ}\right)+[t_{h.\mathrm{in}}\left(\mathrm{\τ}\right)-t_w\left(\mathrm{\τ}\right)]\·(1-e^{-{\mathrm{\γ}}_3/{\mathrm{\γ}}_4})/({\mathrm{\γ}}_3/{\mathrm{\γ}}_4)---\left(8\right)$

Heat interchanger wall medial temperature:

$\left(m_w{c}_{\mathrm{pw}}\right)d{t}_w/\mathrm{d\τ}=\left({\mathrm{\η}}_c{\mathrm{\α}}_c{A}_{\mathrm{cw}}\right)\·(t_{\mathrm{pc}}-t_w)+\left({\mathrm{\η}}_h{\mathrm{\α}}_h{A}_{\mathrm{hw}}\right)\·(t_{\mathrm{ph}}-t_w)---\left(9\right)$

Difference equation is:

$t_w^{\left(n\right)}=t_w^{(n-1)}+(\left({\mathrm{\γ}}_1\mathrm{\Δ\τ}\right)/\left(m_w{c}_{\mathrm{pw}}\right))(t_{\mathrm{pc}.\mathrm{out}}^{\left(n\right)}-t_w^{(n-1)})+({\mathrm{\γ}}_3\mathrm{\Δ\τ}/m_w{c}_{\mathrm{pw}})(t_{\mathrm{ph}.\mathrm{out}}^{\left(n\right)}-t_w^{(n-1)})---\left(10\right)$

2, plane environmental control system heat interchanger Fault characteristic parameters is analyzed

Fault detection and diagnosis method basic thought is the variation that many faults can be regarded as process factor, and these process factor can lie in the parameter of process model, and they can be permanent, becomes when also can be.No matter adopt which kind of method for diagnosing faults, all need to analyze the phenomenon of the failure of system, the modal fault of heat interchanger is for leaking, blocking and fouling.

(1) leak: it is that mass rate can reduce suddenly that heat exchanger leaks main manifestations, has influence on or the two product.And if ρ _hc _phfor priori is known, then have influence on A _hv _h.ρ is shown as in fault model _ha _hv _hc _phproduct, monitor ρ in real time _ha _hv _hc _phtotal value or ρ _ca _cv _cc _pcmarked change can diagnose out the leakage failure of heat exchanger.

(2) fouling: η _hα _ha _hwfouling can cause the heat transfer coefficient α of heat exchanger _cor α _hreduce, also can block plate passage time serious, heat exchange efficiency is greatly reduced.The η that real-time monitoring is total _cα _ca _cwor η _cα _ca _cwproduct change can diagnose out the fouling fault of heat exchanger.

(3) block: for heat interchanger plugging fault, valid circulation area value A _hor A _chave significant change, so in the middle of operation, the ρ that monitoring is total in real time _ha _hv _hc _phor ρ _ca _cv _cc _pcthe marked change of product can diagnose out the plugging fault of heat exchanger.

ρ _ca _cv _cc _pcutilize the nominal value of these estimated values and correlation parameter to compare, can fault eigenvalue be obtained, carry out fault diagnosis.

3, the correction of plane environmental control system heat-exchanger model and Fault characteristic parameters are estimated

In order to apply STF algorithm, the heat-exchanger model that 3.1 joints are set up is written as following form again.

Systematic observation equation:

$\left[\begin{array}{c}{t}_{c.\mathrm{out}}\left(k\right)\\ t_{h.\mathrm{out}}\left(k\right)\end{array}\right]=\left[\begin{array}{c}{t}_w\left(k\right)+[t_{c.\mathrm{in}}\left(k\right)-t_w\left(k\right)]e^{-{\mathrm{\γ}}_1\left(k\right)/{\mathrm{\γ}}_2\left(k\right)}\\ t_w\left(k\right)+[t_{c.\mathrm{in}}\left(k\right)-t_w\left(k\right)]e^{-{\mathrm{\γ}}_3\left(k\right)/{\mathrm{\γ}}_4\left(k\right)}\end{array}\right]+\left[\begin{array}{c}{n}_{\mathrm{tc}.\mathrm{out}}\left(k\right)\\ n_{\mathrm{th}.\mathrm{out}}\left(k\right)\end{array}\right]---\left(11\right)$

In formula: n _tc.out(k) and n _th.outk () represents t respectively _c.out(k) and t _h.outk (), in the observation noise in k moment, is assumed to be Gaussian distribution here.

System state equation:

$\left[\begin{array}{c}{\mathrm{\γ}}_1\left(k\right)\\ {\mathrm{\γ}}_2\left(k\right)\\ {\mathrm{\γ}}_3\left(k\right)\\ {\mathrm{\γ}}_4\left(k\right)\\ t_{\mathrm{pc}.\mathrm{out}}\left(k\right)\\ t_{\mathrm{ph}.\mathrm{out}}\left(k\right)\\ t_w\left(k\right)\end{array}\right]=\left[\begin{array}{c}{\mathrm{\γ}}_1(k-1)\\ {\mathrm{\γ}}_2(k-1)\\ {\mathrm{\γ}}_3(k-1)\\ {\mathrm{\γ}}_4(k-1)\\ f_1\left(X(k-1)\right)\\ f_2\left(X(k-1)\right)\\ f_3\left(X(k-1)\right)\end{array}\right]+\left[\begin{array}{c}{n}_{{\mathrm{\γ}}_1}(k-1)\\ n_{{\mathrm{\γ}}_2}(k-1)\\ n_{{\mathrm{\γ}}_3}(k-1)\\ n_{{\mathrm{\γ}}_4}(k-1)\\ 0\\ 0\\ n_{t_w}(k-1)\end{array}\right]---\left(12\right)$

$f_1\left(X(k-1)\right)=t_w(k-1)+[t_{c.\mathrm{in}}\left(k\right)-t_w(k-1)](1-e^{-{\mathrm{\γ}}_1(k-1)/{\mathrm{\γ}}_2(k-1)})/({\mathrm{\γ}}_1(k-1)/{\mathrm{\γ}}_2(k-1))---\left(13\right)$

$f_2\left(X(k-1)\right)=t_w(k-1)+[t_{h.\mathrm{in}}\left(k\right)-t_w(k-1)](1-e^{-{\mathrm{\γ}}_3(k-1)/{\mathrm{\γ}}_4(k-1)})/{\mathrm{\γ}}_3(k-1)/{\mathrm{\γ}}_4(k-1)---\left(14\right)$

f ₃(X(k-1))＝t _w(k-1)+((γ ₁(k-1)T)/(m _wc _pw))[t _pc.out(k-1)-t _w(k-1)]+(γ ₃(k-1)T/(m _wc _pw))[t _ph.out(k-1)-t _w(k-1)]

(15)

X(k)＝[γ ₁(k) γ ₂(k) γ ₃(k) γ ₄(k) t _pc.out(k) t _ph.out(k) t _w(k)] ^T(16)

In formula: be respectively γ ₁, γ ₂, γ ₃, γ ₄, t _wat the noise in k moment, be assumed to be Gaussian distribution here; T is observation cycle.

4, based on the plane environmental control system heat interchanger method for diagnosing faults application result of STF and MB

Aircraft flight height is adopted to be the validity check that the environmental control system convection recuperator emulated data of 2km carries out STF algorithm and MB algorithm.The model parameter of heat interchanger is as shown in table 1.

Table 1. heat-exchanger model parameter

Model parameter	Hot limit	Cold limit
			Distance between plates (mm)	5	7.5
The wing number of plies	7	8
			Wing thickness (mm)	0.15	0.15
Plate thickness (mm)	0.5	0.5
			Side plate thickness (mm)	2	2
Passage length (mm)	260	100
			Entering air temperature (DEG C)	250	90
MAF (kg/s)	0.1778	1
			Heat interchanger initial surface temperature (DEG C)	140	140

The Initial value choice of STF algorithm is as follows:

Measurement noise (n _tc.out, n _th.out) and system noise all obey white noise normal distribution.

n _tc.out～N(0,1)，n _th.out～N(0,1)， $n_{{\mathrm{\γ}}_1}~N\left(\mathrm{0,0.01}\right),n_{{\mathrm{\γ}}_2}~N\left(\mathrm{0,0.01}\right),n_{{\mathrm{\γ}}_3}~N\left(\mathrm{0,0.01}\right),$ $n_{{\mathrm{\γ}}_4}~N\left(\mathrm{0,0.01}\right),n_{t_w}~N\left(\mathrm{0,0.01}\right).$

The Selecting parameter of MB algorithm is as follows:

$\widehat{{\mathrm{\γ}}_1^0}=577.2;\widehat{{\mathrm{\γ}}_2^0}=1008.0;\widehat{{\mathrm{\γ}}_3^0}=677.0;\widehat{{\mathrm{\γ}}_4^0}=183.7;{\mathrm{\σ}}_{\widehat{{\mathrm{\γ}}_1}}^{0^2}={\mathrm{\σ}}_{\widehat{r_2}}^{0^2}={\mathrm{\σ}}_{\widehat{{\mathrm{\γ}}_3}}^{0^2}={\mathrm{\σ}}_{\widehat{{\mathrm{\γ}}_4}}^{0^2}=0.01;$ N ₁＝10；β ₁＝β ₂＝β ₃＝β ₄＝10000；

When systems are functioning properly, the Output rusults of STF algorithm and MB algorithm is as shown in Fig. 3-4, and the Initial value choice of result surface STF algorithm and MB algorithm is correct.

The list of table 2. test failure

Fault	Failure-description
		F 1	In the k=400 moment, γ 1From 577.2 steps to 519.5 (0.9*577.2)
F 2	In the k=400 moment, γ 2From 1008.0 steps to 705.6 (0.7*1008.0)
		F 3	In the k=400 moment, γ 3From 677.0 steps to 473.9 (0.7*677.0)
F 4	In the k=400 moment, γ 4From 183.7 steps to 128.6 (0.7*183.7)
		F 5	From k=400 to 600 moment, γ 1Gradual to 519.5 (0.9*577.2) with each step minimizing 0.2885 from 577.2
F 6	From k=400 to 600 moment, γ 2Gradual to 705.6 (0.7*1008.0) with each step minimizing 1.5120 from 1008.0
		F 7	From k=400 to 600 moment, γ 3Gradual to 473.9 (0.7*677.0) with each step minimizing 1.0155 from 677.0

F 8	From k=400 to 600 moment, γ 4Gradual to 128.6 (0.7*183.7) with each step minimizing 0.2755 from 183.7
		F 9	In the k=400 moment, γ 1From 577.2 steps to 461.8 (0.8*577.2)

Table 3. fault detection and diagnosis result

Nine kinds of faults of heat exchanger have carried out Fault monitoring and diagnosis, and result is as shown in table 2.Wherein front four faults are mutation failure, the 5th to the 8th soft fault.Fault F9 is more serious than F1 fault degree, by the Detection and diagnosis Comparative result of F9 and F1, to obtain the recognition capability of proposed algorithm to different order of severity fault.Fault detection and diagnosis result is as shown in table 3, and the diagnosis and detection result of all faults is normal, occurs without false-alarm and wrong report phenomenon.

The diagnosis and detection result of fault " F1 " and " F8 " is as shown in Fig. 5-8.Fig. 9 shows that STF and MB algorithm successfully can distinguish fault " F1 " (figure .9 (a)) and " F9 " (figure .9 (b)) of two different faults degree.Figure 10 shows the Comparative result of STF algorithm and two Kalman filtering algorithm (DMF double model filter), and result shows that STF algorithm has better fault detect performance than DMF algorithm.

By analyzing the computer artificial result of STF algorithm and MB algorithm, can draw to draw a conclusion:

1) both phase step fault of heat interchanger can be detected very soon, usually needs fault (see table 4) just can be detected about ten steps after fault occurs.And accurate suspected fault amplitude, then need the more time, in table 4, the estimated accuracy of fault amplitude all reaches more than 93%.

2) imminent fault STF and MB algorithm can in fault early detection out.Such as, F8 fault was detected in the k=489 moment, and real fault amplitude only has 159.2 ≈ 0.867*183.7 simultaneously.If select less detection threshold β i, then can detect that the fault of more glitch amplitude occurs, but the rate of false alarm of diagnosis and detection can be improved therewith.

3) design process of heat interchanger Fault Diagnosis Strategy is very succinct, mainly comprises two parts:

A) state of operation parameter and state estimation algorithm heat exchanger and fault phase related parameter carry out On-line Estimation.

B) by heat interchanger parameter estimation result input value fault detection and diagnosis algorithm (MB algorithm).

Term of the present invention is as shown in table 4.

Table 4. nomenclature

Non-elaborated part of the present invention belongs to the known technology of those skilled in the art.

Claims (4)

1., based on a plane environmental control system heat interchanger method for diagnosing faults of STF and MB, it is characterized in that: the method step is as follows:

The first step: the heat interchanger Fault characteristic parameters based on STF is estimated

On the basis of heat-exchanger model, by the analysis of heat exchanger fault mode, draw heat interchanger Fault characteristic parameters; STF algorithm is with the inlet port temperature of the cold limit of heat interchanger and Re Bian for input, and real-time online estimates heat interchanger Fault characteristic parameters;

Second step: based on the fault detection and diagnosis of MB algorithm

Select to revise Bayesian Classification Arithmetic as heat interchanger fault detection and diagnosis algorithm; Estimate the input that the Fault characteristic parameters drawn is MB algorithm with STF algorithm, draw the fault amplitude of each Fault characteristic parameters, when the fault amplitude of Fault characteristic parameters is greater than the threshold value preset, be judged to be that the fault that this Fault characteristic parameters is corresponding occurs; Concrete fault detection and diagnosis strategy is as follows:

STF algorithm is estimated that all Fault characteristic parameters drawn carry out following operation respectively:

If ● certain Fault characteristic parameters γ _imB arithmetic result be greater than predetermined threshold duration report to the police, then show that the fault corresponding with certain Fault characteristic parameters occurs; Jump to the 3rd step;

● if above-mentioned condition does not meet, then jump to the first step;

3rd step: fault Amplitude Estimation

The Fault characteristic parameters γ reported to the police _icorresponding equivalent fault amplitude EEFA (estimated equivalent fault amplitude) through type (1) calculates,

$\mathrm{EEFA}:{\widehat{\mathrm{\γ}}}_i\left(k\right)-{\mathrm{\γ}}_i^0---\left(1\right)$

Wherein: EEFA is the equivalent fault amplitude of heat interchanger when breaking down, for heat interchanger normal run time fault characteristic parameter γ _iestimated result, for heat interchanger K moment run time fault characteristic parameter γ _iestimated result;

Jump to the first step.

2. a kind of plane environmental control system heat interchanger method for diagnosing faults based on STF and MB according to claim 1, it is characterized in that: described plane environmental control system heat interchanger is the air-air convection recuperator of plate-fin, intersect the pipe passage of 90 degree and heat exchange flat board by cold limit air duct and Re Bian air duct two to form, hot-air and cold air carry out heat exchange by heat exchange manifold and heat exchange flat board; In the actual use of heat interchanger, the entrance on cold limit and hot limit, outlet temperature are measured; Heat interchanger is often revealed, block and fouling fault, and this three classes fault is " parameter error " type fault, and fault signature shows as the change of the immeasurability parameters such as MAF, air valid circulation area, heat transfer coefficient of heat exchanger; Therefore, be difficult to by the entrance on cold limit and hot limit, outlet temperature the fault detection and diagnosis directly carrying out heat interchanger.

3. a kind of plane environmental control system heat interchanger method for diagnosing faults based on STF and MB according to claim 1, it is characterized in that: the first step of described plane environmental control system heat interchanger method for diagnosing faults, with the entrance on the cold limit of heat interchanger measurement parameter and hot limit, outlet temperature for input, adopt a kind of strong tracking filfer based on heat-exchanger model correction (STF), realize heat interchanger state and parametric joint is estimated, real-time online estimates the heat interchanger fault signature that can not directly measure by measurable parameter; Uncertain and Unmarried pregnancy in strong tracking filfer heat exchanger model has stronger robustness, and the gradual or mutation failure characteristic parameter after heat exchanger reaches stable state has very strong tracking power.

4. a kind of plane environmental control system heat interchanger method for diagnosing faults based on STF and MB according to claim 1, it is characterized in that: the second step of described plane environmental control system heat interchanger method for diagnosing faults, the heat interchanger Fault characteristic parameters estimated with strong tracking filfer is input, MB method is adopted to estimate the fault amplitude showing that each Fault characteristic parameters is corresponding in real time, when the fault amplitude of Fault characteristic parameters is greater than the threshold value preset, be judged to be that the fault that this Fault characteristic parameters is corresponding occurs, thus realize effective aircraft environmental control system heat interchanger fault diagnosis.