CN102072829B - Iron and steel continuous casting equipment oriented method and device for forecasting faults - Google Patents

Abstract

The invention relates to an iron and steel continuous casting equipment oriented method and device for forecasting faults. The device is characterized in that sensors installed on iron and steel continuous casting equipment are used for acquiring real-time status parameters of the equipment; the signals are processed by signal conditioning circuits, amplifying and filtering circuits and A/D conversion modules of data acquisition nodes and the data are sent to handheld spot inspection equipment via radio circuits supporting the Zigbee protocols; the handheld spot inspection equipment receives and analyzes the data; eigenvectors of equipment status data are constructed based on decomposition and construction carried out by wavelet packets, the eigenvectors are used as samples, support vector data description is utilized to train a single-value classifier, and the domesticated classifier is utilized to compute the health index of the equipment and describe the deterioration curve of the equipment so as to further forecast the residual service life of the equipment; the data acquisition node main boards comprise the sensors, the signal conditioning circuits, the amplifying and filtering circuits, single chips and antenna circuits; and the handheld spot inspection equipment comprises a radio communication module, an advanced RISC machines (ARM) minimum system, a keyboard and a display module. The method and the device have the beneficial effects of adopting the radio communication mode and abandoning the traditional industrial buses, being applied in the harsh production equipment such as iron and steel continuous casting lines, reducing the trouble of arranging lines and lowering the application cost.

Description

A kind of failure prediction method and device towards continuous casting of iron and steel equipment

Technical field

The invention belongs to mechanical fault diagnosis and electric powder prediction, be specifically related to a kind of failure prediction method towards continuous casting of iron and steel equipment and device.

Background technology

Domestic manufacturing management of equipment maintenance demand for services is constantly soaring.Press for research and development towards the MRO back-up system of Large-Scale Equipment, for equipment manufacturing enterprise, equipment user enterprise and equipment service enterprise provide comprehensive digitizing solution and information-based integrated technology, promote to make the service sector great-leap-forward development.At present, domestic continuous casting of iron and steel production line still adopts the service mode of correction maintenance and scheduled maintenance, and the maintenance cost height is easy to generate the equipment non-programmed halt, causes tremendous loss.The tradition maintenance mode is difficult to satisfy the requirement of maintenance cost and efficient, needs the more reasonable monitoring and diagnosis of equipment running status efficiently technology badly and reduces the downtime, reduces operation and maintenance cost.

The research of mechanical fault being carried out monitoring, diagnosing has had many achievements, a kind of rotating machinery fault intelligent diagnosing method and the device announced as University Of Chongqing's CN1514209A patent, the disclosed a kind of low-speed heavy-load rotary machinery fault diagnosis method of Shenyang University of Technology's CN1776390 patent No., the disclosed a kind of rotary machinery fault diagnosis of Shanghai Communications University's CN2826392 patent and analysis experimental provision etc.Their common feature all is the research to rotary machinery fault diagnosis method and device, does not carry out the field failure prediction.And Forecasting Methodology and device based on gray theory that the patent of invention CN101201295A of Shanghai Maritime University announces reach prediction rotating machinery future work situation by following each moment rotating machinery vibrating displacement numerical value of gray prediction principle prediction.For continuous casting of iron and steel equipment, the status information parameter of equipment is many, and existing invention is only analyzed the vibration signal of rotating machinery, can not grasp the status information of equipment comprehensively, can not make more comprehensive failure prediction.On the other hand, for a lot of production environment complexity, abominable occasion, by wired mode collecting device status data and to carry out real-time estimate very difficult.Moreover existing invention only is used for fault diagnosis and prediction with the data of gathering, and is not equipment control and production management service and data are stored and uploaded.

Summary of the invention

One of purpose of the present invention is to provide a kind of intelligent trouble Forecasting Methodology towards continuous casting of iron and steel equipment.This method adopts WAVELET PACKET DECOMPOSITION and reconstruct equipment state signal and construction feature vector, with the proper vector training SVDD monodrome sorter that obtains, by the SVDD monodrome sorter of domestication.

Two of purpose of the present invention is to provide a kind of intelligent trouble prediction unit towards steel equipment, between data acquisition equipment and handheld device, adopt radio communication to carry out data transmission, solved the on-the-spot abominable wiring condition problem of commercial production, the dirigibility that has improved spot check.

A kind of failure prediction method towards continuous casting of iron and steel equipment that the present invention proposes, concrete steps are as follows:

When (1) carrying out spot check, wake the data acquisition node that is installed on monitoring of equipment point up by hand-held spot check equipment, begin image data by data acquisition channel is set;

(2) device status data of data acquisition node front end sensors collection after isolation, conversion and filtering are handled, obtains corresponding voltage signal;

(3) voltage signal through sampling and analog to digital conversion, obtains corresponding digital signal at data acquisition node;

(4) the digital signal packing of encoding in data acquisition node sends by less radio-frequency then;

(5) coded data that sends of hand-held spot check equipment receiving data acquisition node is decoded automatically and is stored;

(6) select operation to image data by the function menu of hand-held spot check equipment, carry out fault analysis and life prediction, send instruction simultaneously and change acquisition channel or stop to gather and makes the data acquisition node dormancy;

(7) prediction result is presented on the display module of hand-held spot check equipment.

Among the present invention, fault analysis uses the WAVELET PACKET DECOMPOSITION technology to carry out obtaining of feature extraction, the domestication of SVDD sorter and equipment degenerated curve in the step (6), and concrete steps are as follows:

(1) utilize the average μ of original signal x (t) and standard deviation sigma standardization x (t),

(2) utilize WAVELET PACKET DECOMPOSITION that the signal x (t) ' after standardizing is carried out the j layer and decompose, obtain 2 ^jThe reconstruction signal f of individual different frequency bands _i(t) (i=1 ..., 2 ^j);

(3) calculate each frequency band reconstruction signal f _i(t) energy adopts method for normalizing that each frequency band reconstruction signal is handled, and namely the mark that accounts for the signal gross energy with the energy of each frequency band reconstruction signal is represented, obtains the relative energy of each reconstruction signal

I=1,2 ..., 2 ^j, have

(4) utilize the relative energy E of each the frequency band reconstruction signal after the normalization _j(i) as parameter, structural attitude vector T=[E _j(1), E _j(2) ..., E _j(2 ^j)];

(5) status data with 1000 groups of good equipment is sample, utilizes the proper vector of constructing in the step (4) as learning sample, and training SVDD monodrome sorter obtains among the decision function χ (z) and sample corresponding parameters α _iValue and the value of minimum hypersphere radius R;

(6) with the sampled data of equipment Life cycle for people Г=χ (z)-R, if Г≤0, then health indicator HI=0; Otherwise HI=Г;

(7) with the health indicator HI rendering apparatus performance degradation curve of the equipment Life cycle that obtains;

(8) fault analysis and life prediction: the data of gathering in the spot check process are judged for people (6) in the step, obtained the health indicator HI when time spot check, this desired value and the equipment degenerated curve that obtains are compared the equipment residual life that obtains predicting.

A kind of device towards continuous casting of iron and steel equipment failure prediction that the present invention proposes, formed by data acquisition node mainboard 2 and hand-held spot check equipment 3, data acquisition node mainboard 2 is by sensor 1, signal conditioning circuit 4, amplification filtering circuit 5, single-chip microcomputer 6, antenna circuit 7 and first power module 8 are formed, be connected by industrial cable between sensor 1 and the signal conditioning circuit 4, signal conditioning circuit 4 connects amplification filtering circuit 5, amplification filtering circuit 4 is connected with the data port of single-chip microcomputer 6, and power module 8 connects sensor 1 respectively, signal conditioning circuit 4, amplification filtering circuit 5, single-chip microcomputer 6 and antenna circuit 7; Hand-held spot check equipment 3 is made up of keyboard 9, wireless communication module 10, second source module 11, communication interface 12, interface module 13, display module 14 and ARM minimum system 15, keyboard 9, wireless communication module 10, second source module 11, communication interface 12 and display module 14 are connected with interface module 13 respectively, and the other end of interface module 13 is connected with the data port of ARM minimum system 15.

Among the present invention, described sensor 1 is one to multiple kind in temperature sensor, vibration transducer, pressure transducer, electromagnetic sensor or the eddy current inductor, temperature sensor is converted to corresponding voltage signal with temperature signal, vibration transducer with vibration acceleration signal be converted to voltage signal, pressure transducer is converted to voltage signal with pressure signal, electromagnetic flow transducer is converted to voltage signal with flow signal.Each sensor is installed on the monitoring point of online equipment, and links to each other by the mainboard of cable in data acquisition node.Different sensors links to each other with the respective signal modulate circuit, and the signal of each process conditioning is as the input of multi-way switch, and the output of multi-way switch links to each other with single-chip microcomputer 6 through signal conditioning circuit 2 output terminals.

Among the present invention, described signal conditioning circuit is connected to form successively by electric capacity buffer circuit, amplifier circuit and RC anti-aliasing filter circuit.

Among the present invention, described single-chip microcomputer 6 adopts the wireless singlechip of built-in A/D converter and wireless radio frequency circuit.

Among the present invention, described ARM minimum system 15 arranges module 17, data analysis module 18, failure prediction module 19, task management module 20, data derivation module 21 and serial communication modular 22 by data acquisition module 16, system and forms, data acquisition module 16 is communicated by letter with wireless radio frequency circuit with the antenna circuit on the data acquisition node mainboard by the Zigbee agreement, the duty of control data acquisition module 19 arranges sampling monitoring point, sample frequency, sampling time and store sample data; System arranges system time, display brightness of 17 pairs of hand-held spot check equipment 3 of module etc. and arranges; Data analysis module 18 is realized data management, data analysis and failure prediction; Data management comprises configuration information and the deletion data file of checking measuring point path, parameter; Spot check task in the hand-held spot check equipment 3 of task management module 20 management, measuring point information, execution time, the parameter of setting, the spot check result of each spot check task; Data derive module 21 and by communication interface 12 data are exported to USB flash disk or other memory devices; Serial communication modular 22 receives the spot check task that host computers assign and uploads data to host computer by communication interface 12.

The present invention adopts communication to abandon traditional industrial bus, can be applied in the harsh production environment such as continuous casting of iron and steel line, has reduced the trouble of wiring, has reduced application cost.Simultaneously, the use of hand-held spot check equipment can be carried out fault judgement and the remaining life prediction of equipment (or spare part) in the very first time, and on top of the presence of equipment can be formulated maintenance schedule under the situation that guarantees largest production.On the other hand, the data after the raw data of hand-held spot check equipment collection and the analysis can be used as the historical data of production management system and are used for the backing up maintenance decision-making.

Description of drawings

Fig. 1 is the one-piece construction block diagram of fault prognoses system.

Fig. 2 is the structural diagrams of data acquisition node mainboard 2.

Fig. 3 is the structural diagrams of ARM minimum system 15.

Fig. 4 is the structural diagrams of hand-held spot check equipment 3.

Fig. 5 is the schematic diagram of WAVELET PACKET DECOMPOSITION.

Fig. 6 is the data space distribution plan of Support Vector data description.

Fig. 7 is the process flow diagram of fault diagnosis and Forecasting Methodology.

Fig. 8 is the workflow diagram of fault prognoses system.

Number in the figure: 1 is sensor, and 2 is the data acquisition node mainboard, and 3 are hand-held spot check equipment, 4 is signal conditioning circuit, and 5 is the amplification filtering circuit, and 6 is single-chip microcomputer, 7 is antenna circuit, 8 is first power module, and 9 is keyboard, and 10 is wireless communication module, 11 is the second source module, 12 is communication interface, and 13 is interface module, and 14 is display module, 15 is the ARM minimum system, 16 is data acquisition module, and 17 for system arranges module, and 18 is data analysis module, 19 is the failure prediction module, 20 are the task management module, and 21 for data derive module, and 22 is serial communication modular.

Embodiment

For making the present invention be easy to understand and implement, be further elaborated below in conjunction with accompanying drawing.

Embodiment 1:

As shown in Figure 1, a kind of device of the failure prediction method towards continuous casting of iron and steel equipment is made up of sensor 1 (acceleration vibration transducer, thermocouple temperature sensor, electromagnetic flow transducer, electromagnetic pressure sensor and eddy current inductor), data acquisition node mainboard 2 and the hand-held spot check equipment 3 of gathering the various state parameters of continuous casting installation for casting.The vibration displacement of acceleration vibration transducer monitoring bearing in rotating machinery, the temperature of thermocouple temperature sensor monitoring equipment difference, the flow of electromagnetic flow transducer monitoring chilled water, electromagnetic pressure sensor monitoring oil-air lubrication terminal pressure, eddy current inductor monitoring current of electric.The voltage signal of the output correspondence of each sensor, and insert data acquisition node mainboard 2; Data acquisition node is to be the microcomputer system of core with the CC2430 single-chip microcomputer, to the voltage signal of input amplify, pre-service such as filtering, A/D conversion, data packing, by Zigbee agreement wireless transmission data; Hand-held spot check equipment 3 receives data, analyzes packet.The vibration signal of equipment and the current signal of motor are used for failure prediction, and other signals are used for fault diagnosis.Vibration signal and current signal to the different time sequence in hand-held spot check equipment 3 carry out WAVELET PACKET DECOMPOSITION, obtain the energy function of different frequency range, each parameter vector with relative energy structural attitude vector domestication Support Vector data description, degradation curve by structure equipment, by the vectorial reasoning of spot check data being obtained the degradation of equipment, thus the remaining life of predict device.In hand-held spot check equipment 3, temperature, flow and pressure signal are carried out threshold decision, thus the malfunction of determining apparatus.

As shown in Figure 2, data acquisition node mainboard 2 comprises sensor 1, signal conditioning circuit 4, amplification filtering circuit 5, single-chip microcomputer 6, antenna circuit 7 and first power module 8.The mainboard of monitoring current signal is converted to voltage signal by signal conditioning circuit 4 with current signal.Voltage signal is as the input of amplification filtering circuit 5, and the output of amplification filtering circuit 5 is as the input of single-chip microcomputer 6.In the single-chip microcomputer central processing unit select for use TI company integrated the CC2430 single-chip microcomputer of A/D modular converter and wireless radio frequency circuit.CC2430 is the well-designed low-consumption wireless single-chip microcomputer of Chipcon, is respectively 50Mw and 54mW in transmission and accepting state energy consumption.This module provides the 2.4GHz high frequency less radio-frequency with IEEE 802.13.4 protocol-compliant.The wireless radio frequency modules of CC2430 is added peripheral antenna and is realized radio-frequency receiving-transmitting.First power module 8 adopts battery to give whole data collection node mainboard 2 and sensor 1 power supply.From the signal of sensor 1, by signal amplification, filtering and A/D conversion process, under the communication drivers programmed control, transmit the device status data of gathering by radio-circuit 7 to hand-held spot check equipment 3.

As shown in Figure 3, hand-held spot check equipment 3 comprises ARM minimum system 15, DIM interface module, wireless communication module 10, second source module 11, display module 14, keyboard 9, RS232 serial communication interface, USB interface, gauge tap.Software systems in the ARM minimum system 15 arrange module 17, data analysis module 18, failure prediction module 19, task management module 20, data derivation module 21 and serial communication modular 22 by data acquisition module 16, system and form, the serial communication modular 22 of ARM minimum system 15 is connected with host computer by the RS232 serial communication port, downloads the spot check task or uploads the spot check data; ARM minimum system 15 utilizes wireless communication module 10 to communicate by letter with data acquisition node mainboard 2, collects data to data acquisition node transmit operation instruction or from data acquisition node.An input end of ARM minimum system 15 is that 9, one output terminals of keyboard are display module 14, and display module 14 can adopt LCD display.The second source module 11 of hand-held spot check equipment is given whole hand-held spot check equipment 3 power supplies, use be chargeable lithium cell.

In hand-held spot check equipment, realize the failure prediction of WAVELET PACKET DECOMPOSITION and Support Vector data description (SVDD).WAVELET PACKET DECOMPOSITION is mainly used to extract eigenwert, the structural attitude vector of signal, and SVDD is mainly used to construct the equipment degradation curve and carries out predicting residual useful life.

1. WAVELET PACKET DECOMPOSITION and reconstruct:

According to different scale factor j the Hilbert Space L ²(R) be decomposed into all subspace W _jThe quadrature of (j ∈ Z) and.W wherein _jFor the closure (wavelets Subspace) of wavelet function Ψ (t), carry out the segmentation of frequency according to scale-of-two fraction form.If { h _k} _{K ∈ Z}Be the real wave filter of the corresponding quadrature low pass of quadrature scaling function φ (t), { g _k} _{K ∈ Z}It is the corresponding Hi-pass filter of orthogonal wavelet function ψ (t), wherein g (k)=(1) ⁿH (1-k), then their two yardstick equations and little wave equation are defined as:

$\left\{\begin{array}{c}\mathrm{\φ}\left(t\right)=\sqrt{2}\underset{k\∈Z}{\mathrm{\Σ}}h\left(k\right)u_n(2t-k)\\ \mathrm{\ψ}\left(t\right)=\sqrt{2}\underset{k\∈Z}{\mathrm{\Σ}}g\left(k\right)u_n(2t-k)\end{array}\right.---\left(1\right)$

Yardstick subspace V _jWith wavelets Subspace W _jUse U _jUnified expression.If order

$\left\{\begin{array}{c}{U}_j^0=V_j,j\&Element;\mathrm{<figure-callout\; id="1"\; label="Z\; U\; j"\; filenames="HSA00000331943500011.png"\; state="\{\{state\}\}">Z</figure-callout>}& \\ U_j^{}{}_1^{}=W_j,j\&Element;Z\end{array}---\left(2\right)\right.$

Then the quadrature in Hilbert space decomposes Can use

The decomposition unification be

$U_{j+1}^0=U_j^0\&CirclePlus;{\mathrm{<figure-callout\; id="1"\; label="U\; j"\; filenames="HSA00000331943500011.png"\; state="\{\{state\}\}">U</figure-callout>}}_j^{},j\&Element;Z---\left(3\right)$

The definition subspace

Be function u _n(t) closure space, and Be function u _2n(t) closure space, and make u _n(t) satisfy two yardstick equations

$\left\{\begin{array}{c}{u}_{2n}\left(t\right)=\sqrt{2}\underset{k\&Element;Z}{\mathrm{\&Sigma;}}h\left(k\right)u_n(2t-k)\\ u_{2n}\left(t\right)=\sqrt{2}\underset{k\&Element;Z}{\mathrm{\&Sigma;}}g\left(k\right)u_n(2t-k)\end{array}\right.---\left(4\right)$

Function u by the following formula recursive definition _n, n=0,1,2 ... be called definite wavelet packet by quadrature scaling function φ (t).

WAVELET PACKET DECOMPOSITION

$\left\{\begin{array}{c}{d}_k^{2n}=\frac{1}{2}\underset{l,k\&Element;Z}{\mathrm{\&Sigma;}}{d}_1^n{h}_{l-2k}\\ d_k^{2n+1}=\frac{1}{2}\underset{l,k\&Element;Z}{\mathrm{\&Sigma;}}{d}_1^n{g}_{l-2k}\end{array}\right.---\left(5\right)$

Wavelet package reconstruction

$d_k^{2n}=\underset{l.k\&Element;Z}{\mathrm{\&Sigma;}}{d}_l^{2n}{h}_{k-2l}+\underset{l,k\&Element;Z}{\mathrm{\&Sigma;}}{d}_l^{2n+1}{g}_{k-2l}---\left(6\right)$

Wherein

Be respectively signal in the subspace With

On wavelet packet coefficient.

Wavelet-packet energy

According to formula signal is carried out j layer WAVELET PACKET DECOMPOSITION, the end layer obtains 2 ^jIndividual frequency band, the signal decomposition coefficient of extraction each frequency content from the low frequency to the high frequency.According to formula each frequency band small echo is reconstructed, establishing reconstruction signal is f _i(t) (i=1 ..., 2 ^j).The resultant signal of reconstruct is

$f\left(t\right)=\underset{i=1}{\overset{2^j}{\mathrm{\&Sigma;}}}{f}_i\left(t\right).---\left(7\right)$

If A _kBe the amplitude of each discrete point of reconstruction signal f (t), n for the number of reconstruct signal discrete point, reconstruction signal gross energy is

$E\left[f\left(t\right)\right]=\underset{-\&infin;}{\overset{\&infin;}{\&Integral;}}{f}^2\left(t\right)\mathrm{dt}=\underset{k=1}{\overset{\mathrm{<figure-callout\; id="2"\; label="n\; A\; k"\; filenames="HSA00000331943500011.png"\; state="\{\{state\}\}">n</figure-callout>}}{\mathrm{\&Sigma;}}}{A}_k^{}{}_2^{}.---\left(8\right)$

If A _{I, k}Be each frequency band reconstruction signal f _i(t) amplitude of each discrete point, n is the number of each band signal discrete point, the energy of each frequency band is

$E_j\left[f_i\left(t\right)\right]=\underset{-\&infin;}{\overset{\&infin;}{\&Integral;}}{f}_i^2\left(t\right)\mathrm{dt}=\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{A}_{i,k}^2---\left(9\right)$

Equally also have

$E\left[f\left(t\right)\right]=\underset{i=1}{\overset{2^i}{\mathrm{\&Sigma;}}}{E}_j\left[f_i\left(t\right)\right]---\left(10\right)$

The decomposable process of wavelet packet as shown in Figure 4.The data that hand-held spot check equipment is received are carried out WAVELET PACKET DECOMPOSITION and are extracted the Monitoring Data eigenwert, and with the relative value structural attitude vector of each reconstruct energy.

2. Support Vector data description:

Support Vector data description (SVDD) is a kind of monodrome sorting technique based on Statistical Learning Theory and border thought, is intended to seek to comprise the optimum hypersphere of target class.For sample X={x ₁, x ₂..., x _n, need find a hypersphere that contains minimum volume, make all x _iBe included in this spheroid.For the compactedness that improves the border is introduced gaussian kernel function.

Therefore, decision function is:

$\mathrm{\&chi;}\left(z\right)=1-2\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_i{K}_G(z\&CenterDot;x_i)+\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_i{\mathrm{\&alpha;}}_j{K}_G(x_i\&CenterDot;x_j)---\left(11\right)$

Wherein, α _iServe as reasons the training obtain corresponding to x _iCoefficient,

$K_G(x,y)=\exp \left(\frac{{\left|\right|x-y\left|\right|}^2}{{\mathrm{\&sigma;}}^2}\right)---\left(12\right)$

Hyperspherical radius is:

$R=\sqrt{\frac{1}{n}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\mathrm{\&chi;}\left(x_i\right)}---\left(13\right)$

X wherein _iCan be any support vector, the state of testing data z can be determined by following formula:

3. the concrete implementation step of failure prediction

The concrete implementation step of failure prediction is as follows:

A) utilize the average μ of original signal x (t) and standard deviation sigma standardization x (t),

B) utilize WAVELET PACKET DECOMPOSITION that the signal x (t) ' after standardizing is carried out 4 layers of WAVELET PACKET DECOMPOSITION, obtain 2 ⁴The reconstruction signal f of individual different frequency bands _i(t) (i=1 ..., 2 ⁴);

C) calculate each frequency band reconstruction signal f _i(t) energy adopts method for normalizing that each frequency band reconstruction signal is handled, and namely the mark that accounts for the signal gross energy with the energy of each frequency band reconstruction signal is represented, obtains the relative energy of each reconstruction signal

I=1,2 ..., 2 ⁴, have

D) utilize the relative energy E of each the frequency band reconstruction signal after the normalization ₄(i) as parameter, structural attitude vector T=[E ₄(1), E ₄(2) ..., E ₄(2 ⁴)];

E) status data with 1000 groups of good equipment is sample, utilizes the proper vector of constructing in the step d) as learning sample, and training SVDD monodrome sorter obtains among the decision function χ (z) and sample corresponding parameters α _iValue and the value of minimum hypersphere radius R;

F) with the sampled data of equipment Life cycle for people Г=χ (z)-R, if Г≤0, then health indicator HI=0; Otherwise HI=Г;

G) with the health indicator HI rendering apparatus performance degradation curve of the equipment Life cycle that obtains;

H) equipment Life cycle health indicator HI and the contrast of equipment performance degenerated curve that obtains with the spot check data computation, the degradation of judgment device and predict device remaining life.

Claims (6)

1. failure prediction method towards continuous casting of iron and steel equipment is characterized in that concrete steps are as follows:

When (1) carrying out spot check, wake the data acquisition node that is installed on monitoring of equipment point up by hand-held spot check equipment, begin image data by data acquisition channel is set;

(2) device status data of data acquisition node front end sensors collection after isolation, conversion and filtering are handled, obtains corresponding voltage signal;

(3) voltage signal through sampling and analog to digital conversion, obtains corresponding digital signal at data acquisition node;

(4) the digital signal packing of encoding in data acquisition node sends by less radio-frequency then;

(5) coded data that sends of hand-held spot check equipment receiving data acquisition node is decoded automatically and is stored;

(6) select operation to image data by the function menu of hand-held spot check equipment, carry out fault analysis and life prediction, send instruction simultaneously and change acquisition channel or stop to gather and makes the data acquisition node dormancy;

Wherein: fault analysis uses the WAVELET PACKET DECOMPOSITION technology to carry out obtaining of feature extraction, the domestication of SVDD sorter and equipment degenerated curve, and concrete steps are as follows:

(6.1) utilize the average μ of original signal x (t) and standard deviation sigma standardization x (t),

(6.2) utilize WAVELET PACKET DECOMPOSITION that the signal x (t) ' after standardizing is carried out the j layer and decompose, obtain 2 ^jThe reconstruction signal f of individual different frequency bands _i(t) (i=1 ..., 2 ^j);

(6.3) calculate each frequency band reconstruction signal f _i(t) energy adopts method for normalizing that each frequency band reconstruction signal is handled, and namely the mark that accounts for the signal gross energy with the energy of each frequency band reconstruction signal is represented, obtains the relative energy of each reconstruction signal $E_j\left(i\right)=\frac{E_j\left[f_i\left(t\right)\right]}{E\left[f\left(t\right)\right]},i=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,2^j,$ Have $\underset{i=1}{\overset{2^j}{\mathrm{\&Sigma;}}}{E}_j\left(i\right)=1;$

(6.4) utilize the relative energy E of each the frequency band reconstruction signal after the normalization _j(i) as parameter, structural attitude vector T=[E _j(1), E _j(2) ..., E _j(2 ^j)];

(6.5) status data with 1000 groups of good equipment is sample, utilizes the proper vector of constructing in the step (6.4) as learning sample, and training SVDD monodrome sorter obtains among the decision function χ (z) and sample corresponding parameters α _iValue and the value of minimum hypersphere radius R;

(6.6) with the sampled data of equipment Life cycle for people Γ=χ (z)-R, if Γ≤0, then health indicator HI=0; Otherwise HI=Γ;

(6.7) with the health indicator HI rendering apparatus performance degradation curve of the equipment Life cycle that obtains;

(6.8) fault analysis and life prediction: data substitution (6.6) step of gathering in the spot check process is judged, obtained the health indicator HI when time spot check, this desired value and the equipment degenerated curve that obtains are compared the equipment residual life that obtains predicting;

(7) prediction result is presented on the display module of hand-held spot check equipment.

2. employed failure prediction equipment of a kind of failure prediction method towards continuous casting of iron and steel equipment as claimed in claim 1, formed by data acquisition node mainboard (2) and hand-held spot check equipment (3), it is characterized in that data acquisition node mainboard (2) is by sensor (1), signal conditioning circuit (4), amplification filtering circuit (5), single-chip microcomputer (6), antenna circuit (7) and first power module (8) are formed, be connected by industrial cable between sensor (1) and the signal conditioning circuit (4), signal conditioning circuit (4) connects amplification filtering circuit (5), amplification filtering circuit (5) is connected with the data port of single-chip microcomputer (6), and first power module (8) connects sensor (1) respectively, signal conditioning circuit (4), amplification filtering circuit (5), single-chip microcomputer (6) and antenna circuit (7); Hand-held spot check equipment (3) is made up of keyboard (9), wireless communication module (10), second source module (11), communication interface (12), interface module (13), display module (14) and ARM minimum system (15), keyboard (9), wireless communication module (10), second source module (11), communication interface (12) and display module (14) are connected with interface module (13) respectively, and the other end of interface module (13) is connected with the data port of ARM minimum system (15).

3. the employed failure prediction equipment of a kind of failure prediction method towards continuous casting of iron and steel equipment according to claim 2, it is characterized in that described sensor (1) is one to multiple kind in temperature sensor, vibration transducer, pressure transducer, electromagnetic sensor or the eddy current inductor, temperature sensor is converted to corresponding voltage signal with temperature signal, vibration transducer with vibration acceleration signal be converted to voltage signal, pressure transducer is converted to voltage signal with pressure signal, electromagnetic flow transducer is converted to voltage signal with flow signal; Each sensor is installed on the monitoring point of online equipment, and links to each other with the mainboard of data acquisition node by cable; Different sensors links to each other with the respective signal modulate circuit, and the signal of each process conditioning is as the input of multi-way switch, and the output of multi-way switch links to each other with single-chip microcomputer (6) through signal conditioning circuit (4) output terminal.

4. the employed failure prediction equipment of a kind of failure prediction method towards continuous casting of iron and steel equipment according to claim 2 is characterized in that described signal conditioning circuit (4) is connected to form successively by electric capacity buffer circuit, amplifier circuit and RC anti-aliasing filter circuit.

5. the employed failure prediction equipment of a kind of failure prediction method towards continuous casting of iron and steel equipment according to claim 2 is characterized in that described single-chip microcomputer (6) adopts the wireless singlechip of built-in A/D converter and wireless radio frequency circuit.

6. a kind of device towards continuous casting of iron and steel equipment failure prediction according to claim 2, it is characterized in that described ARM minimum system (15) is by data acquisition module (16), system arranges module (17), data analysis module (18), failure prediction module (19), task management module (20), data derive module (21) and serial communication modular (22) is formed, data acquisition module (16) is communicated by letter with wireless radio frequency circuit with the antenna circuit on the data acquisition node mainboard by the Zigbee agreement, the duty of control data acquisition module (19) arranges the sampling monitoring point, sample frequency, sampling time and store sample data; System arranges module (17) system time, the display brightness of handing spot check equipment (3) is arranged; Data analysis module (18) is realized data management, data analysis and failure prediction; Data management comprises configuration information and the deletion data file of checking measuring point path, parameter; Spot check task in the hand-held spot check equipment (3) of task management module (20) management, measuring point information, execution time, the parameter of setting, the spot check result of each spot check task; Data derive module (21) and by communication interface (12) data are exported to USB flash disk or other memory devices; Serial communication modular (22) receives the spot check task that host computer assigns and uploads data to host computer by communication interface (12).

Images