CN104070066A - Early warning method for self-excited vibration of mill - Google Patents

Abstract

The invention provides an early warning method for self-excited vibration of a mill. The early warning method comprises the steps of S100 to S500, wherein in the step S100, a vibration acceleration signal of the mill in a moment is obtained; in the step S200, the spectrum vector of the vibration acceleration signal in the moment is obtained; in the step S300, the frequency spectrum information entropy of the vibration acceleration signal in the moment is calculated; in the step S400, the smoothing is performed for the frequency spectrum information entropy to obtain a smoothness value of the information entropy; and in the step S500, the early warning discrimination logical volume is calculated based on the steps of S100 to S400 to judge if the early warning is performed. With the calculation of the frequency spectrum information entropy as the core, the early warning method for the self-excited vibration of the mill increases the early warning accuracy of the mill so as to improve the production efficiency and the product quality.

Description

Rolling Mill Self-Vibration method for early warning

Technical field

The present invention relates to the process monitoring that cold rolled sheet is produced, more specifically, is a kind of rolling Mill Self-Vibration method for early warning.

Background technology

Rolling Mill Self-Vibration is the common problem of cold continuous rolling, is one of obstacle of restriction flat cold-rolled sheet output and product quality raising.This vibration is because of its paroxysmal feature, and general vibration measuring instrument and system are not competent, and conventional method for early warning has little time to react.Therefore, conventionally can only rely on five senses by site operation personnel, carry out at any time urgent reduction of speed, in order to avoid cause broken belt or rolled piece blemish, simultaneously unstable at rolling machine system or while closing on roll change, even if in the situation that not vibrating tendency, speed can not be improved, thereby affect production efficiency.

The vibration mode of milling train can be divided into two kinds of forced vibration and self-excited vibrations, and forced vibration is comparatively speaking than being easier to identification and improving, and self-excited vibration is difficult to find a convictive source, therefore reports to the police and just seems particularly important timely.For a long time, many scholars are devoted to the research of milling train self-oscillation problem, and research mainly concentrates on reason and the inherent mechanism that flutter occurs, and how effectively to weaken self-excited vibration, the raising cold-strip steel quality of milling train, also there is scholar from fault diagnosis, analysis and research treatment Countermeasures.But, self-excited vibration is a problem of all maintaining close ties with aspects such as equipment, technique, production operation, basic management, relate to mechanical, electrical, instrument, liquid, the field such as lubricated, be one with the tens of kinds of challenges that factor is relevant, each reason occurring is not quite similar, and also can not suppress completely.Therefore, more effective means are for site operation personnel provides early warning, to take correct valid function to avoid causing harmful effect because of self-excited vibration.

Summary of the invention

Object of the present invention, is to solve the existing deficiency to rolling Mill Self-Vibration judgement, thereby a kind of rolling Mill Self-Vibration method for early warning of innovation is provided.

Rolling Mill Self-Vibration method for early warning of the present invention, for the self-excited vibration of milling train is carried out to early warning, the method comprises the steps:

Step S100, obtains the vibration acceleration signal of a moment milling train;

Step S200, obtains the spectrum vector A[A of the frequency spectrum of this moment vibration acceleration signal ₁, A ₂a _m], wherein, A ₁, A ₂a _mfor frequency corresponding to each spectral line, m is that frequency is counted;

Step S300, utilizes following formula to calculate the spectrum information entropy H(A of this moment vibration acceleration signal):

$P\left(A_i\right)=\frac{f\left(A_i\right)}{\underset{k=1}{\overset{m}{\mathrm{\Σ}}}f\left(A_k\right)}$ i=1,2……m，

$H\left(A\right)=-\underset{i=1}{\overset{m}{\mathrm{\Σ}}}P\left(A_i\right){\log }_{2}P\left(A_i\right)/{\log }_{2}m,$

Wherein, P (A _i) for composing the probability density of vectorial A, f (A _i) be the amplitude that each spectral line is corresponding;

Step S400, carries out smoothing processing to the spectrum information entropy of this moment vibration acceleration signal, obtains this moment spectrum information entropy smooth value S _t:

S _t＝a·H _t(A)+(1-a)S _t-1，

Wherein, S _t-1for a upper moment spectrum information entropy smooth value, S ₀=0, H _t(A) be the spectrum information entropy of this moment vibration acceleration signal, a is smoothing factor, and its span is between 0 to 1;

Step S500, according to variable quantity logical value R and the rolling mill speed logical variable V of this moment spectrum information entropy smooth value logical value P, this moment spectrum information entropy and upper moment spectrum information entropy, calculate early warning and differentiate amount of logic Q, to judge whether to carry out early warning, wherein

Work as S _t≤ H _mtime, P=1, otherwise, P=0;

When | S _t-S _t-1|>=S _mtime, R=1, otherwise, R=0;

When rolling mill speed is more than or equal to V _mtime, V=1, otherwise, V=0;

And, Q=V ∩ [P ∪ R];

In the time of Q=1, early warning is carried out in judgement, and sends early warning signal, and in the time of Q=0, early warning is not carried out in judgement,

Wherein, H _mfor comentropy smooth value threshold value, S _mfor comentropy smooth value rate of change threshold value, V _mfor mill speed threshold value.

Utilize rolling Mill Self-Vibration method for early warning of the present invention, can provide milling train warning function by entropy, and vibration early warning accuracy rate is high, thereby improved production efficiency and product quality.

Brief description of the drawings

Fig. 1 is the schematic flow sheet of rolling Mill Self-Vibration method for early warning of the present invention;

Fig. 2 is the system schematic of implementing rolling Mill Self-Vibration method for early warning of the present invention.

Detailed description of the invention

Below in conjunction with the drawings and specific embodiments, the process step of rolling Mill Self-Vibration method for early warning of the present invention is elaborated.

In general, rolling Mill Self-Vibration method for early warning of the present invention, by obtain the vibration acceleration signal of milling train in each moment, and calculate the spectrum information entropy of this each this vibration acceleration signal of moment, and it is carried out to smoothing processing, the factor such as difference and mill speed in the smooth value of the comentropy based on this moment, itself and a upper moment, judging whether to have self-oscillation to produce and to carry out early warning.

As shown in Figure 1, rolling Mill Self-Vibration method for early warning of the present invention, comprises step S100 to S500, wherein, in step S100, obtains the vibration acceleration signal of a moment milling train; In step S200, obtain the spectrum vector of this moment vibration acceleration signal; In step S300, calculate the spectrum information entropy of this moment vibration acceleration signal; In step S400, this spectrum information entropy is carried out to smoothing processing, obtain the smooth value of this comentropy; And in step S500, based on above step, calculate early warning and differentiate amount of logic, judge whether to carry out early warning.Below in conjunction with accompanying drawing 1,2, these steps are described in more detail.

step S100.

In this step, the vibration acceleration signal of Real-time Obtaining milling train.As shown in Figure 2, by vibration acceleration sensor, the vibration acceleration of rolling-mill housing 11,12,13 is detected.Then the mill speed signal that data acquisition unit 20 obtains by this vibration acceleration signal and from mill control system 30 carries out digitized processing, and is delivered to home server 51 by router four 0.

Conventionally before further processing, can carry out digital filtering processing to the vibration acceleration signal in this moment.Digital filtering processing can realize by multiple means, for example, carries out digital filtering by finite impulse response filter (FIR).

step S200.

In this step, obtain the spectrum vector A[A of the frequency spectrum of this moment vibration acceleration signal ₁, A ₂a _m], wherein, A ₁, A ₂a _mfor frequency corresponding to each spectral line, m is that frequency is counted.

Particularly, the vibration acceleration signal that utilizes step S100 to obtain, can utilize conventional means, obtains the spectrum vector of this vibration acceleration signal frequency spectrum.Each subvector A in this spectrum vector A ₁, A ₂a _mfor frequency corresponding to each spectral line, following step also will be mentioned, and amplitude corresponding to each frequency is f (A _i).I is to the counting of the frequency after this signal analysis, and for example, the frequency of the analysis m that counts can be 400 points, etc.

step S300.

Utilize following formula to calculate the spectrum information entropy H(A of this moment vibration acceleration signal):

$P\left(A_i\right)=\frac{f\left(A_i\right)}{\underset{k=1}{\overset{m}{\mathrm{\Σ}}}f\left(A_k\right)}$ I=1,2 ... m, (formula 1)

$H\left(A\right)=-\underset{i=1}{\overset{m}{\mathrm{\Σ}}}P\left(A_i\right){\log }_{2}P\left(A_i\right)/{\log }_{2}m,$ (formula 2)

Wherein, P (A _i) for composing the probability density of vectorial A, f (A _i) be the amplitude that each spectral line is corresponding.Spectrum information entropy H(A) characterize the degree of uncertainty of this frequency spectrum, its value is larger, and uncertainty degree is higher or spectral pattern is more complicated, and its value is approximately little, and uncertainty degree is lower or spectral pattern is more single.Due to the excitation process of self-excited vibration be also its rumble spectrum feature by uncertain to definite process, therefore the size of frequency spectrum entropy has reflected self-excited vibration a situation arises.Degree of uncertainty is lower, and the possibility of self-excited vibration is larger.

step S400.

Spectrum information entropy to this moment vibration acceleration signal carries out smoothing processing, obtains this moment spectrum information entropy smooth value S _t:

S _t=aH _t(A)+(1-a) S _t-1(formula 3)

Wherein, S _t-1for a upper moment spectrum information entropy smooth value, S ₀=0, H _t(A) be the spectrum information entropy of this moment vibration acceleration signal, a is smoothing factor, and its span is between 0 to 1.Spectrum information entropy is carried out to smoothing processing, is in order to prevent in prealarming process, factor data sign mutation and cause erroneous judgement.

step S500.

According to variable quantity logical value R and the rolling mill speed logical variable V of this moment spectrum information entropy smooth value logical value P, this moment spectrum information entropy and upper moment spectrum information entropy, calculate early warning and differentiate amount of logic Q, to judge whether to carry out early warning, wherein,

Work as S _t≤ S _mtime, P=1, otherwise, P=0;

When | S _t-S _t-1|>=S _mtime, R=1, otherwise, R=0;

When rolling mill speed is more than or equal to V _mtime, V=1, otherwise, V=0;

And, Q=V ∩ [P ∪ R];

In the time of Q=1, early warning is carried out in judgement, and sends early warning signal, and in the time of Q=0, early warning is not carried out in judgement.

Wherein, H _mfor comentropy smooth value threshold value, S _mfor comentropy smooth value rate of change threshold value, V _mfor mill speed threshold value.

Above-mentioned Q is calculated and shown, at rolling mill speed, is also the mill speed of milling train while exceeding a threshold velocity, judgement: 1, whether to exceed certain uncertainty threshold value (be H to the smooth value of this moment comentropy _m), 2, whether this moment comentropy smooth value is than carving existing variation largely (is whether the two difference exceedes S upper a period of time _m), if these two conditions meet one, be judged as and be about to produce self-excited vibration, and carry out early warning to system transmission instruction.In conjunction with Fig. 2, determine and carry out after early warning when server 51, can send early warning signal 53 by man-machine interface 52, and can pass through other equipment, for example printer 54, informs operator, or for early warning record.

In prealarming process, can pass through two kinds of threshold value H _mand S _madjustment, adjust early warning susceptibility and the degree of accuracy.Also can set by threshold value H _mand S _madjustment, set different warning level and report to the police, for example send yellow alarm signal or red alarm signal, represent the occurrence degree of self-excited vibration.

Utilize rolling Mill Self-Vibration method for early warning of the present invention, can provide milling train warning function by entropy, can make the velocity error of early warning be controlled within 100m/ divides, and vibration early warning accuracy rate can reach 93.3%, thereby improve production efficiency and product quality.The method has objectively responded the Vibration Condition of unit, improved the situation of current dependence face operation, for Field Force provides early warning timely and effectively, the scene of being convenient to adjusts in time according to prompting, roll change foundation is provided, thereby can brings significant economic benefit.

Claims (1)

1. a rolling Mill Self-Vibration method for early warning, for the self-excited vibration of milling train is carried out to early warning, is characterized in that, the method comprises the steps:

Step S100, obtains the vibration acceleration signal of a moment milling train;

Step S200, obtains the spectrum vector A[A of the frequency spectrum of this moment vibration acceleration signal ₁, A ₂a _m], wherein, A ₁, A ₂a _mfor frequency corresponding to each spectral line, m is that frequency is counted;

Step S300, utilizes following formula to calculate the spectrum information entropy H(A of this moment vibration acceleration signal):

$P\left(A_i\right)=\frac{f\left(A_i\right)}{\underset{k=1}{\overset{m}{\mathrm{\Σ}}}f\left(A_k\right)}$ i=1,2……m，

$H\left(A\right)=-\underset{i=1}{\overset{m}{\mathrm{\Σ}}}P\left(A_i\right){\log }_{2}P\left(A_i\right)/{\log }_{2}m,$

Wherein, P (A _i) for composing the probability density of vectorial A, f (A _i) be the amplitude that each spectral line is corresponding;

Step S400, carries out smoothing processing to the spectrum information entropy of this moment vibration acceleration signal, obtains this moment spectrum information entropy smooth value St:

S _t＝a·H _t(A)+(1-a)S _t-1，

Wherein, S _t-1for a upper moment spectrum information entropy smooth value, S ₀=0, H _t(A) be the spectrum information entropy of this moment vibration acceleration signal, a is smoothing factor, and its span is between 0 to 1;

Step S500, according to variable quantity logical value R and the rolling mill speed logical variable V of this moment spectrum information entropy smooth value logical value P, this moment spectrum information entropy and upper moment spectrum information entropy, calculate early warning and differentiate amount of logic Q, to judge whether to carry out early warning, wherein

Work as S _t≤ H _mtime, P=1, otherwise, P=0;

When | S _t-S _t-1|>=S _mtime, R=1, otherwise, R=0;

When rolling mill speed is more than or equal to V _mtime, V=1, otherwise, V=0;

And, Q=V ∩ [P ∪ R];

In the time of Q=1, early warning is carried out in judgement, and sends early warning signal, and in the time of Q=0, early warning is not carried out in judgement,

Wherein, H _mfor comentropy smooth value threshold value, S _mfor comentropy smooth value rate of change threshold value, V _mfor mill speed threshold value.