CN105719007A - Method for failure prediction of infrared hot box audio channel - Google Patents

Abstract

The invention discloses a method for failure prediction of an infrared hot box audio channel.With an infrared hot box channel monitoring and diagnostic system as the basis, a failure prediction module is arranged on a monitoring diagnosis platform to predict a system state and judge whether failures occur in a period of time.The failure prediction module comprises a data processing module, a model building module, a state prediction module and a failure judgment module.The data processing module receives and analyzes data uploaded by a diagnostic device, and analyzes the data according to a protocol and stores the data into a database.The model building module extracts data from the database and uses a grey system theory to build a model.The state prediction module predicts signal characteristics according to a system model.The failure judgment module sets a signal characteristic threshold.According to the method, the system model is built through the grey system theory, the system state is predicted, the failure trend is found timely, preventive measures are taken, the operation and maintenance efficiency is improved, and the traffic safety is guaranteed.

Description

A kind of infra-red heat axle voice-grade channel failure prediction method

Technical field

The present invention relates to THDS failure predication field, be specifically related to a kind of infra-red heat axle voice-grade channel failure prediction method.

Background technology

THDS is widely used in the railway system, utilizes infrared thermometry principle, monitors in real time running vehicle axle temperature.This system effectively prevents vehicle overheating of axle bearing, obtains good result, play a significant role in guarantee driving safety.

Infra-red heat axis channel monitors that diagnostic system monitors THDS duty in real time, during abnormal state, judge that fault occurs at infrared main frame or communication line, has differentiated the responsibility of vehicle department and Office of the Telecommunications Authority in time, reduce and send employee, improve O&M efficiency.

At present, infra-red heat axis channel monitors that it after fault occurs, can only be diagnosed by diagnostic system, it is impossible to the fault that prediction will occur.In real system, equipment or circuit are not break down suddenly sometimes, but progressive failure, if the fault of such class progressive failure can be predicted, it is possible to accomplish to prevent trouble before it happens.

Summary of the invention

For solving prior art Problems existing, the invention provides a kind of infra-red heat axle voice-grade channel failure prediction method, use gray system theory to set up system model, prognoses system state, finds fault trend in time, prevents trouble before it happens, improve O&M efficiency, guarantee driving safety.

The present invention is extension and the extension that infra-red heat axis channel monitors diagnostic system, and this system for content refers to utility model " a kind of infra-red heat axis channel monitors diagnostic system ", the patent No.: ZL201420533301.6.

A kind of infra-red heat axle voice-grade channel failure prediction method, monitor based on diagnostic system by infra-red heat axis channel, monitoring, diagnostic platform arranges failure predication module, prognoses system state, judge whether break down in a period of time, described failure predication module includes data processing module, model building module, state prediction module and fault determination module, the data that described data processing module receiving and analyzing diagnostic device is uploaded, and is resolved according to agreement and stores to data base；Described model building module extracted data from data base, uses gray system theory to set up model；Signal characteristic, according to system model, is predicted by described state prediction module；Described fault determination module arranges signal characteristic threshold, during beyond threshold value, it is determined that system will fault, and by fault message output to display and alarm.

Preferably, it was predicted that the specific works flow process of system mode is as follows:

(1) start

(2) the initial data X of signal characteristic is obtained⁽⁰⁾, it is used for setting up system model:

X⁽⁰⁾=[x⁽⁰⁾(1),x⁽⁰⁾(2),…,x⁽⁰⁾(n)](1)

(3) X is calculated⁽⁰⁾Ordered series of numbers level than λ (k):

$\mathrm{\λ}\left(k\right)=\frac{x^{\left(0\right)}(k-1)}{x^{\left(0\right)}\left(k\right)},k=1,2,...,n---\left(2\right)$

(4) judge whether all λ (k) all can hold covering r in:

$r=(e^{-\frac{2}{n+1}},e^{\frac{2}{n+1}})---\left(3\right)$

(5) if all λ (k) are all holding in covering r, step (7) is forwarded to；

(6) if there is λ (k) not holding in covering r, to X⁽⁰⁾Convert, make all λ (k) all fall within and can hold in covering r, forward step (7) to；

(7) initial data X⁽⁰⁾Or the X after conversion⁽⁰⁾Accumulating generation X⁽¹⁾:

$x^{\left(1\right)}\left(k\right)=\underset{i=1}{\overset{k}{\Σ}}{x}^{\left(0\right)}\left(i\right),k=1,2,...,n---\left(4\right)$

X⁽¹⁾=[x⁽¹⁾(1),x⁽¹⁾(2),…,x⁽¹⁾(n)](5)

(8) again by X⁽¹⁾Next-door neighbour average generation Z⁽¹⁾:

z⁽¹⁾(k)=0.5x⁽¹⁾(k-1)+0.5x⁽¹⁾(k), k=2,3 ..., n (6)

Z⁽¹⁾=[z⁽¹⁾(2),z⁽¹⁾(3),…,z⁽¹⁾(n)](7)

(9) primitive form and the primitive form of GM (1,1) model are drawn:

$\frac{{\mathrm{dx}}^{\left(1\right)}}{dt}+{\mathrm{ax}}^{\left(1\right)}=b---\left(8\right)$

x⁽⁰⁾(k)+az⁽¹⁾(k)=b (9)

(10) development coefficient a and Lycoperdon polymorphum Vitt actuating quantity b is obtained:

$B=\left(\begin{array}{cc}-z^{\left(1\right)}\left(2\right)& 1\\ -z^{\left(1\right)}\left(3\right)& 1\\ ...& ...\\ -z^{\left(1\right)}\left(n\right)& 1\end{array}\right)---\left(10\right)$

Y=[x⁽⁰⁾(2),x⁽⁰⁾(3),…,x⁽⁰⁾(n)](11)

(a,b)^T=(B^TB)^-1·B^T·Y(12)

(11) x is obtained⁽¹⁾And x (k+1)⁽⁰⁾(k+1) expression formula:

$\left\{\begin{array}{c}{x}^{\left(1\right)}(k+1)=\left(x^{\left(0\right)}\right(1)-\frac{b}{a})\·e^{-ak}+\frac{b}{a}\\ x^{\left(0\right)}(k+1)=x^{\left(1\right)}(k+1)-x^{\left(1\right)}\left(k\right)\end{array}\right.---\left(13\right)$

(12) relevant parameter, checking system model are calculated；

(13) whether judgment models precision meets system prediction requirement；

(14) if model accuracy meets system prediction requirement, step (16) is forwarded to；

(15) if model accuracy can not meet system prediction requirement, residual error e is used⁽⁰⁾Correction model, makes model accuracy meet system prediction requirement, forwards step (16) to；

(16) use Model forecast system state, forward step (2) to.

Preferably, to X in step (6)⁽⁰⁾Converting, make all λ (k) all fall within and can hold in covering r, described conversion includes translation, logarithm and root conversion.

Preferably, by calculating relevant parameter checking system model in step (12), described relevant parameter is average relative error, mean square deviation ratio or small probability error.

Preferably, for initial data X⁽⁰⁾, supplement fresh information as required or regularly, remove old information, set up metabolism predicting model.

Preferably, described signal characteristic is: line signal frequency shift (FS), peak-to-peak value or signal to noise ratio are predicted；Initial data X⁽⁰⁾For: line signal frequency offset data, peak-to-peak value data or signal-to-noise ratio data.

Preferably, described fault determination module arranges signal characteristic threshold, when two or three signal characteristics are beyond threshold value, it is determined that system will fault, and by fault message output to display and alarm.

Beneficial effect:

The invention provides a kind of infra-red heat axle voice-grade channel failure prediction method, use gray system theory to set up system model, it was predicted that system mode, find fault trend in time, prevent trouble before it happens, improve O&M efficiency, guarantee driving safety.

Accompanying drawing explanation

Fig. 1 is present system structural representation；

Fig. 2 is failure predication functions of modules schematic diagram of the present invention；

Fig. 3 is status predication flow chart of the present invention.

Detailed description of the invention

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described further.The present embodiment is carried out premised on technical solution of the present invention, but protection scope of the present invention is not limited to following embodiment.

As it is shown in figure 1, THDS includes acquisition station, voice-grade channel, monitoring center and monitoring room, acquisition station is provided with infrared main frame, and infrared main frame connects voice-grade channel by modem；Voice-grade channel is provided with rosette and distributing frame, and rosette connects acquisition station, and distributing frame accesses monitoring center；Monitoring center is provided with modem, and monitoring room is provided with monitoring host computer, and distributing frame connects monitoring host computer by modem.Infra-red heat axis channel monitors that diagnostic system includes monitoring terminal, analyzing and diagnosing equipment and monitors diagnostic platform.

Monitoring terminal is deployed in acquisition station, connects voice-grade channel, it is monitored；Analyzing and diagnosing deployed with devices, in monitoring center, accesses voice-grade channel, and is connected with monitoring network；Monitor that diagnostic platform is deployed in monitoring room, be connected with monitoring network.

Failure predication module is the important component part monitoring diagnostic platform, it was predicted that system mode, it is determined that whether break down in a period of time.

As in figure 2 it is shown, failure predication module includes data processing module, model building module, state prediction module and fault determination module.In a preferred embodiment:

Data processing module is stored three parts by data acquisition, data parsing and data and forms, the data that receiving and analyzing diagnostic device is uploaded, and is resolved according to corresponding agreement, stores afterwards to data base.

Model building module extracts the data of 5 days in the past from data base, uses gray system theory to set up system model, for failure predication.

The line signal frequency shift (FS)s of following 5 days, peak-to-peak value and signal to noise ratio, according to system model, are predicted by state prediction module.

Fault determination module is respectively to three kinds of signal characteristic set point, and frequency deviation range is ± 40Hz, and peak-to-peak value ranges for 100 ± 20mV, and SNR ranges is more than 70dB.When there being two or three signal characteristics to go beyond the scope, it is determined that system will fault, and by fault message output to display and alarm.

As it is shown on figure 3, the job step of status predication is as follows:

(1) start

(2) initial data X is obtained⁽⁰⁾:

X⁽⁰⁾=[x⁽⁰⁾(1),x⁽⁰⁾(2),…,x⁽⁰⁾(n)](1)

(3) X is calculated⁽⁰⁾Ordered series of numbers level than λ (k):

$\mathrm{\λ}\left(k\right)=\frac{x^{\left(0\right)}(k-1)}{x^{\left(0\right)}\left(k\right)},k=1,2,...,n---\left(2\right)$

(4) judge whether all λ (k) all can hold covering r in:

$r=(e^{-\frac{2}{n+1}},e^{\frac{2}{n+1}})---\left(3\right)$

(5) if all λ (k) are all holding in covering r, step (6) is forwarded to；

(6) if there is λ (k) not holding in covering r, to X⁽⁰⁾Carry out translating, logarithm or root conversion, make all λ (k) all fall within and can hold in covering, forward step (6) to；

(7) initial data X⁽⁰⁾Accumulating generation X⁽¹⁾:

$x^{\left(1\right)}\left(k\right)=\underset{i=1}{\overset{k}{\Σ}}{x}^{\left(0\right)}\left(i\right),k=1,2,...,n---\left(4\right)$

X⁽¹⁾=[x⁽¹⁾(1),x⁽¹⁾(2),…,x⁽¹⁾(n)](5)

(8) again by X⁽¹⁾Next-door neighbour average generation Z⁽¹⁾:

z⁽¹⁾(k)=0.5x⁽¹⁾(k-1)+0.5x⁽¹⁾(k), k=2,3 ..., n (6)

Z⁽¹⁾=[z⁽¹⁾(2),z⁽¹⁾(3),…,z⁽¹⁾(n)](7)

(9) primitive form and the primitive form of GM (1,1) model are drawn:

$\frac{{\mathrm{dx}}^{\left(1\right)}}{dt}+{\mathrm{ax}}^{\left(1\right)}=b---\left(8\right)$

x⁽⁰⁾(k)+az⁽¹⁾(k)=b (9)

(10) development coefficient a and Lycoperdon polymorphum Vitt actuating quantity b is obtained:

$B=\left(\begin{array}{cc}-z^{\left(1\right)}\left(2\right)& 1\\ -z^{\left(1\right)}\left(3\right)& 1\\ ...& ...\\ -z^{\left(1\right)}\left(n\right)& 1\end{array}\right)---\left(10\right)$

Y=[x⁽⁰⁾(2),x⁽⁰⁾(3),…,x⁽⁰⁾(n)](11)

(a,b)^T=(B^TB)-¹·B^T·Y(12)

(11) x is obtained⁽¹⁾And x (k+1)⁽⁰⁾(k+1) expression formula:

$\left\{\begin{array}{c}{x}^{\left(1\right)}(k+1)=\left(x^{\left(0\right)}\right(1)-\frac{b}{a})\·e^{-ak}+\frac{b}{a}\\ x^{\left(0\right)}(k+1)=x^{\left(1\right)}(k+1)-x^{\left(1\right)}\left(k\right)\end{array}\right.---\left(13\right)$

(12) average relative error, mean square deviation ratio or small probability error, checking system model are calculated；

(13) whether judgment models precision meets requirement；

(14) if model accuracy meets requirement, step (15) is forwarded to；

(15) if model accuracy can not meet requirement, residual error e is used⁽⁰⁾Correction model, makes model accuracy meet requirement, forwards step (15) to；

(16) use Model forecast system state, forward step (1) to.

In the present embodiment, take frequency shift (FS) initial data X⁽⁰⁾For:

X⁽⁰⁾=(x⁽⁰⁾(1),x⁽⁰⁾(2),x⁽⁰⁾(3),x⁽⁰⁾(4),x⁽⁰⁾(5))

=(29,33,33,34,37) (14)

All λ (k) are all holding in covering r, by X⁽⁰⁾Accumulating generation X⁽¹⁾:

X⁽¹⁾=(x⁽¹⁾(1),x⁽¹⁾(2),x⁽¹⁾(3),x⁽¹⁾(4),x⁽¹⁾(5))

=(29,62,95,129,166) (15)

Again by X⁽¹⁾Next-door neighbour average generation Z⁽¹⁾:

Z⁽¹⁾=(z⁽¹⁾(2),z⁽¹⁾(3),z⁽¹⁾(4),z⁽¹⁾(5))

=(46,79,112,148) (16)

Obtain development coefficient a and Lycoperdon polymorphum Vitt actuating quantity b:

$B=\left(\begin{array}{cc}-z^{\left(1\right)}\left(2\right)& 1\\ -z^{\left(1\right)}\left(3\right)& 1\\ -z^{\left(1\right)}\left(4\right)& 1\\ -z^{\left(1\right)}\left(5\right)& 1\end{array}\right)=\left(\begin{array}{cc}-46& 1\\ -79& 1\\ -112& 1\\ -148& 1\end{array}\right)---\left(17\right)$

$Y=\left(\begin{array}{c}{x}^{\left(0\right)}\left(2\right)\\ x^{\left(0\right)}\left(3\right)\\ x^{\left(0\right)}\left(4\right)\\ x^{\left(0\right)}\left(5\right)\end{array}\right)=\left(\begin{array}{c}33\\ 33\\ 34\\ 37\end{array}\right)---\left(18\right)$

${(a,b)}^T={\left(B^{T}B\right)}^{-1}\·B^T\·Y=\left(\begin{array}{c}-0.0372\\ 30.6536\end{array}\right)---\left(19\right)$

Then the primitive form of GM (1,1) model and primitive form are:

$\frac{{\mathrm{dx}}^{\left(1\right)}}{dt}-0.0372x^{\left(1\right)}=30.6536---\left(20\right)$

x⁽⁰⁾(k)-0.0372z⁽¹⁾(k)=30.6536 (21)

So x⁽¹⁾(k+1) expression formula is:

x⁽¹⁾(k+1)=852.7615e^0.0372k-824.0215(22)

By x⁽¹⁾(k+1) x can be obtained⁽⁰⁾(k+1)。

Draw X⁽⁰⁾The analogue value, calculate relative error Δ_k, can obtain average relative error further is:

$\frac{1}{4}\underset{k=2}{\overset{5}{\Σ}}{\mathrm{\Δ}}_k=1.61\%---\left(23\right)$

Model accuracy meets system prediction requirement, this model prediction frequency shift (FS) available.

The frequency shift (FS) of following 5 days is:

(x⁽⁰⁾(6),x⁽⁰⁾(7),x⁽⁰⁾(8),x⁽⁰⁾(9),x⁽⁰⁾(10))=(37,39,40,42,44) (24)

After known 3 days, frequency shift (FS) will beyond former setting threshold value 40Hz.

In like manner, line signal peak-to-peak value and signal to noise ratio are predicted.When there being two or three signal characteristics to go beyond the scope, it is determined that system will fault.

In sum, the invention provides a kind of infra-red heat axle voice-grade channel failure prediction method, use gray system theory to set up system model, it was predicted that system mode, find fault trend in time, prevent trouble before it happens, improve O&M efficiency, guarantee driving safety.

The present invention has passed through above-described embodiment and accompanying drawing illustrates clear; without departing from the spirit and substance of the case in the present invention; those skilled in the art can make respective change and correction according to the present invention, and these variations and modifications broadly fall into the protection domain of the claims in the present invention.

The present invention is not directed to that method is all same as the prior art maybe can adopt prior art to be realized.

Claims (7)

1. an infra-red heat axle voice-grade channel failure prediction method, it is characterized in that, monitor based on diagnostic system by infra-red heat axis channel, monitoring, diagnostic platform arranges failure predication module, prognoses system state, judge whether break down in a period of time, described failure predication module includes data processing module, model building module, state prediction module and fault determination module, the data that described data processing module receiving and analyzing diagnostic device is uploaded, are resolved according to agreement and are stored to data base；Described model building module extracted data from data base, uses gray system theory to set up model；Signal characteristic, according to system model, is predicted by described state prediction module；Described fault determination module arranges signal characteristic threshold, during beyond threshold value, it is determined that system will fault, and by fault message output to display and alarm.

2. a kind of infra-red heat axle voice-grade channel failure prediction method according to claim 1, it is characterised in that the specific works flow process of prognoses system state is as follows:

(1) start

(2) the initial data X of signal characteristic is obtained⁽⁰⁾, it is used for setting up system model:

X⁽⁰⁾=[x⁽⁰⁾(1),x⁽⁰⁾(2),…,x⁽⁰⁾(n)](1)

(3) X is calculated⁽⁰⁾Ordered series of numbers level than λ (k):

$\mathrm{\λ}\left(k\right)=\frac{x^{\left(0\right)}(k-1)}{x^{\left(0\right)}\left(k\right)},k=1,2,...,n---\left(2\right)$

(4) judge whether all λ (k) all can hold covering r in:

$r=(e^{-\frac{2}{n+1}},e^{\frac{2}{n+1}})---\left(3\right)$

(5) if all λ (k) are all holding in covering r, step (7) is forwarded to；

(6) if there is λ (k) not holding in covering r, to X⁽⁰⁾Convert, make all λ (k) all fall within and can hold in covering r, forward step (7) to；

(7) initial data X⁽⁰⁾Or the X after conversion⁽⁰⁾Accumulating generation X⁽¹⁾:

$x^{\left(1\right)}\left(k\right)=\underset{i=1}{\overset{k}{\Σ}}{x}^{\left(0\right)}\left(i\right),k=1,2,...,n---\left(4\right)$

X⁽¹⁾=[x⁽¹⁾(1),x⁽¹⁾(2),…,x⁽¹⁾(n)](5)

(8) again by X⁽¹⁾Next-door neighbour average generation Z⁽¹⁾:

z⁽¹⁾(k)=0.5x⁽¹⁾(k-1)+0.5x⁽¹⁾(k), k=2,3 ..., n (6)

Z⁽¹⁾=[z⁽¹⁾(2),z⁽¹⁾(3),…,z⁽¹⁾(n)](7)

(9) primitive form and the primitive form of GM (1,1) model are drawn:

$\frac{{\mathrm{dx}}^{\left(1\right)}}{dt}+{\mathrm{ax}}^{\left(1\right)}=b---\left(8\right)$

x⁽⁰⁾(k)+az⁽¹⁾(k)=b (9)

(10) development coefficient a and Lycoperdon polymorphum Vitt actuating quantity b is obtained:

$B=\left(\begin{array}{cc}-z^{\left(1\right)}\left(2\right)& 1\\ -z^{\left(1\right)}\left(3\right)& 1\\ ...& ...\\ -z^{\left(1\right)}\left(n\right)& 1\end{array}\right)---\left(10\right)$

Y=[x⁽⁰⁾(2),x⁽⁰⁾(3),…,x⁽⁰⁾(n)](11)

(a,b)^T=(B^TB)^-1·B^T·Y(12)

(11) x is obtained⁽¹⁾And x (k+1)⁽⁰⁾(k+1) expression formula:

$\left\{\begin{array}{c}{x}^{\left(1\right)}(k+1)=\left(x^{\left(0\right)}\right(1)-\frac{b}{a})\·e^{-ak}+\frac{b}{a}\\ x^{\left(0\right)}(k+1)=x^{\left(1\right)}(k+1)-x^{\left(1\right)}\left(k\right)\end{array}\right.---\left(13\right)$

(12) relevant parameter, checking system model are calculated；

(13) whether judgment models precision meets system prediction requirement；

(14) if model accuracy meets system prediction requirement, step (16) is forwarded to；

(15) if model accuracy can not meet system prediction requirement, residual error e is used⁽⁰⁾Correction model, makes model accuracy meet system prediction requirement, forwards step (16) to；

(16) use Model forecast system state, forward step (2) to.

3. a kind of infra-red heat axle voice-grade channel failure prediction method according to claim 2, it is characterised in that to X in step (6)⁽⁰⁾Converting, make all λ (k) all fall within and can hold in covering r, described conversion includes translation, logarithm and root conversion.

4. a kind of infra-red heat axle voice-grade channel failure prediction method according to claim 2, it is characterized in that, by calculating relevant parameter checking system model in step (12), described relevant parameter is average relative error, mean square deviation ratio or small probability error.

5. a kind of infra-red heat axle voice-grade channel failure prediction method according to claim 2, it is characterised in that for initial data X⁽⁰⁾, supplement fresh information as required or regularly, remove old information, set up metabolism predicting model.

6. a kind of infra-red heat axle voice-grade channel failure prediction method according to any one of Claims 1 to 5, it is characterised in that described signal characteristic is: line signal frequency shift (FS), peak-to-peak value or signal to noise ratio are predicted；Initial data X⁽⁰⁾For: line signal frequency offset data, peak-to-peak value data or signal-to-noise ratio data.

7. a kind of infra-red heat axle voice-grade channel failure prediction method according to claim 6, it is characterized in that, described fault determination module arranges signal characteristic threshold, when two or three signal characteristics are beyond threshold value, decision-making system will fault, and by fault message output to display and alarm.